Healthcare & Life Sciences

Four agentic workflows you can build for life sciences for R&D

Fri, 21 Nov 2025 17:00:00 +0000

AI agents, powered by generative AI, are rapidly transforming industries by acting as intelligent, collaborative partners that can interpret goals, plan multi-step actions, and work independently across systems, marking a significant shift in how businesses can find, understand, and act on their data. Our recent blog outlines how AI agents are transforming several industries.

Below we describe how to create a modular, end-to-end platform that accelerates the discovery and preclinical optimization of novel therapeutic candidates through a multi-agentic system. The system is designed to move from a high-level disease concept to a set of lead candidates with a high probability of success, regardless of the specific disease or therapeutic modality.

We see few key roles to be played by specialized AI agents, each based on a specialized open-weight model from Google, which can in turn be fine-tuned and trained for even more specialized purposes. Given the below agents are all based on open weight models, it gives a lot of room to further fine tune and train these models to build powerful agents.

Four agents you can build for life sciences

1. MedGemma: "The strategic intelligence agent"

- Expertise: Deep comprehension and synthesis of unstructured biomedical text, medical imaging, clinical data, and scientific literature.
- Function: Acts as a specialized knowledge agent. When directed by the Cognitive Orchestrator, it executes deep search and synthesis across biomedical corpora (e.g., PubMed, patient text records, other modalities such as chest x-rays) to extract findings, build cohorts, and summarize knowledge. MedGemma is especially useful for use cases requiring strict version control (e.g. regulated products), lower inference costs, or requiring substantial adaptation to specific use cases. Additionally its fast performance and efficient cost makes it very suitable for high volume medical use cases where speed and cost are of importance, a lot of those use cases are very common in LifeSciences

2. TxGemma: "The preclinical analyst"

- Expertise: Predicting functional and safety properties of therapeutic molecules
- Function: Predicts preclinical properties of drug candidates in silico, such as pharmacokinetics, permeability, toxicity, or efficacy.
- TxGemma Blog

3. Gemini 2.5 Pro: "The cognitive orchestrator agent"

Expertise: Advanced multi-step reasoning, dynamic planning, and contextual understanding to manage the end-to-end drug discovery workflow.
Function: Directs the specialized AI agents by interpreting high-level goals, sequencing tasks, evaluating results, and dynamically adapting the workflow to help the scientists achieve the final therapeutic objective.This orchestrator also accesses various tools. A tool can be a complete, specialized agent (like MedGemma) or a specific model endpoint (like AlphaFold), and is given a clear, natural language description of its function. For example, the MedGemma tool might be used as: "A tool that searches and synthesizes biomedical literature to identify potential disease targets based on a given pathology."
Note: For uses cases needing a version locked model and change control users have the option of using Gemma (Open Source) for this orchestration

4. AlphaFold-2 & molecular docking tools: "The molecular architect"

Expertise: Predicting the precise 3D structure of molecular targets and simulating how candidate molecules physically interact (dock) with them.
Function: Creates the essential structural blueprint of the drug-target interaction, enabling structure-based design, virtual screening, and specificity analysis

Here’s the step-by-step process

Phase 1: Find the target

A scientist prompts the system (e.g., "Find novel targets for Parkinson's"). The MedGemma agent ("AI Research Analyst") instantly scans millions of publications and clinical data to identify promising biological targets. The Orchestrator delivers a concise report, and the scientist approves the final target.

Phase 2: Generate candidates

The AlphaFold agent ("Molecular Architect") builds a 3D model of the target. Then, the TxGemma agent performs virtual screening, testing thousands of potential drug "keys" to see how they "fit" the target "lock," creating a shortlist of candidates.

Phase 3: The "Design-test-refine" loop

This is the core engine for rapidly improving candidates.

Predict: TxGemma ("Preclinical Analyst") runs a virtual simulation on each candidate, predicting its real-world performance (e.g., potency, toxicity).
Triage: The Orchestrator sorts them: "Promote" (looks excellent), "Archive" (a dead end), or "Optimize" (promising, but flawed).
Refine: "Optimize" candidates are automatically refined to fix their specific flaw and are sent right back into the loop.

This Design -> Dock -> Predict -> Refine cycle runs thousands of times on Google Cloud's high-performance computing, iterating on drug designs at a speed impossible in a physical lab.

Phase 4: Nominate lab-ready leads

After the loop, the Orchestrator presents the human scientist with the final, highly-optimized lead candidates. The scientist makes the final selection, and MedGemma re-engages to help design the optimal strategy for real-world lab testing.

By moving the costly "test-and-fail" part of discovery into this rapid, in-silico workflow, we can focus our lab resources on candidates with the highest probability of success, creating a faster, more intelligent path to new therapies.

Reference architecture

This diagram shows the foundational services and how data flows between them and how services work together.

Executing this sophisticated, iterative workflow requires a robust, scalable, and secure cloud platform. Google Cloud provides a comprehensive suite of services that map directly to the needs of each AI agent and the overall workflow, ensuring data integrity, compliance, and computational power.

How to get started using Google Cloud

Vertex AI Search is the core service for this agent's function. It can create a sophisticated Retrieval-Augmented Generation (RAG) system over a corpus of private biomedical data, such as internal research documents, PubMed literature, and clinical trial data. This directly enables the agent to answer natural language queries and synthesize information with citations.
Vertex AI. Google Cloud offers managed, optimized AlphaFold environments and integrations. For high-throughput needs, Vertex AI Training with GPU or TPU acceleration can run thousands of protein folding and docking simulations in parallel. Use Vertex AI Agent Builder to create agents.

^{We would like to thank Ryan Ye Min Thein (Customer Engineer, Google Cloud) and Justin Chen (Clinician Specialist, Google Health) for their contributions}

The Blueprint: How Giles AI transforms medical research with conversational AI

Tue, 28 Oct 2025 19:00:00 +0000

Welcome to The Blueprint, a new feature where we highlight how Google Cloud customers are tackling unique and common challenges across industries using the latest AI and cloud technologies. We hope to inspire others looking to innovate in their work.

Giles AI is a London-based startup that helps healthcare and life sciences organizations quickly extract insights from fragmented data, whether that data is available in an online repository (e.g. PubMed, NICE, the FDA etc.), local documents, or internal IP. Users can connect to internal and external data repositories and upload documents and images to the Giles AI platform; this integration allows users to the combined knowledge base for insights more quickly and efficiently, using natural language prompts and an intuitive interface.

The challenge:

As Giles AI grew in popularity, our incumbent cloud provider struggled to cope with complex data flows, new LLMs, and external APIs. Latency increased, slowing the user interface and impacting critical activities. Engineers also required a more agile development environment. Security is also a foundational feature of Giles AI and everything we build — with clinical, medical, and healthcare standards in mind, sensitive data must be protected at every step, both at rest and in transit.

The solution:

Giles AI leveraged Google Cloud’s modular, API-friendly, microservices-based architecture to minimize latency, easily manage complex clinical data flows in real-time, and capitalize on the latest and greatest AI foundation models as they are released.

Backend service orchestration in Google Kubernetes Engine and lightweight microservices in Cloud Run are complemented by specialized workloads in Compute Engine to keep the Giles AI platform available, flexible, and scalable without the heavy management and maintenance demands of legacy infrastructure. Cloud Load Balancing ensures efficiency.

Cloud SQL, Cloud Storage, and Document AI help the Giles AI platform manage structured and unstructured data and extract insights. Under the hood, Vertex AI handles model selection and prompt orchestration. The system is model-agnostic by design, enabling Giles AI to route queries to the most appropriate language model including hundreds available through Model Garden on Vertex AI.

With this highly flexible approach, Giles AI is able to deliver numerous healthcare and life sciences use cases from systematic literature reviews and regulatory reviews to meta-analyses, data extraction, and patient eligibility screening — all with high levels of accuracy and agreement.

To enhance security, we’re using Cloud Armor to defend against web-based attacks, and Security Command Center to proactively manage risks and secure our cloud environment. Google Cloud regional databases help Giles AI localize data at rest — a critical need given healthcare regulations.

The architecture:

The outcome:

What we love about Vertex AI is that it supports our AI workflow experimentation. In simple terms, this means we can plug any LLM of our choice in and out of our workflow, drawing from the hundreds of models available in the Model Garden on Vertex AI. This provides amazing flexibility and efficiency, which is key to our success.

So far, the results of our migration have been impressive.

One of Giles AI’s early customers achieved an 85% reduction in the time required for clinical research tasks and over 94% response accuracy, with references provided when they wanted to be certain and verify. This customer was so compelled with the results that they went on to make a significant investment into the company and became a strategic partner.

Latency, uptime, and scalability have all improved significantly, even with complex, multi-layered data queries. From an internal perspective, Giles AI has seen an increase in developer velocity, with infrastructure-as-code and managed services reducing engineering overheads.

The Giles AI generative AI assistant interface

Looking to the future, our team at Giles AI is excited for the potential of Google Cloud’s AI foundation models designed for the medical community. These include MedGemma, a family of open-source AI models tailored for medical applications, and TxGemma, a suite of open therapeutic-language models derived from Gemma 2 that help streamline drug discovery and development.

With these powerful tools on the horizon, Giles AI is poised to deliver smarter, more verticalized decision-making across the entire healthcare R&D pipeline. For clients, this means turning complex data into real-world breakthroughs, faster than ever before.

Inside the AI-powered assistant helping doctors work faster and better at Seattle Children’s Hospital

Thu, 18 Sep 2025 16:00:00 +0000

Though its name may suggest otherwise, Seattle Children’s is the largest pediatric healthcare system in the world.

While its main campus is in its namesake city, Seattle Children’s also encompasses 47 satellite hospitals across Alaska, Montana, Idaho, and Washington, and patients come from as far away as Hawaii for treatment. For more than 100 years, Seattle Children’s has helped kids across the Western U.S. get healthy and stay healthy, regardless of the ability to pay.

With so much ground to cover and diverse patient populations to treat, Seattle Children's has always looked to new technologies to bring improved, consistent care to its patients and providers. Generative AI is now the latest advance in their medical toolkit.

It started roughly two decades ago, when Seattle Children’s created its pediatric clinical pathways, a set of standardized protocols designed to help clinicians make quicker and more reliable decisions to address dozens of medical conditions. Such pathways were becoming commonplace across medicine, and Seattle Children’s had developed some of the first for children’s unique medical needs.

Innovative as these were, they still required clinicians to thumb through indexes and long binders of information to find what they needed for a given ailment. And in healthcare, it’s often the case that every second counts.

Seattle Children’s was already working with Google Cloud on a number of projects, and as we began to explore the potential for generative AI to make the work of our clinicians easier, the clinical pathways seemed like an obvious place to start. Using Vertex AI and Gemini, we were able to quickly develop our Pathways Assistant, which took training from the clinical pathways documentation and supercharged it with not just searchability but conversationality.

Instead of flipping pages, we’d flipped the script on how quickly and reliably clinicians could find the lifesaving information they needed.

aside_block: <ListValue: [StructValue([('title', '2025 ROI of AI in Healthcare and Life Sciences'), ('body', <wagtail.rich_text.RichText object at 0x7f9aad6fd160>), ('btn_text', 'Read now'), ('href', 'https://cloud.google.com/resources/content/roi-of-ai-healthcare-life-sciences'), ('image', None)])]>

The pathways to improved healthcare run through Gemini

“Clinical pathways” are end-to-end treatment protocols for a specific condition or illness. Seattle Children’s pediatric clinical pathways are widely respected and used by hospitals around the globe, providing information on everything from diagnostic criteria to testing protocols to medication recommendations.

Previously, these clinical pathways were documented exclusively in PDFs — hundreds of thousands of pages of them. Performing a traditional search of their contents for the answers clinicians needed delayed their ability to provide treatment in an environment where minutes or even seconds can be critical.

Google Cloud engineers worked with Seattle Children’s informatics physicians, who straddle the worlds of healthcare and technology, to create Pathway Assistant. The new multimodal AI chatbot that responds to spoken or written natural-language queries using the information in those PDFs.

After processing a question, Pathway Assistant searches each PDF’s metadata, which contains semi-structured data in JSON format that’s been extracted from the PDFs by Gemini and curated by clinicians. It then selects the most relevant PDFs, parses the information — including any complex flowcharts, diagrams, and illustrations embedded in them — and answers the clinician’s question in just a few seconds.

Interactive information-finding for accurate decision-making

Pathway Assistant becomes more accurate with use. Healthcare providers can “discuss” clinical pathways with the chatbot, which, instead of answering a question, poses questions of its own if it needs clarification, going back and forth until it’s confident it can answer accurately. The chatbot always displays the specific sections of each PDF that was the source for formulating its answers, helping clinicians confirm the veracity of responses.

The interface also includes a way for users to provide feedback about the accuracy and appropriateness of the chatbot’s analysis and answers. The feedback is then logged in a BigQuery table for future forensic analysis — both by clinicians, who can query the information using natural language, and by the built-in Gemini models, which processes the feedback and summarizes what clinicians found confusing or how to improve the accuracy of future answers.

This reflexive capability enables Pathway Assistant to update the PDFs based on clinicians’ feedback if the inaccuracy stemmed from the PDF’s content. Clinicians are also finding that the metadata is becoming more accurate and requiring less curation. Pathway Assistant even corrects typos in the documentation automatically. And as new clinical pathways are developed, PDFs containing the latest information are added to the PDF library.

This growing collection is housed securely in Google Cloud Storage, and the bigger it gets, the more useful it becomes — which wasn't always the case. Whereas an expanding paper-based collection contained more information, it was also more material to wade through, which is especially challenging in emergency medical situations. Pathway Assistant almost entirely relieves this burden, synthesizing and delivering the most complete information at any time in a matter of seconds.

Ultimately, Pathway Assistant is not a decision-making tool but rather an information-finding tool. Research into critical, evidence-based guidelines that used to take hours now takes minutes.

This speed and effectiveness helps clinicians make the right decisions more quickly at the point of care, drastically reducing research time and improving patient safety and outcomes. Ultimately, clinicians can spend more time with more patients, not with more PDFs.

Ask any physician, they’ll tell you that’s what the best medical technology enables them to do — focus on the patient, not paperwork.

Now available: Cloud HSM as an encryption key service for Workspace client-side encryption

Mon, 18 Aug 2025 16:00:00 +0000

Organizations in highly-regulated sectors, such as government, defense, financial services, and healthcare, are required to meet stringent standards to safeguard sensitive data. Client-side encryption (CSE) for Google Workspace is a unique, privacy-preserving offering that keeps customer data confidential and enables the customer to be the sole arbiter of their data, helping them adhere to rigorous compliance regimes.

