2024 – The Year of Production | Highlights from the AWS Life Sciences Leaders Symposium 2024

Last week, we hosted our 6th Annual AWS Life Sciences Leaders Symposium in Boston. Over 450 executives from 240 top life sciences organizations joined us in person for an intense full-day discussion on how life sciences organizations can leverage generative AI, machine learning, and data analytics on AWS to revolutionize drug discovery and clinical development.

Centered on this year’s theme “Data + Generative AI : Unlocking Real-World Value in Life Sciences,” leaders from 13 pioneering life sciences organizations and AWS partners took the stage to showcase tangible innovations to advance cutting-edge science, and discussed key trends that will shape the deployment of generative AI in life sciences in the coming year. Here’s a quick recap of the key highlights.

2024: The year of Production

In the opening keynote, co-presented with NVIDIA and Genentech, I highlighted key trends shaping the deployment of generative AI in life sciences R&D and clinical development. We’ve doubled down with our customers, partners, and investments across our technology stack to transform this vision into a reality. And this is proven by the fact that 9 out of the top 10 pharma companies globally (by revenue) use AWS for generative AI and machine learning.

We explored the top 3 considerations for any organization planning their move to GenAI from PoC to Production:

• Extensibility and choice to keep pace with the proliferation of models with domain-specific capabilities
• Ability to rapidly deploy new workflows and pipelines for both well defined use cases like protein folding, where AWS HealthOmics shines, as well as non-deterministic use cases like target identification, which AI agents can address
• the crucial roles of proprietary data and management

Attendees were not only introduced to AWS’ comprehensive generative AI solutions tailored for life sciences, but also to strategies to help organizations remove data bottlenecks in applications, make it as accessible as possible while remaining protected. The keynote also provided actionable guidance on specialized models for life sciences. More on the key takeaways from the keynote on a follow-up blog coming shortly.

Kimberly Powell, GM of Healthcare at NVIDIA, highlighted how AWS and NVIDIA are collaborating to streamline the training and deployment of specialized models, and scale resources needed for interpreting bio-molecular data. By integrating NVIDIA’s BioNeMo framework with Amazon SageMaker, researchers can use powerful biological models for for accelerated drug discovery & development, benefiting from AWS’s scalability, security, and operational excellence. She also explained how NVIDIA NIM’s, a broad library of AI models for drug discovery, medical imaging and genomics, can now be accessed via AWS Marketplace and orchestrated using AWS HealthOmics, our purpose-built service for biological data analysis.

John Marioni, Senior Vice President and Head of Computational Sciences at Genentech, concluded the keynote by outlining a future where generative AI fundamentally reshapes drug discovery. He highlighted how digitally connected labs can enhance treatment development and disease targeting by continuously experimenting, analyzing, and iteratively learning from massive volumes of data. John also described how autonomous agents could use reasoning abilities to orchestrate complex multistep tasks through logical sequencing – significantly boosting researcher productivity, decision-making, and heralding a new era of innovation through improved experiment design and enhanced data search capabilities.

Here is the recording of the keynote, in case you wanted to revisit.

Choice is Key

Embracing generative AI is essential for organizations to stay competitive, but adapting to its rapid evolution of new models and workflows can be daunting. In the complex life sciences industry, the operative word is ‘choice’—there’s no one-size-fits-all solution, instead, its selecting the right strategy from a range of available options. Our session on ‘Building R&D Workflows with Generative AI – Keeping Pace with Emerging Models‘ de-mystified this, and provided guidance around which services (from AWS’ library of 40+ services for generative AI and ML) organizations need to invest in, tailored to specific use cases. Additionally, the session also explored the best methods for customizing workloads on your data, focusing on domain adaptation, task adaptation, workflow integration, and enhancing user experience. Attendees gained specific guidance on when to train or fine-tune models for domain-specific tasks, versus when to focus on prompt-engineering with off-the-shelf models as they plan for future projects.

