Drug Discovery

AI supports drug discovery by generating predictive chemistry and safety outputs, and by speeding up model testing and deployment for research teams.

Business impact

• number of AI/ML models running — AstraZeneca reported more than 100 AI/ML models running on its Predictive Insight Platform.
• time to iterate and deploy a new machine-learning model — Insilico Medicine reported a reduction from 50 days to 3 days.

Data requirements

• molecular and chemical data (tabular) — Needed for models that predict DMPK, ADME, safety, and molecular stability.
• drug-discovery research data (text) — Used in the discovery workflow described by the case studies.

AI methods and techniques

• predictive modeling — Used to generate DMPK, ADME, and safety predictions for candidate molecules.
• parallel model testing — Insilico Medicine used parallel testing instead of a sequential in-house pipeline to speed iteration.
• molecular property prediction — NYU researchers trained a model to learn chemical patterns associated with stability and predict hydrogen atom positions.

Industries

Real-world evidence

5 documented case studies on record.

Companies using this: AstraZeneca, Insilico Medicine, Nestlé, Crossr, New York University.