Google Workspace CSE adds another layer of encryption to your organization's data — like files, emails, meetings, and events — in addition to the default encryption that Google Workspace provides. CSE can be especially beneficial for organizations that store sensitive and regulated data because it can provide:

Confidentiality for organizations working with sensitive intellectual property, healthcare records, and financial data.
Compliance support for organizations in highly-regulated industries that have ITAR and EAR requirements.
Data sovereignty for organizations that need demonstrative data control using encryption keys that can be held at a defined boundary, such as a specific geographic location or within a nation’s borders.

To help highly-regulated organizations meet their encryption key service obligation, we are now offering Cloud Hardware Security Module (HSM) for Google Workspace (CHGWS,) bringing Google Cloud’s highest levels of compliance classifications to Workspace CSE customers. Cloud HSM is a highly available and scalable, fully managed key management service operated at cloud scale with hardware-backed keys stored in FIPS 140-2 Level 3 compliant HSMs (hardware security modules).

Available today in the U.S., and globally in the coming months, CHGWS offers a convenient, flat pricing model that makes it easy to set up and maintain.

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud security products'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac3447cd0>), ('btn_text', ''), ('href', ''), ('image', None)])]>

Use Cloud HSM to help meet regulatory obligations

Cloud HSM is engineered to support cloud workloads that are subject to the most stringent security and regulatory mandates, and has undergone comprehensive audits and achieved compliance with regulations and certifications including FedRAMP High, DISA IL5, ITAR, SOC 1/2/3, and PCI DSS.

A cornerstone of Cloud HSM for Google Workspace’s security posture is its reliance on FIPS 140-2 Level 3 validated Marvell LiquidSecurity HSMs. Specifically, the service uses models CNL3560-NFBE-2.0-G and CNL3560-NFBE-3.0-G, running firmware versions 3.4 build 09. This validation level is critical, as it indicates that the cryptographic modules have met the highest standards of security for hardware and software components.

This extensive list of certifications provides strong assurance to customers in highly regulated market segments that their key management and data protection needs are met in accordance with the most demanding regulatory and compliance frameworks.

Our emphasis on comprehensive compliance can help simplify the burdens faced by these organizations, and can allow them to confidently deploy and manage their encryption keys while satisfying their legal and audit requirements.

While security and compliance are paramount, Google Cloud also recognizes the critical importance of high availability and scalability for its customers. CHGWS can help address these needs by offering a highly available and standards-compliant CSE key service that can be deployed rapidly, often in minutes.

Our rapid deployment capability, combined with inherent high availability, can help ensure that critical encryption services are always accessible, minimizing potential disruptions to operations.

How does Cloud HSM for Google Workspace work?

CHGWS can enhance privacy and compliance for Google Workspace CSE. The data is encrypted end-to-end and can only be decrypted by users who have permission to access it.

Encrypting data

Step 1: When a user creates content in Google Workspace, the CSE library generates a data encryption key (DEK) that is sent to the CHGWS service.

Step 2: The CHGWS service verifies the user’s identity using a customer-managed identity provider and Google Cloud IAM.

Step 3: The CHGWS service then encrypts the DEK using a customer-managed encryption key (CMEK) stored in Cloud HSM, and sends the encrypted DEK back. Then the CSE library encrypts the content using the DEK, and the encrypted DEK is stored with the content.

Reading encrypted data

When a user tries to access encrypted content the process unfolds in reverse.

Step 4: First, the CSE library sends the encrypted DEK stored with the content to CHGWS service. CHGWS service verifies the user’s identity using the customer-managed identity provider.

Step 5: CHGWS service uses the CMEK stored in Cloud HSM to decrypt the DEK, and sends it back.

Step 6: The CSE library uses the decrypted DEK to decrypt the content.

All the encrypt and decrypt operations using CMEK are always performed inside the HSM. The CMEK never leaves the HSM protection boundary to ensure that customers maintain full control over their encryption keys and data access.

Generating audit logs using Cloud Logging: As with all Google Cloud services, Cloud HSM service writes audit logs that record administrative activities and accesses in your Google Cloud resources. Audit logs help you answer "who did what, where, and when?" in your Google Cloud resources, with the same level of transparency as in on-premises environments. This is part of our comprehensive Access Transparency offering.

Enabling audit logs can help your security, auditing, and compliance entities monitor Google Cloud data and systems for possible vulnerabilities or external data misuse. You can learn more about KMS Audit Logging here.

Regional availability and our SLA

Our Cloud HSM service provides 99.95% uptime for encryption and decryption operations. You can review Cloud Key Management Service and Cloud HSM SLA for more details. To get started, please see our onboarding instructions.

Manipal Hospitals and Google Cloud partner to transform nurse handoffs with GenAI

Fri, 11 Jul 2025 16:00:00 +0000

As one of India's largest healthcare providers, Manipal Hospitals serves nearly 7 million patients annually across 37 hospitals. To deliver clinical excellence and patient-centric care at a high standard, we are continually embracing technology. One of our most significant operational challenges we consistently face is the nurse handover process—a critical but time-consuming task. To make nurse handovers more efficient, safe and accurate, we entered a strategic partnership with the Google Cloud Consulting team to co-develop a generative AI solution, leveraging the power of Google Cloud.

Rethinking time-consuming, error-prone nurse handoffs

The process of transferring essential information about a patient's condition and care plan from an outgoing nurse to an incoming one is crucial for ensuring continuity of care and patient safety. However, with more than 10,500 beds across our hospitals, the sheer volume of data required for a comprehensive handover meant our nurses routinely added an extra 90 minutes to their shifts for both creating and receiving these reports. This lengthy process could directly affect patient care, as it could lead to fatigue and potential mistakes, and also reduce job satisfaction for our vital nursing staff. We needed a way to make this process faster, more accurate, and less of a burden.

Building a trusted solution on Google Cloud

Our joint Manipal-Google team knew that for a clinical tool to be adopted by over 5,000 nurses, it had to be both fast and trustworthy. The primary challenge with any generative AI application in healthcare is ensuring accuracy and minimizing the risk of AI “hallucinations.”

The solution's architecture, designed by the Google Cloud Consulting team, addresses this head-on by leveraging multiple Google Cloud components. Patient data from our TrakCare system is securely transferred in near real-time to a data lake on Google Cloud. When a nurse requests a handover summary, a serverless Cloud Run application orchestrates a multi-stage process.

Critically, instead of passing pages of raw data directly to the AI, the system first uses intelligent, time-based filters to extract only the most relevant clinical information for the specific shift. This structured, pre-processed data is then sent to Gemini on Vertex AI. This “controlled generation” approach was a key innovation; it ensures Gemini summarizes only the most pertinent facts, dramatically improving the accuracy and consistency of the final ISBAR (Identify, Situation, Background, Assessment, and Recommendation) report. The ability of Gemini to understand complex medical terminology, medication names, and clinical procedures without specialized fine-tuning was a game-changer, accelerating the entire development process.

How our partnership delivered results

By combining Manipal's deep clinical expertise and Google Cloud Consulting's technical leadership, our joint approach provides a blueprint for enterprise-grade AI implementation:

From ideation to production: The Google Cloud Consulting team led the engagement from the initial idea all the way to a production-ready solution now used by thousands of nurses daily. The project started with a focused Minimum Viable Product (MVP) to prove the technology's value before scaling.
User-centric design: The solution was not built in a vacuum. The Google team conducted over eight rounds of deep discussion and evaluation sessions directly with our nurses. This ensured the final ISBAR summary format was not just technically impressive, but clinically useful from day one.
Agile and iterative rollout: The solution was piloted at one hospital initially to test its performance and safety in a real-world setting. With a successful pilot, the solution is live in 23 of Manipal hospitals, and used by more than 5,000 every day. At full scale, it is projected to help save significant nurse hours on a daily basis. This phased approach, managed jointly, has allowed us to gather feedback and ensure smooth adoption.

Ensuring better patient care

The generative AI solution we implemented has yielded impressive returns. The 70% reduction in handoff time—from 90 minutes down to 20—frees our nurses to focus more on direct patient needs and care. It also makes the process less vulnerable to errors that can arise from handwritten notes and human fatigue.

This project, delivered in partnership with Google Cloud Consulting, is a prime example of how we are pioneering the future of healthcare in India, helping us scale the delivery of quality care across the length and breadth of the country.

^{We’d like to give special thanks to Google Cloud Consulting team - Naveen Poosarla, Gopala Dhar, Rupjit Chakraborty, Hem Anand, Amit Dutta, Nishant Welpulwar, Preetam Dey and Shikha Saxena - for designing and developing the solution. We are grateful to the Manipal Hospitals team - Saroja Jaykumar, Sunil Bhattacharjee - in delivering this successful project.}

aside_block: <ListValue: [StructValue([('title', 'Try Google Cloud for free'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac15cdfa0>), ('btn_text', 'Get started for free'), ('href', 'https://console.cloud.google.com/freetrial?redirectPath=/welcome'), ('image', None)])]>

Cloud CISO Perspectives: The global threats facing EU healthcare

Mon, 30 Jun 2025 16:00:00 +0000

Welcome to the second Cloud CISO Perspectives for June 2025. Today, Thiébaut Meyer and Bhavana Bhinder from Google Cloud’s Office of the CISO discuss our work to help defend European healthcare against cyberattacks.

As with all Cloud CISO Perspectives, the contents of this newsletter are posted to the Google Cloud blog. If you’re reading this on the website and you’d like to receive the email version, you can subscribe here.

aside_block: <ListValue: [StructValue([('title', 'Get vital board insights with Google Cloud'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac0902040>), ('btn_text', 'Visit the hub'), ('href', 'https://cloud.google.com/solutions/security/board-of-directors?utm_source=cloud_sfdc&utm_medium=email&utm_campaign=FY24-Q2-global-PROD941-physicalevent-er-CEG_Boardroom_Summit&utm_content=-&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

The global threats facing European hospitals and health organizations

By Thiébaut Meyer, director, Office of the CISO, and Bhavana Bhinder, European healthcare and life sciences lead, Office of the CISO

Thiébaut Meyer, director, Office of the CISO

As the global threat landscape continues to evolve, hospitals and healthcare organizations remain primary targets for cyber threat actors. To help healthcare organizations defend themselves so they can continue to provide critical, life-saving patient care — even while facing cyberattacks — the European Commission has initiated the European Health Security Action Plan to improve the cybersecurity of hospitals and healthcare providers.

There are two imperative steps that would both support Europe's plan and bolster resilience in our broader societal fabric: Prioritizing healthcare as a critical domain for cybersecurity investment, and emphasizing collaboration with the private sector. This approach, acknowledging the multifaceted nature of cyber threats and the interconnectedness of healthcare systems, is precisely what is required to secure public health in an increasingly digitized world. It’s great to see the European Commission has recently announced funding to improve cybersecurity, including for European healthcare entities.

Bhavana Bhinder, European healthcare and life sciences lead, Office of the CISO

At Google, we have cultivated extensive industry partnerships across the European Union to help healthcare organizations of all levels of digital sophistication and capability be more resilient in the face of cyberattacks.

Collaboration across healthcare organizations, regulators, information sharing bodies and technology providers like Google is essential to get and stay ahead of these attacks.

Cyberattacks targeting the healthcare domain, especially those that leverage ransomware, can take over healthcare systems - completely upending their operations and stopping them from providing life-saving medical procedures, coordinating critical scheduling and payment activities, stopping delivery of critical supplies like blood and tissue donations, and can even render the care facilities physically unsafe. In some cases, these cyberattacks have contributed to patient mortality. The statistics paint a grim picture:

Ransomware attacks accounted for 54% of analyzed cybersecurity incidents in the EU health sector between 2021 and 2023, with 83% financially motivated.
71% of ransomware attacks impacted patient care and were often coupled with patient data breaches, according to a 2024 European Commission report.
Healthcare's share of posts on data leak sites has doubled over the past three years, even as the number of data leak sites tracked by Google Threat Intelligence Group increased by nearly 50% in 2024. In one example, a malicious actor targeting European organizations said that they were willing to pay 2% to 5% more for hospitals — particularly ones with emergency services.
In-hospital mortality shoots up 35% to 41% among patients already admitted to a hospital when a ransomware attack takes place.
The U.K.’s National Health Service (NHS) has confirmed that a major cyberattack harmed 170 patients in 2024.

“Achieving resilience necessitates a holistic and adaptive approach, encompassing proactive prevention that uses modern, secure-by-design technologies paired with robust detection and incident response, stringent supply chain management, comprehensive human factor mitigation, strategic utilization of artificial intelligence, and targeted investment in securing unique healthcare vulnerabilities,” said Google Cloud’s Taylor Lehmann, director, Healthcare and Life Sciences, Office of the CISO. “Collaboration across healthcare organizations, regulators, information sharing bodies and technology providers like Google is essential to get and stay ahead of these attacks.”

Bold action is needed to combat this scourge, and that action should include helping healthcare providers migrate to modern technology that has been built securely by design and stays secure in use. We believe security must be embedded from the outset — not as an afterthought — and continuously thereafter. Google's secure-by-design products and services have helped support hospitals and health organizations across Europe in addressing the pervasive risks posed by cyberattacks, including ransomware.

Secure-by-design is a proactive approach that ensures core technologies like Google Cloud, Google Workspace, Chrome, and ChromeOS are built with inherent protections, such as:

Encrypting Google Cloud customer data at rest by default and data in transit across its physical boundaries, offering multiple options for encryption key management and key access justification.
Designing Google Workspace with security-first principles, incorporating AI defenses that block over 99.9% of spam, phishing, and malware, and client-side encryption.
Building security and compliance into ChromeOS, which powers Chromebooks, to help protect against ransomware attacks. ChromeOS boasts a record of no reported ransomware attacks. Its architecture includes capabilities such as Verified Boot, sandboxing, blocked executables, and user space isolation, along with automatic, seamless updates that proactively patch vulnerabilities.
Providing health systems with a secure alternative through Chrome Enterprise Browser and ChromeOS for accessing internet-based and internal IT resources crucial for patient care.
Committing explicitly in our contracts to implementing and maintaining robust technical, organizational, and physical security measures, and supporting NIS2 compliance efforts for Google Cloud and Workspace customers.

Our products and services are already helping modernize and secure European healthcare organizations, including:

In Germany, healthcare startup Hypros has been collaborating with Google Cloud to help hospitals detect health incidents without compromising patient privacy. Hypros’ innovative patient monitoring system uses our AI and cloud computing capabilities to detect and alert staff to in-hospital patient emergencies, such as out-of-bed falls, delirium onset, and pressure ulcers. They’ve tested the technology in real-world trials at leading institutions including the University Hospital Schleswig-Holstein, one of the largest medical care centers in Europe.
With the CUF, Portugal's largest healthcare provider with 19 hospitals and clinics. CUF has embraced Google Chrome and cloud applications to enhance energy efficiency and streamline IT operations. ChromeOS is noted in the industry for its efficiency, enabling operations on machines that consume less energy and simplifying IT management by reducing the need for on-site hardware maintenance.
For the Canary Islands 112 Emergency and Safety Coordination Center, which is migrating to Google Cloud. Led by the public company Gestión de Servicios para la Salud y Seguridad en Canary Islands (GCS) and developed in conjunction with Google Cloud, this migration is one of the first in which a public emergency services administration has moved to the public cloud. They’re also using Google Cloud’s sovereign cloud solutions to help securely share critical information, such as call recordings and personal data, with law enforcement and judicial bodies.

