AWS for Healthcare & Life Sciences

Transforming Pharma R&D with Generative AI

  • The-Challenge-icon

    The Challenge

  • Explore-The-icon

    Explore the Demo

  • Start-Building-Today-icon

    Start Building Today

I Want to Design a New Drug Candidate for

  • Breast-Cancer-icon

    Breast Cancer

  • Diabetes-icon

    Diabetes

Progress Tracker

Continue to Next Step?

Progress Tracker

Continue to Next Step?

Drug discovery is challenging

Scientists need to know which drug candidates have the best chance of success as early as possible.

2xIcon

2x

more distinct drugs in
development, but flat approvals

13%Icon

<13%

assets in Phase 1
go to launch

novelIcon

Novel asset costs a$2.8B and
takes ~12 years of development,
with a 90% failure rate

Why is it so hard?

CHALLENGES
Complexity of biology

Complexity of biology

Diverse, unstructured data

Diverse, unstructured data

Differences between in vitro, animal, and human test results

Differences between in vitro,
animal, and human test results

Safety considerations

Safety considerations

Enhanced efficacy requirements

Enhanced efficacy
requirements

Regulatory hurdles

Regulatory hurdles

This is made further complex since drug discovery is
evolving.

icon

Traditional model

Mass production, one size-fits-all
therapies

icon

New paradigm

Personalized medicines & precision
therapeutics

icon
Leveraging data within and across health continuum
integral to accelerating drug discovery/development
icon

AI can transform pharma R&D, making it more efficient.

USE CASES
  • Scientific knowledge extraction
  • Compound screening
  • Protein, molecule,
    and drug-vector design
  • Indication selection and
    population phenotyping
Biology FMs
  • Scientific knowledge extraction
  • Compound screening
  • Protein, molecule,
    and drug-vector design
  • Indication selection and
    population phenotyping
Agentic Frameworks
  • Scientific knowledge extraction
  • Compound screening
  • Protein, molecule,
    and drug-vector design
  • Indication selection and
    population phenotyping

Designing new proteins with generative AI

NLP-POWERED, NO-CODE APPROACH TO ITERATING ON PROTEIN STRUCTURES

AmazonBedrockIcon

Amazon Bedrock

AgentsAmazonBedrockIcon

Agents for
Amazon Bedrock

EvolutionaryScale

ESM3 by EvolutionaryScale

The opportunity: Unlocking new
levels of R&D productivity

potential
Novel protein designs to expand R&D potential
MoreAccurate
More accurate 3D structure and function predictions
Simultaneous
Simultaneous multi-property optimization
(like binding, stability, manufacturability)
productivity

Start building with generative AI on AWS today

Qr
Learn more about
AWS for Healthcare
& Life Sciences
Qr
Get started with
Amazon Bedrock
Qr
Access ESM3
on AWS
Qr
Explore re:Invent
sessions on generative
AI for drug discovery