AWS for Industries

Top-10 re:Invent 2023 Announcements Important for Healthcare and Life Sciences

Occasionally, new technologies emerge that fundamentally alter how businesses operate. But without the appropriate tools—controls, safeguards, and user experiences—the promise of new technologies can get stuck.  In the past year, many have been captivated by the potential of Generative AI to revolutionize drug discovery, precision medicine, and healthcare delivery. But what tools will turn the adoption flywheel for the industry?

AWS, born with a spirit of reinvention, has a proven track record of transforming industries. We recognize the importance of aligning reinvention with a broad and deep toolkit that fits customer needs at different levels of the technology stack. These are essential elements for translating exciting technological advancements into tangible productivity.

At Re:Invent 2023, AWS unveiled new tools that harness the power of Generative AI, which will enable reinvention in the Healthcare and Life Sciences industries. These tools empower builders and business users to leverage Generative AI at every stage, from an AI assistant for your enterprise data, to application orchestration agents, to multi-model evaluation, to fine-tuning and continuous training — even with no-code. These are packaged in toolkits that come with operational safeguards, and integration with tools the industry uses to today like IDEs, business intelligence dashboards, and directly in the AWS Console.

We believe Generative AI is only as useful as your data foundation to support it, populated by your private healthcare and life sciences datasets that are multi-modal and diverse. That’s why along with the AI tools introduced this year, AWS has matured its offerings even further in data storage, data governance, and data collaboration to meet the demands of new AI applications.

The services launched this year will dramatically shorten the time it will take to apply Generative AI to evolve healthcare provider workflows, patient experiences, medical imaging diagnostics, the life sciences R&D process, clinical trials, biomanufacturing, and commercialization. In fact, we heard from two industry leaders in Life Sciences how they were already using Generative AI on AWS to improve business functions, research, and engagement.

Lidia Fonseca, chief digital and technology officer at Pfizer, joined CEO Adam Selipsky on stage to discuss how the company uses artificial intelligence (AI) and Amazon Web Services (AWS) to achieve the scale to treat more than 1.3 billion people with medicines and vaccines in 2022. Fonseca shows how Pfizer centralized its data, cultivated strong AI talent, and built a secure global foundation in the cloud while saving tens of millions of dollars annually. Pfizer and AWS created the Scientific Data Cloud, which aggregates data from hundreds of laboratory instruments to make it simpler and faster for scientists to search the data. On AWS, Pfizer built VOX, a generative AI solution that uses large language models from Amazon SageMaker and Amazon Bedrock, to accelerate research, predict product yield, and help it deliver more medicines to patients. Watch the keynote presentation on-demand, and review the top three take-aways from the session in this blog.

We also heard from Marc Berson, Chief Innovation Officer of Gilead, who joined Shaown Nandi, AWS Director of Technology for Industries and Strategic Accounts, to discuss how generative AI is helping his organization accelerate new therapeutic development. Watch the innovation session video on-demand.

Top-10 Launch Announcements important for HCLS

1. AWS HealthScribe automatically creates clinical notes from patient-clinician conversations using generative AI. Transcribe patient visits, generate preliminary clinical notes, and extract insights that help clinicians to quickly revisit highlights of their patient visits and easily accept, reject, or edit content suggestions. Every AI-generated summary statement comes with traceable transcript references that make it easier for clinicians or scribes to quickly verify accuracy and locate the source of the insight. Simplify implementation with an integrated conversation and generative AI service that is specifically trained for the healthcare industry. Read more in the blog announcement.HealthScribe joins our growing portfolio of healthcare and life sciences-focused services including AWS HealthOmics,  AWS HealthImagingAWS HealthLake, and Amazon Comprehend Medical. 

In fact, we saw Werner Vogels, CTO of AWS, feature AWS HealthImaging and how it can be used to improve Radiologists workflows in his keynote. Watch the keynote session on-demand here.