Building partnerships and sharing information

Information sharing is a vital component of securing the healthcare sector against evolving cyber threats. Google actively develops partnerships with information sharing and analysis centers (ISACs) across more than ten critical infrastructure sectors, including a robust ambassadorship with the Health Information Sharing and Analysis Center (Health-ISAC), and with the European Union Agency for Cybersecurity (ENISA).

We believe that information sharing must extend beyond threat intelligence to encompass data-supported conclusions regarding effective practices, counter-measures, and successes. Reducing barriers to sophisticated and rapid intelligence-sharing, coupled with verifiable responses, can be the decisive factor between a successful defense and a vulnerable one.

Our engagement with organizations including the international Health-ISAC and ENISA underscores our commitment to building trust across many communities, a concept highly pertinent to the EU's objective of supporting the European Health ISAC and the U.S.-based Health-ISAC’s EU operations.

Protecting European health data with Sovereign Cloud and Confidential Computing

We’re committed to digital sovereignty for the EU and to helping healthcare organizations take advantage of the transformative potential of cloud and AI without compromising on security or patient privacy.

We’ve embedded our secure-by-design principles in our approach to our digital sovereignty solutions. By enabling granular control over data location, processing, and access, European healthcare providers can confidently adopt scalable cloud infrastructure and deploy advanced AI solutions, secure in the knowledge that their sensitive patient data remains protected and compliant with European regulations like GDPR, the European Health Data Space (EHDS), and the Network and Information Systems Directive.

Additionally, Confidential Computing, technology that we helped pioneer, has helped narrow that critical security gap by protecting data in use.

Google Cloud customers such as AiGenomix leverage Confidential Computing to deliver infectious disease surveillance and early cancer detection. Confidential Computing helps them ensure privacy and security for genomic and related health data assets, and also align with the EHDS's vision for data-driven improvements in healthcare delivery and outcomes.

Building trust in global healthcare resilience

We believe that these insights and capabilities offered by Google can significantly contribute to the successful implementation of the European Health Security Action Plan. We are committed to continued collaboration with the European Commission, EU member states, and all stakeholders to build a more secure and resilient digital future for healthcare.

To learn more about Google’s efforts to secure and support healthcare organizations around the world, contact our Office of the CISO.

aside_block: <ListValue: [StructValue([('title', 'Join the Google Cloud CISO Community'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac0902700>), ('btn_text', 'Learn more'), ('href', 'https://rsvp.withgoogle.com/events/ciso-community-interest?utm_source=cgc-blog&utm_medium=blog&utm_campaign=2024-cloud-ciso-newsletter-events-ref&utm_content=-&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

In case you missed it

Here are the latest updates, products, services, and resources from our security teams so far this month:

Securing open-source credentials at scale: We’ve developed a powerful tool to scan open-source package and image files by default for leaked Google Cloud credentials. Here’s how to use it. Read more.
Audit smarter: Introducing our Recommended AI Controls framework: How can we make AI audits more effective? We’ve developed an improved approach that’s scalable and evidence-based: the Recommended AI Controls framework. Read more.
Google named a Strong Performer in The Forrester Wave for security analytics platforms: Google has been named a Strong Performer in The Forrester Wave™: Security Analytics Platforms, Q2 2025, in our first year of participation. Read more.
Mitigating prompt injection attacks with a layered defense strategy: Our prompt injection security strategy is comprehensive, and strengthens the overall security framework for Gemini. We found that model training with adversarial data significantly enhanced our defenses against indirect prompt injection attacks in Gemini 2.5 models. Read more.
Just say no: Build defense in depth with IAM Deny and Org Policies: IAM Deny and Org Policies provide a vital, scalable layer of security. Here’s how to use them to boost your IAM security. Read more.

Please visit the Google Cloud blog for more security stories published this month.

aside_block: <ListValue: [StructValue([('title', 'Learn something new'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac0902730>), ('btn_text', 'Watch now'), ('href', 'https://www.youtube.com/watch?v=t_roGtOtlDE'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

Threat Intelligence news

What’s in an ASP? Creative phishing attack on prominent academics and critics of Russia: We detail two distinct threat actor campaigns based on research from Google Threat Intelligence Group (GTIG) and external partners, who observed a Russia state-sponsored cyber threat actor targeting prominent academics and critics of Russia and impersonating the U.S. Department of State. The threat actor often used extensive rapport building and tailored lures to convince the target to set up application specific passwords (ASPs). Read more.
Remote Code Execution on Aviatrix Controller: A Mandiant Red Team case study simulated an “Initial Access Brokerage” approach and discovered two vulnerabilities on Aviatrix Controller, a software-defined networking utility that allows for the creation of links between different cloud vendors and regions. Read more.

Please visit the Google Cloud blog for more threat intelligence stories published this month.

Now hear this: Podcasts from Google Cloud

AI red team surprises, strategies, and lessons: Daniel Fabian joins hosts Anton Chuvakin and Tim Peacock to talk about lessons learned from two years of AI red teaming at Google. Listen here.
Practical detection-as-code in the enterprise: Is detection-as-code just another meme phrase? Google Cloud’s David French, staff adoption engineer, talks with Anton and Tim about how detection-as-code can help security teams. Listen here.
Cyber-Savvy Boardroom: What Phil Venables hears on the street: Phil Venables, strategic security adviser for Google Cloud, joins Office of the CISO’s Alicja Cade and David Homovich to discuss what he's hearing directly from boards and executives about the latest in cybersecurity, digital transformation, and beyond. Listen here.
Beyond the Binary: Attributing North Korean cyber threats: Who names the world's most notorious APTs? Google reverse engineer Greg Sinclair shares with host Josh Stroschein how he hunts down and names malware and threat actors, including Lazarus Group, the North Korean APT. Listen here.

To have our Cloud CISO Perspectives post delivered twice a month to your inbox, sign up for our newsletter. We’ll be back in a few weeks with more security-related updates from Google Cloud.

How AI & IoT are helping detect hospital incidents — without compromising patient privacy

Tue, 24 Jun 2025 16:00:00 +0000

Hospitals, while vital for our well-being, can be sources of stress and uncertainty. What if we could make hospitals safer and more efficient — not only for patients but also for the hard-working staff who care for them? Imagine if technology could provide an additional safeguard, predicting falls, or sensing distress before it's even visible to the human eye.

Many hospitals today still rely on paper-based processes before transforming critical information to digital systems, leading to frequent — and sometimes, remarkably absurd — inefficiencies. In-person patient monitoring, while standard practice, can be slow, incomplete, and subject to human error and bias. In one serious incident, shared by hospital staff, a patient fell shortly after getting out of bed at 5 a.m. and wasn’t discovered until the routine 6:30 a.m. check. Events like this underscore the need for continuous, 24/7 in-room monitoring solutions that can alert staff immediately in high-risk and emergency situations.

Driven by a shared vision to enhance patient care, healthcare innovator Hypros and Google Cloud joined forces to develop an AI-assisted patient monitoring system that detects and alerts staff to in-hospital patient emergencies, such as out-of-bed falls, delirium onset, or pressure ulcers. This innovative privacy-preserving solution enables better prioritization of care and a strong foundation for clinical decision-making — all without the use of invasive cameras.

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud AI and ML'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac34682e0>), ('btn_text', 'Start building for free'), ('href', 'http://console.cloud.google.com/freetrial?redirectPath=/vertex-ai/'), ('image', None)])]>

Privacy-preserving, AI-assisted patient monitoring

While the need for 24/7 patient monitoring is clear, developing these solutions raises important concerns around privacy and professional conduct. Privacy is paramount in any patient-monitoring technology for both the individuals receiving care and the professionals providing it. Even seemingly simple aspects, such as interventions within the patient’s immediate surroundings, require strict compliance with hospital hygiene policies — a lesson reinforced during the COVID-19 pandemic.

It's crucial to monitor and correct any mistakes without singling out individuals. By using tools like low-resolution sensors, we can protect people's identities and reduce the risk of unfair judgment, keeping the focus squarely on improving care. This approach is especially valuable, since the root cause of errors, more often than not, extend beyond the individual. As a result, ethical technology deployment of monitoring, AI or otherwise, means ensuring that the efficiencies or insights gained never compromise fundamental rights and well-being.

Figure 1: Patient monitoring device from Hypros.

The approach for continuous patient monitoring hinges on two key innovations:

Non-invasive IoT devices: Hypros developed a novel battery-powered Internet of Things (IoT) device that can be mounted on the ceiling. This device uses low-resolution sensors to capture minimal optical and environmental data, creating a very low-detail image of the scene. The device is designed to be non-invasive, preserving anonymity while still gathering the crucial information needed to detect any meaningful changes in a patient’s environment or condition.
Two-stage AI workflow: Hypros employ a two-stage machine learning (ML) workflow. Initially, they trained a camera-based vision model using AutoML on Vertex AI to label sensor data from simulated hospital scenarios. Next, they use this labeled dataset to train a second model to interpret low-resolution sensor data.

The following sections explain how Hypros implemented these innovations into their patient monitoring solution, and how Google Cloud assisted Hypros in this endeavor.

Low resolution, high information: Securing patient privacy

To address the critical need for patient privacy while enabling effective hospital bed monitoring, Hypros developed a compact, mountable IoT device (see Figure 1) equipped with low-resolution optical and environmental sensors. This innovative solution operates on battery power, facilitating easy installation and relocation to various bed locations as needed.

Figure 2: How a bed with a patient scene is abstracted to low resolution sensor data.

The device's low-resolution optical sensors are effective for protecting patient privacy, they also can make data interpretation and analysis more complex. Additionally, low sampling rates and environmental factors can introduce noise and sparsity into the data, resulting in an incomplete representation of human behavior in the hospital. The combination of low-resolution imaging, limited sampling rates, and environmental noise creates a complex data landscape that requires sophisticated algorithms and interpretive models to extract meaningful insights.

Figure 3: Real-world data: Bed sheets changed by Staff, and Patient gets into bed. This is a “simple” scenario.

Despite these challenges, Hypros’ device represents a significant advancement in privacy-preserving patient monitoring, offering the potential to enhance hospital workflow efficiency and patient care without compromising individual privacy.

Patient monitoring with AI: Overcoming low-resolution data challenges

While customized parametric algorithms can partially interpret sensor data, they have difficulty handling complex relationships and edge cases. ML algorithms offer clear advantages, making AI a vital tool for a patient monitoring system.

However, the complexity of their sensor data makes it difficult for AI to independently learn the detection of critical patient conditions, and thus, unsupervised learning techniques would not yield useful results. In addition, manual data labeling can quickly become expensive as tight monitoring sends readings every few seconds, quickly producing large volumes of data.

To solve these issues, Hypros adopted an innovative approach that would allow AI to learn how to detect scenarios from their monitoring devices with minimal labeling effort. They found that using pre-trained AI models, which require fewer examples to learn a new image-based task, can simplify labeling image data. However, these models struggled to interpret their low-resolution sensor data directly.

Therefore, they use a two-step process. First, they train a camera-based vision model using camera data to produce a larger, labelled dataset.Then, they transfer these labels to concurrently recorded sensor data, which they use to train a patient monitoring model. This unique approach enables the system to reliably detect events of interest, such as falls or early signs of delirium, without compromising patient privacy.

Driving healthcare innovation with Google Cloud

Hypros relied heavily on Google Cloud to build their patient monitoring system, particularly its data and AI services. The first crucial step was collecting useful data to train their AI models.

They began by replicating a physical hospital room environment within their offices. This controlled setting enabled them to simulate various realistic scenarios, gather data, and record video. During this phase, they also collaborated closely with hospitals to ensure that the characteristics specific to each use case were accurately determined.

Next, they trained a camera-based vision model with AutoML on VertexAI to label sensor data. This process was remarkably straightforward and efficient. Within approximately two weeks, their initial AutoML camera-based vision model used for labeling achieved an average precision exceeding 91% across all confidence thresholds. Already impressive, the actual performance was higher as labeling discrepancies artificially lowered the results.

Subsequently, they labeled various video recordings from hospital beds and correlated these labels with their device data for model training. This approach allowed the model to learn how to interpret sensor data sequences by observing and learning from the corresponding video. For training use cases that didn’t incorporate video information, they relied on data or simulation methodologies from their hospital partners.

The speed of development cycles is also a critical competitive advantage. Therefore, they mapped every step in their workflow and model development cycles (see Figure 4) to the following Google Cloud services:

Cloud Storage: Stores all raw data, enabling easy rollbacks and establishing a clear baseline for ongoing improvements.
BigQuery: Stores labeled data for easier querying, and analysis. Easy access to the right data helps them iterate, analyze, debug, and refine their models more efficiently.
Artifact Registry: Hosts their custom Docker images in ETL and training pipelines. Fewer downloads, shorter builds, and better software dependency management provides smoother, more optimized operations.
Apache Beam with Dataflow Runner: Processes large volumes of data at high speed, keeping their pipelines fast and maximizing their development time.
Vertex AI: Provides a unified platform for model registration, experiment tracking, and visualizing results in TensorBoard; training is done with TensorFlow and TFRecords, using customized resources (like GPUs) and easy deployment options simplify rolling out new model versions.

Figure 4: Simple workflow directed graph to highlight technologies used

With Google Cloud’s ability to handle petabytes of data, they know their workflows are highly scalable. Having a powerful, flexible platform lets them focus on delivering value from data insights, rather than worrying about infrastructure.

Further possibilities: Distilling nuanced information

The development of their system has sparked more ideas about ways hospitals can benefit from using sensor data and AI. They see three main areas of care where continuous patient monitoring can help: patient-centric care for better outcomes, staff-centric support to optimize their time, and environmental monitoring for safer spaces.

Some potential use cases include:

People detection: Anonymously detect individuals to improve operations, such as bed occupancy for patient flow management.
Fall prevention and detection: Alert staff about patient falls or flag restless behavior to prevent them.
Pressure ulcers: Monitor 24/7 movement to aid clinical staff in repositioning patients effectively to prevent the development of pressure ulcers (bedsores).
Delirium risk indicators: Track sleep disruption factors like light and noise, which are potential indicators of delirium risk (final correlation requires additional data from other sources).
General environmental analysis: Monitor temperature, humidity, noise, and other environmental data for smarter building responses in the future (e.g., energy savings through optimized heating) and more effective patient recovery.
Hand hygiene compliance: Anonymously track hand disinfection compliance to improve hygiene practices in combination with solutions like the Hypros’ Hand Hygiene Monitoring solution – NosoEx.