Anthropic demonstrated the capabilities of its newly launched state-of-the-art model, Claude 3, available on Amazon Bedrock, to solve complex life sciences use cases. The session offered a detailed overview of how HCLS organizations can utilize Claude in foundational roles, including as an AI collaborator, a data transformer, and a knowledge synthesizer. Anthropic was founded to create the world’s safest and most capable LLM, and we are working together to accelerate the responsible deployment of generative AI. And with Claude 3 on Amazon Bedrock, customers can seamlessly integrate their model and data within the same cloud environment, innovating with generative AI right where your data lives. You can watch the session recording here.

Data + Generative AI: Real-world implementations

Startups are pioneering groundbreaking drug discovery innovations using generative AI – which is why it was essential for our attendees to learn firsthand about the swift, agile innovation from these thought leaders.

Terray Therapeutics shared how their AWS-powered connected labs platform is transforming small molecule discovery by developing technologies and stitching together the advancements across data, hardware, and generative AI. Their platform merges experimentation with computation, enabling them to iterate 10,000 times faster and 125,000 times cheaper than traditional methods, streamlining the discovery of novel leads for complex targets and speeding up pipelines. Vertex Pharmaceuticals demonstrated their ML-powered innovations around molecular activity studies and image segmentation workflows to dramatically increase scientist productivity. Built on AWS, these workflows enable scientists to predict molecular properties and automate image analysis at petabyte scale, while optimizing costs and reducing operational burdens – helping Vertex discover transformative medicines for people with serious diseases. Click here to watch the session recording.

Partner success with AWS: Fueling generative AI innovation in life sciences

AWS stands out not only for our extensive generative AI, ML, and data solutions but also for our trusted partner network. Two of our key life sciences partners, ZS and PwC showcased their industry-leading generative AI solutions for R&D and clinical development, built using AWS. ZS’s scientific knowledge mining solution combines knowledge graphs with LLMs, their FM-based digital twins for better clinical trial predictions. And PwC showcased an innovative, gen AI-powered solutions for clinical trial protocol optimization, built on Amazon Bedrock. More importantly, they clarified the essential people, process, and technology requirements that are fundamental for any organization navigating this transition – especially true with CEOs try to answer “What is our generative AI strategy?”, while CIOs grappling with, “How do we implement this?” For full session recording, visit this page.

From PoCs to Production

Experimentation is crucial when building with generative AI, and companies refine their approaches with each test. Yet, many remain stalled in the proof-of-concept stage, unable to progress to production despite promising results. The panel discussion on ‘Moving Generative AI From POCs to Enterprise Use’, featuring leaders from Takeda Pharmaceuticals, Insilico Medicine, and EPAM, explored essential considerations for life sciences organizations aiming to transition their POCs to robust enterprise applications. Key takeaways included prioritizing use cases with clear business value, shifting mindsets to embrace new paradigms, and leveraging replication to enable agile experimentation without overwhelming resources. Watch the discussion here.

Data: Your generative AI differentiator

Throughout the day, one unifying thread connecting all the sessions emerged – data. At AWS, we firmly believe data is your differentiator. It is the core to the success of every experiment, every analysis, and every patient outcome. This made Gilead’s session on ‘Maximizing AI Through Robust Data Foundations’ the perfect finale to the day’s discussions. The session illustrated how Gilead’s comprehensive data strategy and cloud infrastructure, built on AWS, empowers innovation with generative AI across the value chain, from target assessment to commercialization. It also delved into how their data platform, built on a data mesh architecture, centralizes data, scales management, and fosters organizational collaboration, thereby accelerating innovation, enhancing governance, and increasing agility. Detailed session recording here.

Its evident that the generative AI revolution is here, presenting both immense opportunities and complex challenges for life sciences leaders. Navigating this rapidly changing environment requires strategic, bold leadership, a readiness to question assumptions, and a dedication to continuous learning and adaptation. With 2024 dubbed as the “Year of Production” for generative AI in life sciences, and the imperative to act is NOW. Our team at AWS is prepared to help you swiftly deploy your high-impact use cases while maintaining a commitment to ethical and responsible AI practices.

The path to production is open. Discover more by visiting our website.

All session recordings and presentations can now be accessed here.