2. Amazon Bedrock Agents enable generative AI applications to execute multistep tasks across company systems and data sources. Across healthcare & life sciences this opens up the opportunity for automating tasks that require enterprise knowledge such as those in health systems, research databases, clinical trial systems, and commercial databases, and provides orchestrated access to those systems. For researchers, this can be used to carry out a broad internal search for prior studies at an organization before embarking on a new research project. For clinical labs and providers this can be used to enroll study members or patients which requires interacting with multiple transactional software systems. For biomanufacturers, this can be used to automate multi-step supply chain work orders across different systems. For Payors, this can be used to develop applications that automate multi-step approval processes. Read more in the blog announcement.

Bedrock Agents work in concert with the new Bedrock Knowledge Bases to create applications that synthesize, summarize, and recommend information using Retrieval Augmented Generation (RAG). Read more in the blog announcement. For data science teams that want to build custom models to reflect the unique data modalities in healthcare, biomedicine, biology, and chemistry, Bedrock Fine Tuning and Continued Pre-training simplifies the process. Read more in the blog announcement.

Additionally, Amazon Bedrock has added new models that can be accessed via a single API. Anthropic’s Claude 2.1 enables the submission of biomedical or research prompts of up to 200,000 tokens in length (about a 500 page document) Read more in the blog announcement. Titan Multimodal Embeddings model lets you prompt with diverse biomedical data all at once, such as a full set of medical images, lab test results, and medical records. Read more in the blog announcement.

3. Amazon Q (Preview) is your AI assistant designed to understand your business, data, code, and operations, with enterprise controls that keep model and data secure to your organization. For healthcare and life sciences business users, Amazon Q connects to your private datasets or enterprise software and can conduct tasks like analyzing clinical observations from trials, discovering trends across troves of R&D data, creating summaries from manufacturing records, and preparing for compliance investigations. Read more in the blog announcement.

For business analysts, Amazon Q in QuickSight can generate compelling stories by examining data, surfacing key insights into executive summaries, and confidently answering questions of data not answered by dashboards Read more in the blog announcement. For developers and IT professionals, Amazon Q helps you get started building applications on AWS, researching best practices, resolving errors, and getting assistance in coding new features for your applications. Read more in the blog announcement.

4. New generative AI capabilities for Amazon DataZone (Preview) HCLS customers have large, complex, and growing data estates across research, clinical, manufacturing, and commercial domains. With the changes in businesses through trends, mergers, acquisitions, and divestments, metadata and data understanding is lost. Manually creating this metadata can be a cumbersome and expensive task. AI recommendations for descriptions in DataZone uses generative AI to identify data tables and columns required for analysis, which enhances data discoverability. This enables data consumers (such as data analysts, data engineers, and data scientists) to have more contextualized data at their fingertips to inform their analysis. The auto-generated descriptions enable a richer search experience, as search results are now also based on detailed descriptions, possible use cases, and key columns. Read more in the blog announcement.

5. AWS Clean Rooms ML and Differential Privacy (Preview) Apply machine learning with your partners without sharing underlying data. We introduce the first model specialized to help companies create lookalike population segments. With AWS Clean Rooms ML lookalike, you can train your own custom model, and you can invite partners to bring a small sample of their records to collaborate and generate an expanded set of similar records while protecting everyone’s underlying data. Today this is launched for marketing use cases, and in the coming months, we will release a healthcare model. Users can get started by experimenting with the service today. Read more in the blog announcement. Additionally, AWS Clean Rooms Differential Privacy is a new fully managed capability to help you prevent the re-identification of your users. It obfuscates the contribution of any individual’s data in collaboration insights. Read more in the blog announcement.

6. NVIDIA brings BioNeMo to AWS: NVIDIA BioNeMo, a generative AI platform for drug discovery, is available now on Amazon SageMaker. This enables pharmaceutical companies to speed up drug discovery by simplifying and accelerating the training of models using their own data. Read more in the blog announcement. NVIDIA also now offers MONAI as Hosted Cloud Service. With NVIDIA MONAI cloud APIs, solution providers can more easily integrate AI into their medical imaging platforms, enabling them to provide supercharged tools for radiologists, researchers, and clinical trial teams to build domain-specialized AI models. Read more in the blog announcement.