Instead of stockpiling sensor data, their system uses advanced AI models to interpret and connect data from multiple streams — turning simple raw readings into practical insights that guide better decisions. Real-time alerts also bring timely attention to critical situations — ensuring patients receive the swift and focused care they deserve, and staff can perform at their very best.

The path forward with patient care

Already, Hypros’ patient monitoring system is gaining momentum, with real-world trials at leading institutions like UKSH (University Hospital Schleswig-Holstein) in Germany. As highlighted by their recent press release, the UKSH recognizes the potential of their solutions to transform patient care and improve operational efficiency. In addition, their clinical partner, the University Medical Center Greifswald, has experienced benefits firsthand as an early adopter.

Dr. Robert Fleishmann, a managing senior physician and deputy medical director at the University Medical Center Greifswald, is convinced of its usefulness, saying:

“The prevention of delirium is crucial for patient safety. The Hypros patient monitoring solution provides us with vital data to examine risk factors (e.g., light intensity, noise levels, patient movements) contributing to the development of delirium on a 24/7 basis. We are very excited about this innovative partnership."

This positive feedback, alongside the voices of other customers, fuels Hypros' ongoing commitment to revolutionize patient care through ethical and data-driven technology.

By harnessing the power of AI and cloud computing, in close collaboration with Google Cloud, Hypros is dedicated to developing privacy-preserving patient monitoring solutions that directly address critical healthcare challenges such as staffing shortages and the ever-increasing need for enhanced patient safety.

Building on this foundation, Hypros envisions a future where their AI-powered patient monitoring solutions are seamlessly integrated into healthcare systems worldwide. The goal is to empower clinicians with real-time, actionable insights, ultimately improving patient outcomes, optimizing resource allocation, and fostering a more sustainable and patient-centric healthcare ecosystem for all.

^{We’d like to thank Tobias Gebhardt, Karsten Fritz, Marcelo Simon, and Felix Eggert from Hypros, and Laura Heidrich, Dr. Stefan Ebener, Benjamin Schantze, and Chris Kremer from Google Cloud for their contributions to this project.}

Accelerating AI in healthcare using NVIDIA BioNeMo Framework and Blueprints on GKE

Tue, 18 Mar 2025 20:00:00 +0000

The quest to develop new medical treatments has historically been a slow, arduous process, screening billions of molecular compounds across decade-long development cycles. The vast majority of therapeutic candidates do not even make it out of clinical trials.

Now, AI is poised to dramatically accelerate this timeline.

As part of our wide-ranging, cross-industry collaboration, NVIDIA and Google Cloud have supported the development of generative AI applications and platforms. NVIDIA BioNeMo is a powerful open-source collection of models specifically tuned to the needs of medical and pharmaceutical researchers.

Medical and biopharma organizations of all sizes are looking closely at predictive modeling and AI foundation models to help disrupt this space. With AI, they’re working on accelerating the identification and optimization of potential drug candidates to significantly shorten development timelines and address unmet medical needs. This has become a significant turning point for analyzing DNA, RNA, and protein sequences, and chemicals, predicting molecular interactions, and designing novel therapeutics at scale.

With BioNeMo, companies in this space gain a more data-driven approach to developing medicines while reducing reliance on time-consuming experimental methods. But these breakthroughs are not without their own challenges. The shift to generative medicine requires a robust tech stack, including: powerful infrastructure to build, scale, and customize models; efficient resource utilization; agility for faster iteration; fault tolerance; and orchestration of distributed workloads.

Google Kubernetes Engine (GKE) offers a powerful solution for achieving many of these demanding workloads, and when taken together with NVIDIA BioNeMo, GKE can accelerate work on the platform. With BioNeMo running on GKE, organizations can achieve medical breakthroughs and new research with levels of speed and effectiveness that were unheard of before.

In this blog, we’ll show you how to build and customize models and launch reference blueprints using NVIDIA BioNeMo platform on GKE

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud AI and ML'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac39405b0>), ('btn_text', 'Start building for free'), ('href', 'http://console.cloud.google.com/freetrial?redirectPath=/vertex-ai/'), ('image', None)])]>

NVIDIA’s BioNeMo platform on GKE

NVIDIA BioNeMo is a generative AI framework that enables researchers to model and simulate biological sequences and structures. It places major demands for computing with powerful GPUs, scalable infrastructure for handling large datasets and complex models, and robust managed services for storage, networking, and security.

GKE offers a highly scalable and flexible platform ideal for AI and machine learning — and particularly the demanding workloads found in biopharma research and development. GKE's autoscaling features ensure efficient resource utilization, while its integration with other Google Cloud services simplifies the AI workflow.

NVIDIA’s BioNeMo platform offers two potential synergistic components:

1. BioNeMo Framework: Large-Scale Training Platform for Drug Discovery AI

A scalable, open-source, training system for biomolecular AI models like ESM-2 and Evo2. It provides an optimized environment for training and fine-tuning biomolecular AI models. Built on NVIDIA NeMo and PyTorch Lightning, it offers:

Domain-Specific Optimization: Provides performant biomolecular AI architectures that can be scaled to billions of parameters (eg: BERT, Striped Hyena) along with representative model examples (e.g., ESM-2, Geneformer) built with CUDA-accelerated tooling tailored for drug discovery workflows.
GPU-accelerated performance: Delivers industry-leading speed through native integration with NVIDIA GPUs at scale, reducing training time for large language models and predictive models.
Comprehensive open-source resources: Includes programming tools, libraries, prepackaged datasets, and detailed documentation to support researchers and developers in deploying biomolecular AI solutions

Explore the preprint here for details.

2. BioNeMo Blueprints: Production Ready Workflows for Drug Discovery

BioNeMo Blueprints provide ready-to-use reference workflows for tasks such as protein binder design, virtual screening, and molecular docking. These workflows integrate advanced AI models like AlphaFold2, DiffDock 2.0, RFdiffusion, MolMIM, and ProteinMPNN to accelerate drug discovery processes. These blueprints provide solutions to patterns identified across several other industry use cases. Scientific developers can try NVIDIA inference microservices (NIMs) at build.nvidia.com and access them to test via a NVIDIA developer license.

The following graphic shows the components and features of GKE used by the BioNeMo platform. In this blog, we demonstrate how to deploy these components on GKE, combining NVIDIA’s domain-specific AI tools with Google Cloud’s managed Kubernetes infrastructure for:

Distributed pretraining and finetuning of models across NVIDIA GPU clusters
Blueprint-driven workflows using NIMs
Cost-optimized scaling via GKE’s dynamic node pools and preemptible VMs

Figure 1: NVIDIA BioNeMO Framework and BioNeMo Blueprints on GKE

Solution Architecture of BioNeMo framework

Here, we will walk through setting up the BioNeMo framework on GKE to perform ESM2 pretraining and fine-tuning.

Figure 2: BioNeMo framework on GKE

The above diagram shows an architectural overview of deploying the NVIDIA BioNeMo Framework on GKE for AI model pre-training, fine-tuning, and inferencing. Here's a breakdown from an architectural perspective:

GKE: The core orchestration platform including the control plane managing the deployment and scaling of the BioNeMo Framework. This is deployed as a regional cluster, and can be optionally configured as a zonal cluster.

Node Pool: A group of worker nodes within the GKE cluster, specifically configured with NVIDIA GPUs for accelerated AI workloads.
Nodes: Individual machines within the node pool, equipped with NVIDIA GPUs.
NVIDIA BioNeMo Framework: The AI software platform running within GKE, enabling pre-training, fine-tuning, and inferencing of AI models.

Networking:

Virtual Private Cloud (VPC): A logically isolated network within GCP, ensuring secure communication between resources.
Load Balancer: Distributes incoming traffic to the BioNeMo services running in the GKE cluster, enhancing availability and scalability.

Storage:

Filestore (NFS): Provides high-performance network file storage for datasets and model checkpoints.
Cloud Storage: Object storage for storing datasets and other large files.

NVIDIA NGC Image Registry: Provides container images for BioNeMo and related software, ensuring consistent and optimized deployments.

Steps

We have published an example to pre-train, fine-tune, and infer an ESM-2 model using BioNeMo Framework on GKE in Pretraining and Fine-tuning ESM-2 LLM on GKE using BioNeMo Framework 2.0 GitHub repo. Here is an outline of the steps for pretraining:

1. Create a GKE cluster

code_block: <ListValue: [StructValue([('code', 'gcloud container clusters create "gke-bionemo-esm2" \\\r\n --num-nodes="1" \\\r\n --location="<GCP region / zone>" \\\r\n --machine-type="e2-standard-2" \\\r\n --addons=GcpFilestoreCsiDriver'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9aad6fdd90>)])]>

2. Add node pool with NVIDIA GPUs

code_block: <ListValue: [StructValue([('code', 'gcloud container node-pools create "gke-bionemo-esm2-np" \\\r\n--cluster="gke-bionemo-esm2" \\\r\n--location="<GCP region / zone>" \\\r\n--node-locations="<GCP region / zone>" \\\r\n--num-nodes="1" \\\r\n--machine-type="g2-standard-2" \\\r\n--accelerator="type=nvidia-l4,count=1,gpu-driver-version=LATEST" \\\r\n--placement-type="COMPACT" \\\r\n--disk-type="pd-ssd" \\\r\n--disk-size="300GB"'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac31f09d0>)])]>

3. Run the pretraining job

code_block: <ListValue: [StructValue([('code', 'kubectl apply -f pretraining/'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac31f0970>)])]>

4. Visualize results in TensorBoard

code_block: <ListValue: [StructValue([('code', 'kubectl port-forward -n bionemo-training svc/tensorboard-service 8080:6006'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac0d29100>)])]>

Open a web browser pointing to http://localhost:8080 to see the loss curves. The details for fine-tuning and inference are laid out in the GitHub repo.

Solution Architecture of BioNeMo Blueprints

The below graphic shows a BioNeMo Blueprint that is deployed on GKE for inferencing. From an infrastructure standpoint, the components used across the Compute, Networking and Storage layer are similar to Figure 2:

NIMs are packaged as a unit with runtime and model-specific weights. Blueprints deploy one or more NIMs using Helm charts. Alternatively, they can be deployed using gcloud or docker commands and configured using kubectl commands. Each NIM needs a minimum of one NVIDIA GPU accessible through a GKE node pool.
Three NIMs—AlphaFold2, DiffDock, and MolMIM—are deployed as individual Kubernetes deployments. Each deployment uses a GPU and a NIM container image, mounting a persistent volume claim for storing model checkpoints and data. Services expose each application on different ports. The number of GPUs can be configured to a higher value for better scalability.

Figure 3: NIM Blueprint on GKE

Steps

We have an example of deploying a BioNeMo blueprint for Generative Virtual Screening at Generative Virtual Screening for Drug Discovery on GKE GitHub repo. The setup steps, such as GKE cluster, node pool, and mounting filestore, are similar to BioNeMo training. The below steps give an outline of deploying the BioNeMo blueprint and using it for inference:

1. Deploy the BioNeMo blueprint

code_block: <ListValue: [StructValue([('code', 'kubectl create -f nim-bionemo-generative-virtual-screening.yaml'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac0d29370>)])]>

2. Use port forwarding to interact with the pod

code_block: <ListValue: [StructValue([('code', 'kubectl port-forward pod/<molmim-pod> 8010:8000 &'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac110da30>)])]>

3. Test MolMIM NIM locally using a curl statement. The output will have the generated molecule.

code_block: <ListValue: [StructValue([('code', 'curl -X POST \\\r\n-H \'Content-Type: application/json\' \\\r\n-d \'{\r\n "smi": "CC1(C2C1C(N(C2)C(=O)C(C(C)(C)C)NC(=O)C(F)(F)F)C(=O)NC(CC3CCNC3=O)C#N)C",\r\n "num_molecules": 5,\r\n "algorithm": "CMA-ES",\r\n "property_name": "QED",\r\n "min_similarity": 0.7,\r\n "iterations": 10\r\n}\' \\\r\n"http://localhost:8011/generate"'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f9ac110d880>)])]>

NVIDIA BioNeMo Blueprints workflows can be adapted to various domain-specific use cases beyond drug discovery. For example, researchers can leverage generative AI models like RFdiffusion and ProteinMPNN in protein engineering to design stable protein binders with high affinity, drastically reducing the experimental iteration cycles.

By integrating modular NIM microservices with scalable platforms like GKE, industries ranging from biopharma to agriculture can deploy AI-driven solutions tailored to their unique challenges, enabling faster insights and more efficient processes at scale.

Conclusion

As we've explored in this blog post, GKE provides a robust and versatile platform for deploying and running both NVIDIA BioNeMo Framework and NVIDIA BioNeMo Blueprint. By leveraging GKE's scalability, container orchestration capabilities, and integration with Google Cloud's ecosystem, you can streamline the development and deployment of AI solutions in the life sciences and other domains.

Whether you're accelerating drug discovery with BioNeMo or deploying generative AI models with NIMs, GKE empowers you to harness the power of AI and drive innovation. By leveraging the strengths of both platforms, you can streamline the deployment process, optimize performance, and scale your AI workloads seamlessly.

Ready to experience the power of NVIDIA BioNeMo on Google Cloud? Get started today by exploring the BioNeMo Framework and NIM catalog, deploying your first generative AI model on GKE, and unlocking new possibilities for your applications.

^{We'd like to thank the NVIDIA team members who helped contribute to this guide, Juan Pablo Guerra, Solutions Architect, and Kushal Shah, Senior Solutions Architect.}

Bupa: Building a cloud-first digital platform for the future of healthcare

Wed, 09 Oct 2024 08:00:00 +0000

Consumer expectations of healthcare have changed. Accustomed to using digital products and services in other walks of life, people now rightly demand a similar experience with their healthcare. At Bupa, we’re committed to delivering on those desires, which is why we are working with Google Cloud to build our new, cloud-first digital customer platform.

Driven by our ambition to be the world’s most customer-centric healthcare company, our Bupa Blua experience provides digital access to care when needed, and the tools to enable customers to be proactive with their health. Using Google Cloud technology, we are expanding the technical foundations to power Blua, allowing us to develop innovative ways to improve our customers’ wellbeing, reduce the cost of healthcare, and deliver the highly personalized experience our customers expect.

Prescribing the right architecture for digital innovation

We are building from a position of strength. Bupa boasts a powerful health database, and with our Bupa app we already offer our customers a wide range of digital services. However, our legacy infrastructure was restricting our ability to innovate quickly. We knew that with the right cloud infrastructure and a robust data ecosystem, we could build on our firm foundation to deliver truly transformative care for our customers.

Google Cloud was the obvious choice. Not only for its market-leading data analytics solutions, AI expertise, and healthcare-specific APIs, but also for its broader commitment to healthcare innovation across Alphabet’s subsidiaries.

