Devany West

A pressing challenge in applying AI to drug discovery is limited access to high-quality, appropriately sized datasets, particularly for early-stage biotech firms. Without robust, diverse, and validated data, AI/ML models struggle to deliver reliable predictions or actionable wet lab outcomes. Open-source and federated learning platforms aim to bridge this gap by providing access to proprietary AI models trained on extensive preclinical, safety, and target product profile data. This workshop explores data-centric challenges and collaborative platforms that democratize AI for small molecule discovery.

Participants will explore:

• The opportunities of open data and federated learning platforms, as well as privacy-preserving data sharing in collaborative drug discovery environments
• Strategies for integrating proprietary and shared datasets to improve model generalizability and predictive power
• Approaches for validating AI-generated hypotheses in wet lab settings, including alignment of experimental design with model outputs
• Techniques for ensuring data quality, chemical diversity, and reproducibility in training datasets
• Examples of how access to large-scale datasets has accelerated small molecule lead optimization and candidate selection