7. No-Code fine tuning of FMs and natural language data preparation in SageMaker Canvas. Healthcare and life sciences organizations have their own sets of unique vocabulary that generic models are not trained on, and at the same time many customers lack dedicated data science teams to support fine tuning. This new capability in Amazon SageMaker Canvas bridges this gap effectively in a no-code application. SageMaker Canvas now performs fine tuning and evaluation of models without the need to write code. Read more in the blog announcement. Additionally, HCLS data analysts and data scientists who want to explore and prepare their data without using code, can now use Canvas’ FM-powered natural language instructions for data exploration, analysis, visualization, and transformation. Read more in the blog announcement.

If the goal is to build an entirely new foundation model based on healthcare or life sciences data, Amazon SageMaker HyperPod is a purpose-built infrastructure for distributed training at scale. SageMaker HyperPods provide managed infrastructure for distributed training which leads to faster and more cost-effective FM training. This provides monitoring features as well, which actively monitor the cluster health and provide automated node and job resiliency by replacing faulty nodes and resuming model training from a checkpoint. Read more in the blog announcement.

8. Amazon Neptune Analytics. Healthcare and life sciences data teams are building knowledge graphs to understand relationships between their data and do correlative studies. Storing data in graphs and performing analytics on it used to be two separate steps that required separate science tools, needed intricate pipelines, and was challenging to operate. Now, Neptune Analytics provides a single tool for storing the graph and then performing analytics, with an 80x speed improvement versus previous AWS solutions. Read more in the blog announcement.

To build knowledge bases based on other types of data stores, the new vector engine for Amazon OpenSearch Serverless, vector search for Amazon DocumentDB, and vector search for Amazon MemoryDB for Redis makes it easy for you to build RAG applications without needing to manage the underlying vector database infrastructure.  Read more in this series of blogs:  Blog 1 | Blog 2 | Blog 3

9. Advancements for HCLS Computing

EC2 Capacity Blocks for ML allow research and data science teams to reserve the use of Amazon EC2 P5 instances, for a future start date. This will enable healthcare and life sciences data science teams to conduct testing and fine tuning of LLMs with reliable budgetary and timing information. Read more in the blog announcement.

Graviton4 provides up to 30% better compute performance, 50% more cores, and 75% more memory bandwidth than current generation Graviton3 processors, delivering the best price performance and energy efficiency for workloads such as Electronic Health Records (EHR) systems. Equipped with brand-new Graviton4 processors, the new Amazon EC2 R8g (Preview) instances will deliver better price performance than any existing memory-optimized instance. The R8g instances are suitable for your most demanding memory-intensive workloads: big data analytics, high-performance databases, and in-memory caches. Read more in the blog announcement.

Trainium2 is designed to deliver up to 4x faster training than first generation Trainium chips and will be able to be deployed in EC2 UltraClusters of up to 100,000 chips, making it possible to train foundation models (FMs) and large language models (LLMs) in a fraction of the time, while improving energy efficiency up to 2x. Read more in the blog announcement.

10. Advancements for HCLS Storage

Amazon S3 Express One Zone speeds up and reduces the cost of HPC workloads that need fast S3 object access. This is a new storage class that delivers the performance needed by your most demanding, compute-intensive HCLS applications such as genomics, image processing, simulations, and machine learning. With durable, single digit millisecond latency, customers can co-locate storage in the same Availability Zone as their EC2, EKS, and ECS compute resources. For workloads that still need POSIX permissions, Amazon FSx for Luster is still a great alternative.  Read more in the blog announcement.

Amazon FSx for NetApp ONTAP scale-out file systems.  Many HCLS customers are configuring, running, and scaling NetApp ONTAP deployments to provide their enterprise file store. Now they can launch and run fully managed, fully-featured ONTAP file systems in the cloud. Read more in the blog announcement.

AWS EBS Snapshots Archive is low-cost, long-term storage tier meant for your rarely-accessed snapshots that do not need frequent or fast retrieval, allowing you to save up to 75% on storage cost Read more in the blog announcement. Amazon EFS Archive is a new storage class that is cost-optimized for long-lived file data that is accessed a few times a year or less. Read more in the blog announcement.