The lifeblood of personalized care

At the heart of our new technical platform will be data, and Google Cloud’s market-leading data solutions will be essential in helping us make full use of it, giving us deeper insights into our customers’ needs and preferences.

Connecting all our data and insights with Google Cloud’s advanced AI and machine learning solutions will allow us to deliver a range of benefits for our customers, stakeholders, and the business while ensuring the utmost levels of privacy and transparency.

Among the new services include connecting customers to expert healthcare day or night with digital general practitioner (GP) services; the ability to check symptoms in seconds using our app; faster, more convenient, and cost-effective healthcare services; and joined-up healthcare across our health provision and insurance businesses for better outcomes for our customers.

Taking the first steps towards a healthier future

We are just scratching the surface of the power of our partnership, but we’re already starting to address key questions around how to build and deliver the most effective platform. How do we provide more sophisticated services to our customers at the same time as improving ease of use? And, to consider one specific scenario, how can we capture and connect health data for a customer, such as someone requiring cancer treatment, not only to deliver better treatment but also to improve early detection and prevent the cancer from spreading and becoming more serious?

With Bupa’s healthcare expertise and Google Cloud’s infrastructure and technical excellence, we will answer these questions — and many more like them — as we seek to deliver highly personalized and seamless customer interactions across all channels and improve healthcare outcomes for our members. Our journey with Google Cloud is just beginning. The benefits for Bupa in the UK and our millions of customers will be felt for years to come.

Cyber Public Health: A new approach to cybersecurity

Fri, 19 Jul 2024 16:00:00 +0000

At Google, we believe the approach to cloud infrastructure should be informed, in part, by understanding the relative “health” of the Internet. Defining and measuring these vital statistics can help proactively and systemically identify and address conditions that make the internet unhealthy, unsafe and insecure. Crucially, they can be used to help craft a holistic view of the internet that applies the principles and science of public health to cybersecurity — an emerging field known as Cyber Public Health (CPH).

We're excited to announce our support for the practice of CPH, which can help us understand if our individual efforts organizations take to secure their systems are adding up to a greater overall cyber public health benefit. CPH is about managing the risks the internet faces, which can only be done by looking at the bigger picture. That means going beyond vulnerabilities and incidents, and into practices that work to keep internet-connected systems safe and secure.

For example, one way we can measure CPH is to look at the cumulative effect that patching vulnerable systems has on decreasing the spread of malware and improving global system uptime. To do these types of measurements, organizations need to define, measure, and publicly report the equivalent of common health data, or vital statistics, as is done in public health reporting today. With this larger data context, we can understand the overall health of the internet and use that information to employ practices that work in keeping systems safe.

Getting better comprehensive data

Traditional cybersecurity models often react to individual threats, leaving organizations vulnerable to new and evolving attacks. Existing data is often fragmented, siloed, and difficult to obtain, making it challenging to identify trends, patterns, and risk factors at a population level.

Given that many security compromises are not disclosed, little can be learned collectively about what created a particular vulnerability, how it was exploited, what provided a “cure,” and what can ensure prevention of similar vulnerabilities in the future.

As a community, we lack comprehensive data on the overall health of the internet. We believe that CPH can help us broaden our understanding of the internet’s health because it’s principally about measuring and reporting the practices that have been proven to reduce cyber-risk.

From reaction to prediction to protecting the internet

CPH offers a paradigm shift in cybersecurity. By using data-driven insights and fostering collaboration between stakeholders, CPH can help us build a more secure and resilient digital ecosystem. Google Cloud is committed to supporting this new approach by investing in research, developing innovative tools, and promoting information sharing across the cybersecurity community.

The CyberGreen Institute, a champion of CPH and an organization dedicated to measuring the health of the internet, recently co-hosted a workshop with Google Cloud. Rather than focusing reactively upon treating threats and responding to attacks, the CyberGreen Institute empowers people and organizations to take proactive measures to help them avoid and mitigate cybersecurity issues. “Such approaches are analogous to treating a case of malaria through medicine, while leaving the nearby mosquito swamp untouched or developing cancer treatment technology while paying little attention to the population’s tobacco use,” said Adam Shostack, lead author of the workshop report.

The inaugural Cyber Public Health workshop brought together experts from various fields to discuss the future of CPH. The workshop identified key areas for research, including:

Defining the fundamental units of measurement in CPH (including devices, accounts, and users).
Identifying reliable data sources and addressing privacy concerns.
Developing standardized incident reporting forms and metrics.
Investigating the cybersecurity impact of emerging technologies, including AI.

One area of discussion was the concept of Digital Activities of Daily Living (DADLs). Similar to the approach of measuring the impairment of human physical health by assessing the ability to complete daily, routine activities, DADLs extends that concept to digital lives.

“DADLs represent the critical digital tasks that individuals, organizations, and even nations must perform to maintain a healthy and secure cyber ecosystem. Just as ADLs are crucial for physical well-being, DADLs are essential for modern digital well-being,” Josiah Dykstra, director, Strategic Initiatives, Trail of Bits, wrote in a recent CyberGreen blog.

Google Cloud is actively involved in these research efforts, collaborating with leading organizations and researchers to advance the field of CPH.

What’s next

Cyber Public Health is a promising new approach that, together with ideas like those put forth by the public-private PCAST Cyber-Physical Resilience Strategy, has the potential to revolutionize cybersecurity. Google Cloud is proud to be a part of this movement, and we invite you to join us in building a healthier and more secure internet.

We encourage you to learn more about Cyber Public Health and the work of the CyberGreen Institute. The Workshop Report is here.

Together, we can create a safer digital world for everyone.

Unlocking Medical Insights: Secure AI Analyzes Data

Thu, 11 Jul 2024 16:00:00 +0000

With the widespread adoption of wearable fitness devices and health trackers, biomedical researchers face the challenge of how to glean insights from all this information and translate them efficiently into clinical practice. The latest artificial intelligence (AI) tools and cloud computing can accelerate the collection and analysis of insights by processing massive amounts of data quickly and securely. For Neo Christopher Chung, Assistant Professor of Computer Science at the University of Warsaw and Research Associate in the David Geffen School of Medicine at University of California in Los Angeles, “medical imaging and electronic health records provide new opportunities for scientists and clinicians to improve our understanding of disease classification and prognosis.” For Marieke van Buchem, Ph.D. researcher and Innovation Manager at the Leiden University Medical Center in the Netherlands and Visiting Researcher at the Stanford School of Medicine, “the implementation of clinical AI in healthcare settings has been lagging in the past years, but there is an opportunity to gain insights from the enormous amount of electronic health data records. It’s all about moving AI from the research space into clinical practice.”

Both Chung and van Buchem were members of the 2022 cohort of Google Cloud Research Innovators, a global community of researchers from many different organizations driving scientific breakthroughs with Google Cloud. As Research Innovators, they have access to Google Cloud research credits, technical support, and a network of potential collaborators. Van Buchem says, “I enjoy the exchange of ideas with the other Research Innovators and Google experts on how Google Cloud could innovate healthcare and how I could translate these ideas into actual projects in the hospital.” “It’s been interesting to see other people’s processes, and I get inspiration from other deep learning problems,” says Chung. Both researchers are committed to exploring how AI and cloud computing can make the most advanced cancer research tools more accessible to all researchers, and improve diagnoses and prognoses for cancer patients.

Training algorithms to Improve cancer diagnoses

As the Lead AI/ML principal investigator for INFORM (Interpretability of Deep Neural Networks for Radiomics), a European consortium that develops and trains interpretable deep neural networks to make better biomedical predictions, Chung uses AI to study cancer images like CT and PT scans. The huge datasets in the Cancer Imaging Archive and the National Institutes of Health (NIH) can run to hundreds of gigabytes or tens of petabytes. He explains that current AI models work best for the commonly found images they have been trained on, but for privacy and regulatory reasons medical imagery has been more difficult to collect and analyze. Chung’s goal is to start with the deep learning algorithms trained on natural image datasets, then adapt them to process medical images. “It is hard to understand the inner workings of AI,” he says. “An algorithm can make a correct prediction but we don’t know why it’s accurate. That makes it difficult for doctors and patients to trust it. It’s like a black box.” Chung believes transparency will improve both accuracy and trust in AI applications to healthcare.

Chung and Lennart Brocki at the Computational Biology and Machine Learning group at the University of Warsaw set up a project to run large-scale model training and inference using Google Compute Engine’s Graphical Processing Units (GPUs). They have developed an AI tool that can accurately classify tumors and simultaneously explain its decision-making process. They use Google Colab to quickly prototype or test an idea on a smaller scale. “Having a low barrier to entry is incredible,” Chung says. “That means lower overhead costs and no waiting times, no maintenance. It’s also flexible to scale GPU usage up and down. I really appreciate having Google Research credits to try new ideas.”

Using Natural Language Processing to analyze patient communications

As part of the Clinical AI Implementation and Research Lab (CAIRELab) in Leiden, van Buchem applies Natural Language Processing tools to accelerate and scale up the implementation of clinical AI in hospitals. For her current project, she wondered how patient-generated data could help screen and provide resources for at-risk populations. During six months at Stanford School of Medicine’s Boussard Lab, she and her colleague Anne de Hond, under the supervision of Tina Hernandez-Boussard, piloted a program to identify cancer patients at risk for developing depression. To understand the care process, she interviewed social workers, oncologists, and psychiatrists about their workflow with patients. Then, using Google Cloud’s storage and compute capabilities as well as the open-source Bidirectional Encoder Representations from Transformers (BERT), she trained a large language model on public data and fine tuned it to identify concerning patient messages sent through Stanford’s patient portal.

Van Buchem’s pilot showed that the model was able to distinguish concerning from non-concerning patient messages, and her results were accurate across demographic subgroups. “This project would have taken weeks to conduct on a laptop,” she says. “On Google Cloud, the pretraining only took a few days. The scale and speed were a big advantage. I could iterate faster, and I had extra GPUs, memory–everything I needed in one place.”

Envisioning the future of healthcare

Both Chung and van Buchem continue to refine and advance their projects, and envision a future where cloud computing and AI can transform healthcare research as well as clinical practices. They see cloud computing democratizing access to data and resources for researchers across the world. They see AI developing personalized treatments and better support for patients, leading to improved outcomes. They see big data yielding insights that help healthcare workers fulfill their critical missions. Van Buchem concludes, “we’re talking about huge amounts of multimodal data, just waiting to be used to support healthcare professionals in decreasing the administrative burden, increasing diagnostic accuracy, improving the flow through the hospital.”

To find out how you can get started with generative AI for higher education, download the new 10-step public sector guide. With domain-specific use cases and customer stories from Ed Tech innovators like Varsity Tutors, IBL Education, and more, it offers a comprehensive guide to kickstart your gen AI journey.

Cloud CISO Perspectives: How Google is helping to improve rural healthcare cybersecurity

Fri, 28 Jun 2024 16:00:00 +0000

Welcome to the second Cloud CISO Perspectives for June 2024. In this update, Taylor Lehmann, director, Office of the CISO, shares remarks he made to the National Security Council this month on the steps Google is taking to help rural healthcare networks become more secure and resilient against cyberattacks.

--Phil Venables, VP, TI Security & CISO, Google Cloud

aside_block: <ListValue: [StructValue([('title', 'Get vital CISO Insights with Google Cloud'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac0156d90>), ('btn_text', 'Visit the hub'), ('href', 'https://cloud.google.com/solutions/security/leaders?utm_source=cloud_sfdc&utm_medium=email&utm_campaign=FY23-Q2-global-PROD418-email-oi-dgcsm-CISOPerspectivesNewsletter&utm_content=ciso-hub&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

How Google is helping to improve rural healthcare cybersecurity

By Taylor Lehmann, director, Office of the CISO, Google Cloud

Healthcare organizations have wrestled for decades to protect complex and critical technologies that are vital to their core mission of helping sick people get better. The proper functioning of our society depends on the ability of people to receive timely healthcare, yet cyberattacks against healthcare organizations are making it harder — and the attacks are getting worse.

In the first half of just this year, attacks on hospitals and their suppliers have disabled payment systems, prevented patients from receiving the care they need, and in some cases, have made it unsafe to be a patient physically located inside an impacted care facility. Hospitals and clinics are pushed to the brink, with some being forced to permanently close.

Taylor Lehmann, director, Office of the CISO, Google Cloud

Rural communities across America are especially vulnerable to these threats. Estimates suggest more than 60 million people are served by 1,800 to 2,100 rural hospitals and clinics, many of which are critical access hospitals located more than 35 miles from another hospital.

Thirty-five miles might not seem like a long distance, but a cyberattack can force someone suffering from a catastrophic brain injury to be diverted from their closest hospital to one further away. The first 60 minutes after an injury or other health emergency can be vital to a patient’s survival, enabling diagnosis and rapid medical interventions. If they can’t get the care they need in that “golden hour,” then the likelihood that the patient will not survive the diversion trip from the nearest hospital to another facility increases.

For patients and staff who remain inside an impacted hospital and can’t be moved, their experience changes too. When computers deliver and coordinate care suddenly stop functioning, other services deteriorate. Radiology services needed to diagnose strokes, systems in the NICU that keep very sick babies under constant surveillance (and warm), bedside medication administration systems to ensure proper medication delivery and dosages, and even basic electronic medical records (EMR) for patients have all been degraded or stopped by cyberattacks.

Today, we are bringing these [secure-by-design] technologies to healthcare organizations, some substantially discounted and many others at no cost, to help improve their agility to defeat cyber threats, and mitigate cyber risks that may otherwise undermine their availability.

While clinicians do their best to keep track of everything with paper and pen during a cyberattack that takes down their EMR system, no access to patient medical records can slow or even halt simple procedures that saves lives. We don’t have to imagine these real-world consequences of cyberattacks against healthcare because we’ve seen them happen, repeatedly.

All of this presumes that a cyberattack isn’t impacting multiple medical facilities in the same vicinity, and hopefully, the hospitals to which patients are diverted are capable of treating patients with the same level of care.

The White House, Department of Health and Human Services, the Health Sector Coordinating Council, and others are putting significant effort into identifying systemic challenges, and working with organizations including Google to come up with real and defined solutions to improve cyber resilience for rural health facilities. We’re excited to see this new direction, and we’re here to support communities and health systems.

The Biden-Harris administration published a fact sheet on June 10 summarizing the White House response to these attacks. Recognizing the unique role that healthcare organizations play in their communities, regions, and across the nation, the White House emphasized the public-private partnership needed to better secure hospitals and other healthcare organizations.

As an early innovator and proponent of secure-by-design technology, Google has been working across industries to provide access to and onboarding support to implement the same security tools and practices that keep Google safe to organizations of all types. Today, we are bringing these technologies to healthcare organizations, some substantially discounted and many others at no cost, to help improve their agility to defeat cyber threats, and mitigate cyber risks that may otherwise undermine their availability.

Information sharing is a vital component of securing the healthcare sector. We need better mechanisms to capture and share information that include and surpass threat intelligence.

We support the White House’s efforts in achieving that outcome. We believe organizations, including Google, can help in a few different, unique, and important ways, and we welcome the opportunity to contribute.

1. Secure by design, secure by default

We know that many health systems have acquired and operate technology that was built for interoperability, but not with strong security measures in mind.

At Google, we develop secure by design technologies that have been engineered with security from the get-go, not bolted on afterwards. Fortunately, the U.S. government and other governments around the world have been encouraging and, in some cases, mandating shifts to secure by design and by default technologies. Critical to the security and resiliency of healthcare technology, secure by design and by default encourages four essential principles:

How customers actually use products, even when those uses are inadvertently risky;
How the developer ecosystem can encourage vulnerability and error prevention;
How grounding software in properties that remain consistent even when under attack can strengthen resilience; and
How understandability and assurance can verify those grounding properties, even at scale.

Technology that shows up in a hospital must be secure by design and by default. It must be increasingly easy to maintain, upgrade, patch, and eventually replace when needed. It must not add more complexity to already complex environments. It needs to work safely, after it has experienced an attack, or indeed, during an attack. The makers of these technologies know that the only way to achieve these outcomes is to ensure that protections are built in from the start.

2. Share information on threats, countermeasures, and successes

Information sharing is a vital component of securing the healthcare sector. We need better mechanisms to capture and share information that include and surpass threat intelligence. This includes data-supported conclusions about which practices work, and ensuring that they are informed — but not solely driven by — incidents and failures.

As part of Google’s pursuit of this goal, we have been developing partnerships with multiple information sharing and analysis centers, including the Health ISAC, across more than 10 critical infrastructure sectors — and we plan on doing more. We are eager to support organizations such as the Health ISAC and Sector Coordinating Council continue to get stronger at executing their key function: sharing information.

Google will put our own collaboration and security products into the hands of hospitals and healthcare organizations that need them, most at no or very discounted cost.

We need to reduce barriers to sharing information, too. More organizations should be sharing information at increasing levels of sophistication: It’s just not enough anymore to merely consume it. Organized, rapid intelligence-sharing, and verifiable responses can mean the difference between a successful defense and a vulnerable one.

3. Put Google’s security tools in the hands of hospitals

Google will put our own collaboration and security products into the hands of hospitals and healthcare organizations that need them, most at no or very discounted cost. We are offering products, implementation services, and support to eligible organizations to support their adoption. Organizations interested in more details on the following offerings should email rural-health@google.com.

Let’s take a look at what Google will be providing to healthcare organizations.

Chrome Enterprise Browser and ChromeOS can help health systems safely access and use internet-based and internal technology resources they use to operate their facilities and deliver patient care. Working together, Chrome and ChromeOS offer a more secure alternative than other browser and operating system combinations.

Google Workspace Enterprise Essentials Plus is also included in this program. Google Workspace, which supports compliance with HIPAA, is a collaboration platform that pairs productivity applications (including Docs, Slides, Sheets, and Drive), messaging applications (such as Gmail and Chat), identity platforms (Cloud Identity Premium), and a suite of sophisticated security tools to keep data safe. Workspace helps organizations simplify communication between administrators, clinicians, and patients securely.

4. Grow cybersecurity expertise through education

We also believe in training more cybersecurity professionals. Google.org grants help fund cybersecurity clinics at universities and colleges, which support rural and underserved hospitals in their communities. We are in the process of providing $25 million to 25 U.S. cybersecurity clinics.

We are also helping establish additional clinics at universities and colleges who put cybersecurity-focused students and faculty into their communities, to help them better secure their IT systems. Schools including the Eastern Washington University, Massachusetts Institute of Technology, Rochester Institute of Technology, Tougaloo College, Turtle Mountain Community College, and the University of Texas, are working to secure small, underserved, and rural healthcare systems and public health agencies through these programs.

To support education and training efforts, we’re making courses available from our Mandiant Academy program.

Google Cloud and Mandiant have built a program that up-levels the healthcare industry and key industry partners. Our offer includes:

Mandiant education and training courses

Education and training are crucial to securing rural and underserved hospitals and clinics. To support education and training efforts, we’re making courses available from our Mandiant Academy program.

We will be giving the Health ISAC 20 on-demand training courses, at no charge, that it can disperse to its members. We will also be giving the Health ISAC credits for 10 public, instructor-led scheduled courses that it can distribute to its members. Members will have the opportunity to earn certifications in incident response and threat intelligence following their training or through their own independent study.

Also from our Mandiant Academy, we’ll be offering discounts on three popular courses.

AIM for Health ISAC: In partnership with the Health ISAC, we are offering our Applied Intelligence Mentorship program (AIM). From Mandiant Intelligence, the AIM program is a four-week immersive mentorship designed to develop skilled cyber threat intelligence (CTI) practitioners through direct coaching and practical skills application.
ThreatSpace: Mandiant Consulting is also making its immersive educational experience ThreatSpace available to help incident response teams hone their skills against realistic APT-level attacks in a consequence-free environment.
Digital Forensics and Incident Response Bootcamp: This intensive, 10-day bootcamp teaches the fundamental investigative techniques needed to respond to today’s threat actors and intrusion scenarios. After eight days of classroom learning, students spend two days doing hands-on exercises that take them through adversary activity and the process of responding to a nation-state threat.

This is just the beginning. We’re developing no cost and discounted offerings of these technologies and services for organizations in need. To learn more, please email us at rural-health@google.com.

aside_block: <ListValue: [StructValue([('title', 'Join the Google Cloud CISO Community'), ('body', <wagtail.rich_text.RichText object at 0x7f9ac0156640>), ('btn_text', 'Learn more'), ('href', 'https://rsvp.withgoogle.com/events/ciso-community-interest?utm_source=cgc-blog&utm_medium=blog&utm_campaign=2024-cloud-ciso-newsletter-events-ref&utm_content=-&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

Threat Intelligence news

Cloaked and covert: Uncovering UNC3886 espionage operations: Mandiant has released new research on UNC3886, a suspected China-nexus cyber espionage actor that has targeted prominent strategic organizations on a global scale. Read more.

Now hear this: Google Cloud Security and Mandiant podcasts

From bad IP to trafficking busts, meet the human side of threat intelligence: Threat intelligence is one of those terms whose meaning changes depending on the listener. Brandon Wood, product manager, Google Threat Intelligence, tells hosts Anton Chuvakin and Tim Peacock about what folks are getting wrong about TI. Listen here.
Cloud incident confessions: Top 5 mistakes that lead to breaches: Mandiant consultants Omar ElAhdan and Will Silverstone discuss securing hybrid clouds and how organizations misunderstand their own attack surfaces. Listen here.

To have our Cloud CISO Perspectives post delivered twice a month to your inbox, sign up for our newsletter. We’ll be back in two weeks with more security-related updates from Google Cloud.

Chugai Pharmaceutical: Accelerating drug discovery through AI, machine learning and data analysis

Fri, 07 Jun 2024 16:00:00 +0000

Chugai Pharmaceutical is one of the leading companies in the pharmaceutical industry driving digital transformation (DX) and has been selected as a ‘DX Brand’ by Japan’s Ministry of Economy, Trade, and Industry and the Tokyo Stock Exchange for three consecutive years. We were also named the “Digital Transformation Platinum Company 2023-2025” under the DX Stocks program in 2023. Digital transformation is the key to our operations and helping us break new ground in drug discovery and testing.

Our current focus is on digitizing our Research and Development (R&D) sectors. In the past, developing and testing new drugs involved numerous rounds of trial and error. This process typically incurs a lot of time and costs for pharmaceutical companies, taking from 10 to 15 years just to produce a single drug. Against this backdrop, we are using AI to accelerate innovation and reduce drug discovery times.

Furthering R&D efforts in the cloud

Many pharmaceutical companies are now using machine learning (ML) technology to artificially create antibody drugs and develop more effective drugs. At Chugai, we consider digital transformation as a business-wide operation, with the DX stock selection being one of our key achievements. In particular, we are driving major changes to our R&D infrastructure, such as building our own protein structure estimation system.

Powerful solutions are necessary to accelerate the speed of R&D using AI. To this end, we migrated our IT infrastructure to the cloud. With our previous on-premises approach, it would have taken several months and tens of millions of Japanese yen to start any initiative. Our highly specialized professionals, including data scientists, have now moved to Google Cloud to encourage further innovation.

In 2022, we began a trial run to add Google Cloud to the Chugai Cloud Infrastructure (CCI), a next-generation multi-cloud infrastructure. We implemented the CCI to standardize infrastructures that were individually deployed and optimized as part of the company’s cloud migration. This will play a crucial role in our "Chugai Digital Vision 2030" of becoming a top innovator in providing healthcare solutions that will change society using digital technology, which we set out to fulfill in 2020. The vision encompasses three main strategies: strengthening the digital platform, optimizing all value chains, and undergoing a digital transformation for drug discovery and development.

We are constructing the CCI in stages, and we launched the main cloud platform integration infrastructure as phase one in April 2023. We are currently in phase two and are now moving forward with multi-cloud integration, including Google Cloud. At the same time, we are strengthening our infrastructure as a platform by improving our functionality and operational efficiency. We believe Google Cloud has an advantage over other cloud platforms in terms of its services in the ML and AI fields, as well as its analytics infrastructure. We aim to provide these functions to users in a timely manner as a CCI service.

We also believe that Google Cloud has the capabilities that professionals working in drug discovery at Chugai Pharmaceutical are looking for. While we have not fully implemented Google Cloud, we plan to utilize it to develop and operate AI for drug discovery in the future, starting with DeepMind, a subsidiary of Google, and its 3D protein structure prediction system, AlphaFold2. As AlphaFold2 can now run on Vertex AI, we should be able to integrate it seamlessly within our system.

Although AlphaFold2 can estimate protein structures extremely accurately from sequence information, the amount of resources that the CPU and GPU demand is large, and several issues in procuring resources and implementation still persist.

To address this, we’re developing our own protein structure estimation system. It’s still in the early stages of development, but we aim to create a system that anyone in the company can access, so they can easily infer up to 1,000 protein structures in a day.

Leveraging BigQuery for data analysis and app development

Since each project is handled individually, we aren’t standardizing our ML operations at this time. Instead, we plan to increase the speed of system integration, and the value we provide to users by continuously improving the model with ML and BigQuery for data analysis and app development.

The other benefit of using BigQuery is that we can simply input the data and then analyze it. In addition to powering our AI for drug discovery, we also use BigQuery for web access analysis and application development, and anticipate using it in other aspects of our operations too.

Expanding the use of Google Cloud to Tech Kobo

At Chugai Pharmaceutical, we have a subsidiary called Tech Kobo, which develops applications in-house in a cloud environment.

We are now deploying Google Cloud across the organization incrementally. In the future, Tech Kobo will also be able to apply ML models that are implemented by data scientists on Google Cloud. As integration between systems becomes more seamless with Cloud Run and Cloud Functions, there may be an opportunity to create an API for these models so that more users can use them freely.

Tech Kobo will also tap on Google Cloud to develop its mobile applications, and at the same time, develop its mobile backend as a service (mBaaS) using Firebase, a backend cloud computing stack by Google.

For us, Google Cloud is more than just an infrastructure provider. It also facilitates our team members in producing greater results and honing their skills. Looking ahead, we hope to improve and shape the future of drug discovery, gaining knowledge and skills from different markets around the world.

Introducing Isolator: Enabling secure multi-party collaboration with healthcare data

Thu, 11 Apr 2024 16:00:00 +0000

With more than a decade of experience building AI and machine learning models, Google Cloud understands both the benefits and challenges that await those looking to take advantage of these new technologies. Organizations around the world are now looking to AI solutions to solve some of their thornier problems, and to do so they need a safe, secure way when testing or developing products using important data.

In healthcare and life sciences in particular, generative AI has shown how it can have a significant impact. This includes improving access to information and insights, reducing administrative burdens for clinicians, and accelerating the pace of drug discovery and clinical trials.

Today at Google Cloud Next, we are announcing the availability of Isolator, a solution for secure infrastructure and data processing that adds a vital layer of protection to healthcare data used in collaborations between parties. Isolator can enable multi-party collaboration on work that involves handling raw and unprocessed, sensitive, and regulated data in an isolated, private, and compliant environment on Google Cloud.

Isolator can empower multiple parties to partner on sensitive data collaborations much more easily than before, and we expect cross-industry interest in how Isolator can help spur creative technology solutions. Some examples that Isolator can help with include:

Building custom models: Isolator can help develop custom models using Vertex AI and our foundational models that automate high-toil, administrative tasks such as writing discharge summaries.
Leverage complex data: Isolator can advance adoption of our Medical Image Suite, giving researchers and data scientists the efficiencies to improve workflows, and at a lower cost than keeping data on-premise.
Discovering new treatments: Isolator can be used with our Multiomics Suite to simplify drug discovery and clinical trials by streamlining workflows, enabling safe multi-party collaboration, and cutting development timelines.
Scaling advanced data analytics: Isolator can assist in reducing costs and improving results by moving and transforming data from siloed, on-premise databases to feature-rich and massively scalable analytic and data processing services such as Healthcare Data Engine and BigQuery.
Collaborating on research: Isolator is especially good at helping to build, package, and share data sets for use in model development and research. It can assure the integrity of data, and it can bring transparency into all actions taken against a data set.

Available through Google Cloud’s Consulting Services, customers and their partners can use Isolator to set up an isolated instance within their IT environment that allows them to work with sensitive healthcare data workloads while maintaining privacy, security, integrity and traceability.

It also can aid our customers when migrating their workloads to Google Cloud, and can help establish the secure boundaries necessary to create strong AI and machine learning models by securing data end to end.

At its core, Isolator is an environment built with Chrome Enterprise Premium, Google’s industry-first, widely-adopted Zero Trust technology. Coupled with other features built into Google Cloud, including VPC Service Controls, Chrome Enterprise, and encryption at-rest, in-use, and in- transit, Isolator can provide identity-based access that prevents data from leaving the secured Isolator environments — without requiring customers to set up and deploy special devices, data loss prevention controls, or VPN connections.

Isolator can also work with technology such as Google Confidential Space, which offers encryption for data in use with strong hardware isolation and remote attestation capabilities for even greater control.

Importantly, Isolator also comes built with detection and alerting capabilities that log data access activity, detect security misconfigurations, and report violations. This can help ensure customer security teams have full control over who is granted access to their sensitive healthcare data.

In just a few clicks, customers can provide secure access to data and machine learning resources on Google Cloud for any collaborator, anywhere in the world, and from any device, as long as the customer organization’s security policies are continuously met. If the security of an environment protected with Isolator changes, access to data is suspended until the necessary protections are restored. Isolator is built on top of Google Cloud Services and inherits controls and assurances described on our Compliance page.