Watch all re:Invent Breakout Session Videos on-demand

Healthcare Session Videos:

  • HLC202 | Reimaging healthcare delivery by migration critical workloads- featuring Geisinger
  • HLC204 | Improving patient outcomes using generative AI in healthcare – featuring UC San Diego Health
  • HLC305 | Building a medical research platform with AWS HealthOmics – featuring Stanford
  • AMZ204 | Beyond the EHR –Delivering timely, accessible care with One Medical – featuring One Medical
  • CON320 | Building for the future with AWS serverless services – featuring Nationwide Children’s Hospital
  • AIM213 | Enhance your document workflows with generative AI – featuring Centene
  • IMP205 | Modern digital experiences to accelerate mission impact – featuring National Marrow Donor Program
  • BIZ103 | How the U.S. Army uses AWS Wickr to deliver lifesaving telemedicine – featuring NETCCN
  • IMP208 | Using data to prevent heart disease and sudden cardiac death – featuring Memorial Hermann Health System

Life Sciences Session Videos:

  • LFS202 | Accelerating life sciences innovation with generative AI on AWS – featuring Gilead
  • LFS203 | Building a life science data strategy to accelerate insights – featuring Johnson & Johnson
  • API310 | Scale interactive data analysis with Step Functions Distributed Map – featuring Vertex Pharmaceuticals
  • BSI203 | Enhance your applications with Amazon QuickSight Embedded Analytics – featuring Honeywell Life Sciences
  • ANT331 | Build an end-to-end data strategy for Analytics and Generative AI
  • NTA204 | Accelerate your digital transformation with a robust cloud foundation – featuring Bristol Myers Squibb
  • NTA213 | 0 to 25 PB in one year – featuring Caris Life Sciences
  • AIM215 | Omics Innovation with AWS HealthOmics – Amgen’s Path to Faster Results – featuring Amgen
  • AIM222 | Amazon Lex reshapes CX with conversational workflows and generative AI – featuring Abbott

Read some of the top news from last week for HCLS:

It was a busy week, and we’re excited to bring you more exciting announcements, videos, and recaps of re:Invent 2023. In the meantime, to learn more about generative AI in Healthcare and Life Sciences visit, 

Kelli Jonakin, Ph.D.

Kelli Jonakin, Ph.D.

Kelli Jonakin is the Worldwide Head of Marketing for Healthcare, Life Sciences, and Genomics Industry verticals at AWS. She comes with a background in pharmaceutical research, with a special focus on development and commercialization of biologics. Kelli received her Ph.D. in Pharmacology and Systems Biology from the University of Colorado, and received an NIH post-doctoral fellowship grant to study Biochemistry at the University of Wisconsin-Madison.

Lee Tessler

Lee Tessler

Lee Tessler, Ph.D. is a Principal Technology Strategist for the Healthcare & Life Sciences industry at AWS. His focus is on cloud architectures for modernizing R&D, clinical trials, manufacturing, and patient engagement. Prior to joining AWS, he launched products in the areas of bioinformatics, drug discovery, diagnostics, lab instruments, and pharma manufacturing. Lee holds a Ph.D. in computational biology from Washington University in St. Louis and Sc.B. from Brown University.

Chris McCurdy

Chris McCurdy

Chris McCurdy is a Global Solutions Architect & Manager with over 20 years of hands-on architecture, development and team lead experience focusing the past 10+ years on the success of the Healthcare and Life Sciences industries. He is an evangelist of GxP/HIPAA compliance and IoT & AI/ML Technologies.

Oiendrilla Das

Oiendrilla Das

Oiendrilla Das is Customer Advocacy Lead for Life Sciences and Genomics Marketing for AWS. She comes from a background in life sciences marketing, with a specialty focus on life sciences and cloud computing. Oiendrilla holds an MBA degree in marketing and completed her engineering in Biotechnology prior to her MBA degree.

James Wiggins

James Wiggins

James Wiggins is a senior healthcare solutions architect at AWS. He is passionate about using technology to help organizations positively impact world health. He also loves spending time with his wife and three children.