At Google Cloud, we believe that sharing our knowledge and capabilities to build AI boldly and responsibly will benefit organizations and society as a whole. That approach helps drive our desire to create solutions such as Isolator. The power that safe and secure collaboration can provide to organizations working in healthcare (and beyond) is transformative and one of the main reasons why healthcare providers, payers, and researchers have put their trust in Google Cloud.

For more information on how to get started with Isolator today, please contact a Google Cloud sales specialist through our Google Cloud Consulting Services

How to set compliance controls for your Google Cloud Organization

Mon, 18 Mar 2024 16:00:00 +0000

Assured Workloads is a modern cloud solution that allows companies to more easily run regulated workloads in many of Google Cloud’s global regions. Assured Workloads can help you ensure comprehensive data protection and regulatory compliance across your Google Cloud Organization. It allows you to apply specific security and compliance controls to a folder in support of your compliance requirements. Assured Workloads supports many compliance programs to create regulated boundaries in Google Cloud.

Assured Workloads in action

Many companies have requirements to meet multiple global compliance standards. For example, if your company must adhere to compliance requirements in more than one geographic region, such as FedRAMP High in the U.S. and General Data Protection Regulation (GDPR) in the European Union, Assured Workloads can help you to easily create regulatory boundaries using a folder structure that meets your needs.

As a best practice, we would recommend placing the data subject to FedRAMP High requirements in one folder, while data subject to GDPR can be processed in a separate EU Regions and Support folder. Each folder serves as a logical boundary that maintains your compliance controls while allowing you to maintain visibility into your data in a single Google Cloud Organization.

Assured Workloads folders supporting EU Regions and FedRAMP High

In some cases, you may want compliance controls to apply to the entire Organization, not just a single folder. For example, your Google Cloud Organization is designed to only process data subject to FedRAMP High requirements, and doesn’t need to adhere to other compliance requirements. Instead of creating a FedRAMP High Assured Workloads environment for each of your folders, we recommend creating a single Assured Workloads environment at the Organization level and treating it as the parent node in your Resource Hierarchy.

Assured Workloads at Organization level

By treating Assured Workloads folder as the parent node, you are enforcing a compliance boundary that applies to the entire Organization: each folder and project created in this hierarchy retains the policies and controls enforced by Assured Workloads.

Assured Workloads can also help with addressing compliance requirements for existing Organizations in Google Cloud running production workloads: you can move your existing folders into the Assured Workloads folder. Before moving any projects to an Assured Workloads folder, we recommend performing a move analysis to uncover any non-compliant resources. With the analyzeWorkloadMove API, you can compare your current configurations to your desired compliance state and determine whether your project is:

Processing data in locations that would be deemed non-compliant in your Assured Workloads folder;
Relying on non-compliant services and features; and
Restricted by Organization Policies that may contradict or are otherwise incompatible with the Policies in your Assured Workloads folder.

You can take the analyzeWorkloadMove API’s findings report and proactively resolve resource violations so that they’re abiding by your compliance requirements. Once you’ve made these changes, you can move the projects to your Assured Workloads and rely on Assured Workloads Monitoring for alerts and updates.

Learn more and get started with Assured Workloads

Google Cloud customers can get started with a free trial of Assured Workloads. You can check out how Assured Workloads helped Iron Mountain’s InSight product achieve and maintain compliance with government standards. And we encourage you to learn more by reviewing these resources:

Healthcare's AI transformation is already underway

Tue, 12 Mar 2024 12:00:00 +0000

AI is making a powerful impact all around us, but few sectors stand to benefit as profoundly as healthcare. Doctors and nurses are already starting to use AI as a capable assistant — getting help with medical notes, for example, or using AI in medical imaging to help with disease detection. This is just the beginning of the transformational journey that will ultimately help all of us be healthier.

From research to real-world applications: Google Cloud advances its healthcare AI tools

At Google Cloud, we build AI that can lighten the load for healthcare professionals while transforming the patient experience, and to accelerate this mission, today, we announced three important advancements: the general availability of Vertex AI Search for Healthcare, now integrated with MedLM and Healthcare Data Engine (HDE); HDE international availability; and new capabilities in MedLM for customers to test.

What's new and why it matters:

Healthcare search gets smarter, and saves time: Vertex AI Search for Healthcare significantly improves how healthcare professionals and other employees find critical information within health records and medical documents. Medical information is traditionally hard to search, given complex vocabulary and abbreviations. With this tool, those nuances are understood – it's there in one intelligent search. Plus, thanks to integration with MedLM, Google Cloud’s family of models fine-tuned for the healthcare industry, Vertex AI Search for Healthcare can generate answers to questions about the patient record, making the information easier to find and digest. With healthcare workers stretched thin, imagine all that time saved when AI summarizes records and pinpoints exactly what a clinician needs to know. Less time wasted means more time focused on patients. (Vertex AI Search for Healthcare is integrated today with HDE, and MedLM integration is available for select early access customers.)
Not just for search, but also for factuality while using large language models: Vertex AI Search uses the organization’s data to search, grounding gen AI outputs in this data to reduce the risks of hallucinations or inaccurate responses. In addition, this tool can cite and link to original, internal sources of the information, giving the user confidence in where information is coming from.
Medical understanding deepens: Vertex AI Search now integrates with MedLM, Google Cloud's family of medically-tuned AI models and tools designed to understand complex medical language. This powerful combo gives even more context-rich answers to healthcare workers' questions.
Healthcare-specific data platform goes global: Healthcare Data Engine (HDE), the platform healthcare and life science companies need to make healthcare data more useful, is now available in countries around the world. HDE enables an interoperable, longitudinal record of patient data, and provides clinical insights in FHIR format, the healthcare industry standard. Since AI is only as good as the data it’s using, this is a critical tool that companies all over the world can now take advantage of.
Building a high quality data foundation becomes easier: HDE has a simplified pay-as-you-go pricing model to better align with the value customers get from the platform. We’ve also made HDE easier to deploy, upgrade, and manage by creating a managed service version of the product. And to make getting data into HDE easier, we’ve worked with Google Research to introduce a new data mapping tool called Data Mapper that brings a graphical mapping interface into the product.
More Google Research coming to Cloud customers: Two new capabilities are coming to MedLM for early customer testing: first, MedLM for Chest X-ray, which can help with classification of chest x-rays for operational, screening, and diagnostics use cases; and second, a task-specific API called Condition Summary, which will be available soon to allowlisted customers.

Beyond the hype: healthcare leaders are transforming with this technology

Organizations like HCA Healthcare, Highmark Health, MEDITECH, Oscar Health, and Telus Health are already leveraging Google Cloud's generative AI tools to drive meaningful change.

HCA Healthcare is collaborating with Google Cloud to streamline documentation processes, easing the workload for doctors and nurses. Highmark Health is exploring ways to enhance personalized care delivery for members through AI-powered insights. MEDITECH has integrated AI into its EHR system to improve data search and summarization capabilities and is now working with Google Cloud to support the auto-generation of clinical documentation at critical points in clinician workflow.

Meanwhile, Oscar Health is leveraging AI to reduce confusion, simplify experiences, and improve satisfaction for consumers and providers. This includes streamlining processes that have historically made the healthcare experience cumbersome, including common friction points like claim denials and referrals. And, Telus Health combined a strong data foundation with search and other gen AI tools to put its healthcare data to work for nearly 70 million lives it serves worldwide.

Vertex AI Search is also driving a significant impact for life sciences companies. Insmed Incorporated, a global biopharmaceutical company on a mission to transform the lives of patients with serious and rare diseases, began collaborating with Google Cloud in 2023 to harness the power of gen AI to reduce the time and increase the efficiency of developing and delivering much-needed medicines.

Using Vertex AI Search with external data, Insmed has transformed the way it uses PubMed, a free and widely-used resource supporting the search and retrieval of life sciences literature. A typical search for information on a disease or area of treatment takes considerable time as users sift through multiple retrieved results. With Vertex AI Search-enabled capability, Insmed users now receive a generative response that curates the most relevant information and gives readers a grounded summarization of what they’re looking for. The result is far more tailored than what a traditional search would yield, and also includes specific articles and links. This contextual approach to search has the potential to create efficiencies throughout the research and development process.

In addition, Insmed’s Market Research group is using Vertex AI Search to rapidly search, retrieve, and summarize information from presentations and finalized research reports. The team can now quickly cull through vast amounts of internal data to answer a specific question and have the appropriate reference at their fingertips. Looking beyond Market Research, Insmed is embarking on a proof of concept to enable enterprise search using all company wide data. powered by Vertex AI Search.

These examples demonstrate how generative AI is moving beyond theoretical benefits into tangible improvements within the healthcare industry. As these advances roll out globally, they hold the potential to transform healthcare organizations and systems worldwide.

Focusing on what matters most

From research to real-world applications, Google Cloud's advancements hold the promise of easing the burdens on healthcare workers and improving patient outcomes. By supporting human expertise with AI-powered tools, we move closer to a future where personalized, data-driven care is accessible to everyone, and where healthcare professionals can focus on what matters most – the well-being of their patients.

A window into protein folding: Lowering the barriers for AlphaFold Inferencing

Mon, 11 Mar 2024 16:00:00 +0000

The open-source tool Vertex AI AlphaFold Inference Pipeline has enabled biotech companies in streamlining protein-folding activities, accelerating their go to market timeline. It addresses key challenges in protein structure prediction by unleashing the power of parallel processing, optimizing compute resources, and scaling to meet high-throughput demands. Furthermore, it ensures reproducibility, lineage analysis, flexibility, adaptability, and seamless integration with upstream and downstream systems – all within Vertex AI as the one-stop platform. With this tool, researchers can unlock new possibilities, make groundbreaking discoveries faster than ever before, and drive end-to-end efficiency in their biotech drug discovery efforts.

However, even with Google Cloud's efforts to make the AlphaFold algorithm more accessible to biotech firms, many bioscience organizations still struggle to integrate this technology seamlessly into their researchers' workflows.

The biggest challenge is this: scientists who obsess over protein shapes aren't usually coding ninjas or cloud wizards. Asking them to wrestle with complicated setups just to get a glimpse of a protein is like asking a chef to build their own oven before they can cook dinner. It's not the best recipe for success (or tasty results).

Solution Overview

To reduce the friction, we are making our Vertex AI AlphaFold Inference Pipeline easier to use, including introducing a user-friendly AlphaFold Portal – think of it like protein modeling for beginners. We empower scientists, irrespective of their prior experience with cloud computing, to derive protein structures with minimal effort. The portal eliminates the need to engage with intricate coding (like Python on a Jupyter notebook), enabling users to focus on protein inference results iterations.

The Google Cloud AlphaFold repository now includes the option to deploy this serverless portal, which offers a streamlined, secure, and centralized way to manage protein folding experiments. Launch new experiments with a single click, simplifying workflows and saving valuable time.

Centralized Pipelines

The portal makes researchers' work more efficient in several ways:

Centralized access: Multiple researchers can access the portal through a single web address instead of running their own Jupyter notebook instances or deploying infrastructure on separate projects.
Streamlined protein folding: Researchers can run protein folding pipeline jobs under their usernames and filter simulation results based on other researchers' work. This allows for easy comparison and fine-tuning.
Enhanced collaboration: Previously, each researcher needed to run their own Jupyter notebook instance to run each protein-folding job. Now, they can collaborate more easily by accessing and comparing simulation results in a centralized location.

1- AlphaFold Portal Dashboard

Consider this dashboard to be the central hub for protein folding endeavors. Users can personalize the display, expertly filter results, and utilize designated link buttons to directly access protein resources. The need to navigate through complex configuration or executions has now been simplified.

Are you prepared to engage in protein folding? With just two clicks, your sequence (in FASTA format) will be processed and simulated. The UI will auto select recommendations for the optimal GPU machine configuration based on the type and size of your protein. However, if you are not satisfied with the suggested settings, you have the option to expand the advanced settings and customize them to your desired specifications.

2 - New Protein Folding

Furthermore, we have integrated a preview function for your protein models. Tapping into an open-source visualization tool, you can now seamlessly explore the intricate molecular structures without leaving the interface.

3 - Protein structure visualization

This tool empowers everyone in your biotech organization to harness the power of protein folding, regardless of their cloud or coding experience. Executing this highly complex and compute intensive workload seamlessly on a streamlined, optimized infrastructure, ensuring efficiency and ease of use.

Getting started

If you're a Google Cloud newbie, no worries! We recommend checking out the Getting Started page to get familiarized with Google Cloud. Then, create a project to house all this protein-folding magic.

To proceed, follow the instructions provided in the open-source Google Cloud AlphaFold repository, accessible via the link. This repository contains convenient, pre-built templates that will assist you in setting up all the necessary components. Kindly note that this part of the process may require some technical expertise. If you encounter any challenges or require guidance, your dedicated GCP representative is readily available to assist you in navigating the complexities of the cloud.

Medical Text Processing on Google Cloud

Wed, 07 Feb 2024 17:00:00 +0000

The FDA has a history of using real world evidence (RWE) as an integral component of the drug approval process. Moreover, RWE can mitigate the need for placebos in some clinical trials. The clinical records that make RWE evidence useful, however, often reside in unstructured formats, such as doctor’s notes, and must be “abstracted” into a clinical structured format. Cloud technologies and AI can help accelerate this process, making it significantly faster and more scalable.

Leading drug researchers are starting to augment their clinical trials with real world data for their FDA study submissions because it saves time and is more cost effective. Once the patient’s care concludes, the vast amounts of historical unstructured patient medical data ends up being a contributor to increasing storage needs. Unstructured data is key and critical in clinical decision support systems. In their original unstructured format, insights need a human to review the unstructured data. With no discrete data points from which insights can be quickly drawn, unstructured medical data can result in increased care gaps and care variances. Simple logic dictates that unassisted human abstraction alone is not fast or accurate enough to abstract all of this patient data. Applied natural language processing (NLP) using serverless software components on Google Cloud provides an efficient way of identifying and guiding clinical abstractors towards a prioritized list of patient medical documents.

How to run Medical Text Processing on Google Cloud

Using Google Cloud’s Vertex Workbench Jupyter Notebooks, you can create a data pipeline that takes raw clinical text documents and processes them through Google Cloud’s Healthcare Natural Language API landing the structured json output into BigQuery. From there, you can build a dashboard that can show clinical text characteristics, e.g., number of labels and relationships. From this, you’ll be able to build a trainable language model that can extract text and be further improved over time by human labeling.

To better understand how the solution addresses these challenges, let’s review the medical text entity extraction workflow:

Document AI for Data Ingestion. The system starts with a PDF file that contains de-identified medical text, such as a doctor’s hand-written notes or other unstructured text. This unstructured data is first processed by Document AI using optical character recognition (OCR) technology to digitize the text and images.
Natural Language Processing. The Cloud Natural Language API includes a set of pretrained models, including models for extracting and classifying medical text. The labels that are generated as part of the output of this service will serve as the “ground truth” labels for the Vertex AI AutoML service where additional, domain specific custom labels will be added.
Vertex AI AutoML. Vertex AI AutoML offers a machine learning toolset for human-in-the-loop dataset labeling and automatic label classification, using a Google model that your team can train with your data, even if team members possess little coding or data science expertise.
BigQuery Tables. NLP processed records are stored in BigQuery for further processing and visualization.
Looker Dashboard. The Looker Dashboard acts as the central “brain” for the clinical text abstraction process by serving visualizations that help the team identify the highest priority clinical documents using metrics like tag and concept “density.”
Python Jupyter Notebook. Use either Colab (free) or Vertex AI (enterprise) notebooks to explore your text data and call different APIs for ingestion and NLP.

The Healthcare Natural Language API

The Healthcare Natural Language API lets you efficiently run medical text entity resolution at scale by focusing on the following optimizations:

Optimizing document OCR and data extraction by using scalable Cloud Functions to run the document processing in parallel.
Optimizing cost and time to market by using completely serverless and managed services.
Facilitating a flexible and inclusive workflow that incorporates human-in-the-loop abstraction assisted by ML.

The following diagram shows the architecture of the solution.

A set of reusable Python scripts that can be run from either a Jupyter notebook or Google Cloud Functions that drive the various stages of an NLP processing pipeline, which converts medical text to structured patient data and a Looker dashboard that acts as the decision support interface for teams of human clinical abstractors.
A set of Google Cloud Storage Buckets to support the various stages of data processing (illustrated below).

Two BigQuery tables, called “Entity” and “Document,” in a dataset called “entity,” are created as the data model for the Looker dashboard.

A Vertex AI dataset used for human-in-the-loop labeling by clinical abstractors and to send labeling requests to the Google Vertex AI Labeling Team for added flexibility and scale.

A Looker dashboard that displays the stack-ranked documents to be processed in order by the human abstractors based on a custom “density” metric, which is the number of data elements (labels) found in those documents. This dashboard will guide the human abstractors to look at the sparsely labeled documents first and let Google’s NLP do the heavy lifting.

A list of documents, by density score, helps human abstractors know which documents need a lot of work versus only a light review.

This Look (view) shows the coded medical text that was mapped to the UMLS clinical ontology by the Google Healthcare Natural Language API.

This Look (view) shows the entity mentions, including the subject of each mention and its confidence score, allowing for loading into a biomedical knowledge graph for further downstream analysis.

This Look (view) shows the entity mentions found in the raw document text.

Future Topics and Next Steps

This demo loaded the entity and document metadata into BigQuery and Looker but didn’t load the rich relationships that come out-of-the-box from the Healthcare Natural Language API. Using those relationships, it is possible to create a biomedical knowledge graph and explore the pathways between disease, treatment, and cohorts, and to help generate new hypotheses linking these facts.

We created a barebones dashboard with Looker. However, Looker has rich functionality, such as the ability to push to channels like chat when a document is available for review or to visualize the patient as a medical knowledge graph of related entities or embedding ML predictions right in the Looker LookML itself. This dashboard should be considered just a starting point for Looker powered clinical informatics.

To learn more about the Healthcare Natural Language API, please visit our product page. To try it yourself for free, please visit this demo link.

For help with loading this example medical text into a Vertex AI dataset for labeling, please contact the Google Cloud Biotech Team.

^{Data Privacy
No real patient data was used for any part of this blog post. Google Cloud’s customers retain control over their data. In healthcare settings, access and use of patient data is protected through the implementation of Google Cloud’s reliable infrastructure and secure data storage that support HIPAA compliance, along with each customer’s security, privacy controls, and processes. To learn more about data privacy on Google Cloud, check out} ^{this link}^.

Building a Clinical Intelligence Engine using MedLM augmented Clinical Knowledge Graphs

Tue, 23 Jan 2024 17:00:00 +0000

Artificial intelligence (AI) has immense potential to empower clinicians by synthesizing patient data and medical knowledge to provide valuable insights. At Google, we understand how AI can benefit the clinician community. We collaborated with Apollo 24|7, the largest multi-channel digital healthcare platform in India, to build their Clinical Intelligence Engine (CIE). This CIE can be used to generate evidence-based insights from the wealth of knowledge captured in their millions of de-identified clinical interaction data points. This solution is designed as an assistive tool to empower clinicians by augmenting their abilities, thereby providing better patient care.

Apollo 24|7’s solution includes multiple approaches. It uses Med-PaLM 2, which is now part of our newly released MedLM and clinical knowledge graph-based models.

Dataset used

We used the MIMIC IV dataset, which is a publicly available electronic health record dataset that includes free-text discharge notes.

Evaluation metrics

Below are the three key metrics that were used for evaluating the experiments:

Modified MAP@K: Mean Average Precision (MAP) is calculated for the top K candidates, in order to judge a recommender-like system. We calculate the average precision for each value up to K in an iterative fashion.
Precision Score: The precision score helps us determine how many false positives are present in the predictions, as compared to the true positives.
Capture Rate: The capture rate metric helps us empirically determine how many of the actually present values in the ground truth were identified correctly by the model. In a way, it is similar to recall.

Experimentation

We experimented with multiple models, ranging from Google Research’s large language model (LLM) called MedLM, to graph models like personalized pagerank, and traditional models like kNN (K-nearest neighbor) and two-tower models.

In this blog, we explain our approach using Personalized Pagerank augmented by MedLM.

Step 1: Build a clinical knowledge graph

We built a comprehensive clinical knowledge graph in Neo4j, which is a graph database.
First, we use Google’s PaLM 2 model and the Vertex Healthcare Natural Language API to extract medical concepts from the unstructured medical text in the form of discharge notes.
These clinical entities include concepts like medicines, dosage, medicine frequency, medicine duration, and symptoms. We link them to create the right relations. The extracted entities are ingested into Neo4j as the graph database.

Fig 1: Clinical knowledge graph

We create a bipartite knowledge graph with only two types of nodes - Admissions & UMLS. The UMLS (Unified Medical Language System) codes are generated by the response of the Healthcare NL API.
In order to retain the context of the entities, we extract the sections (like Symptoms, Allergies, Past Medical History and so on) from the discharge summaries. The medicine, Paracetamol, present in the Medication or Allergy sections means two different things even though the UMLS code is the same. Hence, we create an intermediate node SUMLS, which is a combination of UMLS+Section.

Fig 2: Section Details in clinical knowledge graph

Once we created this knowledge graph, we experimented with the different graph algorithms.

Step 2: Build personalized Pagerank model using GraphSage and Optuna optimization

Based on the clinical knowledge graph created in Neo4j, the baseline Pagerank model was built.
In order to consider the node attributes, the node embeddings were used to represent the similar nodes with nearby vectors and the kNN algorithm was used to connect the top 10 most similar admissions.
Lower weights were given to the edges of popular nodes based on their degree in order to reduce their popularity effect.
Next, a GraphSage model was trained to create the neighborhood embeddings of the nodes present in the bipartite knowledge graph.
For optimizing the weights of the edges, Optuna was used.

Step 3: MedLM augmented results

MedLM harnesses the power of Google’s MedLM, and is aligned to the medical domain to more accurately answer medical questions. It can be used to facilitate rich, informative discussions, answer complex medical questions, and find insights in complicated and unstructured medical texts. It is also used to help draft short- and long-form responses and summarize documentation and insights from internal data sets and bodies of scientific knowledge.

In our experiments, we fed the output of the Personalized Pagerank model to MedLM as context and generated the final response, which had a higher accuracy.

We observed that the final responses generated by this approach using MedLM and clinical knowledge graphs were grounded in factuality and are far more accurate by reducing the false positives and boosting the true positives.

“This solution built on MedLM augmented with a clinical knowledge graph can analyze a patient's medical records and generate insights on relevant medications, laboratory evaluations, medical procedures, and potential diagnoses for the clinician to review. By generating these evidence-based insights, this gen AI solution aims to enhance the clinical workflows, reduce errors, and improve patient outcomes. And it is super important to understand that this is just the tip of the iceberg in terms of the AI’s capabilities where it is so powerful, but yet always assistive to the clinicians.” - Abdussamad M, Engineering Lead at Apollo 24|7.

This solution is not intended to replace the clinician's expertise but rather to augment the clinician’s skills and experience.

Fast track end-to-end deployment with Google Cloud Consulting (GCC)

The partnership between Google Cloud and Apollo 24|7 is just one of the latest examples of how we’re providing AI-powered solutions to solve complex problems to help organizations drive the desired outcomes. With Google Cloud Consulting (GCC), Apollo was able to perform repeated iterations and experiments to build the final solution, thereby empowering the business. Apollo entrusted GCC to collaborate with their teams to build the state of the workflows for their business requirements. The GCC portfolio provides a unified services capability, bringing together offerings across multiple specializations, into a single place. This includes services from learning to technical account management to professional services and customer success. See Google Cloud Consulting’s full portfolio of offerings.

^{Disclaimer: MedLM is still in the preview phase in India and it is not approved for production use}

Nuclera aims to accelerate drug discovery with Google DeepMind AlphaFold2 on Vertex AI

Mon, 18 Dec 2023 17:00:00 +0000

Nuclera, a UK and US-based biotechnology company, is collaborating with Google Cloud to serve the life science community, marrying Nuclera’s rapid protein access benchtop system with Google DeepMind’s pioneering protein structure prediction tool, AlphaFold2 (ref 1) served on Google Cloud’s Vertex AI machine learning platform.

With proteins representing 95% of drug targets, the demand to obtain multiple active protein variations to aid in drug discovery is constantly increasing. In particular, reliable protein structure prediction is a prerequisite for compound/biologics lead development.

The breakthrough AI tool, AlphaFold2 (released by DeepMind in 2021), has thrilled the structural biology and drug discovery communities in recent years by taking a huge leap forward in protein structure prediction accuracy (ref 2).

This coming together of technologies from Nuclera and Google presents a new integrated system for drug developers to optimize protein construct design to accelerate their drug discovery process. High quality structures in minutes to hours will soon be a reality, enabling laser-guided protein design. Additionally, reliable structures for proteins thought “impossible” to characterize experimentally will be made accessible.

Making protein that matters

Accessibility of proteins for lab-based research is fundamental to drug discovery, and is notoriously difficult and expensive to achieve, meaning time and resource limitations are imposed on research potential.

Nuclera is motivated to better human health by making proteins accessible, enabling life science researchers to obtain active proteins from DNA through its benchtop eProtein Discovery system (see Figure 1). Nuclera’s technology integrates cell-free protein synthesis and digital microfluidics on Smart Cartridges, allowing rapid progress on protein projects through an automated, high-throughput, benchtop protein access system.

Click here to view video

How AlphaFold2 fits within Nuclera - guided protein design

Widely hailed as a breakthrough in biological research and a leap in the development of vaccines and synthetic materials, AlphaFold2 is an AI model developed by DeepMind for predicting the 3D structure of a protein based on its 1D amino acid sequence.

AlphaFold2 running on Google Cloud’s Vertex AI is set to become an integral feature in Nuclera’s cloud based software, to improve the quality and obtainability of proteins. Currently Nuclera’s cloud software allows their customers to make informed decisions from expression and purification screen results to identify optimal protein constructs to scale up as well as optimal conditions to scale up proteins. The integration of AlphaFold2 into the eProtein Discovery Software increases the quality of constructs screened on the system by offering an addition in silico filter during the experiment design phase, which translates to a higher probability of identifying a truly optimal target protein on which to build discovery programs. Furthermore, AlphaFold2 will help eProtein Discovery users gain deep insights into possible target protein constructs, including any impacts on drug interactions, structural features, and folding.

Implementation of Alphafold2 on Vertex AI pipelines

While the immense power of the AlphaFold2 algorithm is undeniable, it is important to note that Alphafold2 requires serving infrastructure and an operational model.

Generating a protein structure prediction is a computationally-intensive task. Running inference workflows at scale can be challenging — these challenges include optimizing inference elapsed time, optimizing hardware resource utilization, and managing experiments.

The Vertex AI solution for AlphaFold 2 is designed for inference at scale by focusing on the following optimizations:

Optimizing inference workflow by parallelizing independent steps.
Optimizing hardware utilization (and as a result, costs) by running each step on the optimal hardware platform. As part of this optimization, the solution automatically provisions and deprovisions the compute resources required for a step.
Describing a robust and flexible experiment tracking approach that simplifies the process of running and analyzing hundreds of concurrent inference workflows.

Nuclera will use the Vertex AI platform as a foundation for a scalable and resource-efficient AlphaFold pipeline, as well as other Google Cloud services to expose the pipeline through an API and integrate it with its eProtein Discovery system.

What’s involved with the implementation set-up?

The first objective that AlphaFold2 and Nuclera will achieve kicks off with creating a scalable API service that accesses an execution of AlphaFold2 in Google Cloud. Second, an analytics dashboard will be built which allows users to visually and quantitatively compare predicted 3D structures for protein variants. Third, a protein of interest (POI) recommendation feature will propose possible synthetic protein variants (isoforms, truncations, mutations, orthologs or fusions) to customers using intelligent selection algorithms, taking into account various constraints such as computationally generated scores or conserved domains.

eProtein Discovery/AlphaFold2 - its application significance

The 3D structural insights provided by AlphaFold2 will enable Nuclera and its customers to optimize their protein variation synthesis process and gain deeper insights into the interactions between residues and the 3D folding protein structure.

eProtein Discovery customers worldwide will benefit from the composite predictions delivered in the AlphaFold2 module in the eProtein Discovery Software to build a clearer understanding of their proteins, making faster informed decisions that will ultimately economize on time taken for progress in academic research and drug discovery success.

Shweta Maniar, Global Director, Healthcare & Life Sciences Solutions, Google Cloud,commented that, “AlphaFold2 integrated with Nuclera's eProtein Discovery System is a really exciting demonstration of its practical use in drug discovery, enabling researchers to rapidly and efficiently design and produce proteins with the desired structure and function..”

In partnership with Google Cloud and the awesome capabilities of AlphaFold2, we’re excited to be pioneering AI/ML-assisted drug discovery tools, which we believe will bring forth next-generation therapies at a greater pace than ever before. To learn more and to try out this solution, check our GitHub repository, which contains the components and universal and monomer pipelines. The artifacts in the repository are designed so that you can customize them. In addition, you can integrate this solution into your upstream and downstream workflows for further analysis. To learn more about Vertex AI, visit the product page.

^References

^{1. Jumper, J., Evans, R., Pritzel, A. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).} ^{https://doi.org/10.1038/s41586-021-03819-2}^{2. Ghadermarzi, S., Li, X., Li, Mi, et al. Sequence-Derived Markers of Drug Targets and Potentially Druggable Human Proteins. Front. Genet., 10, 1-18 (2019).}^{https://www.frontiersin.org/articles/10.3389/fgene.2019.01075/full}