Data in Biotech Applying ML/AI to Drug Development with Anil Kane
Nov 26, 2025
Dr. Anil Kane, Executive Director at Thermo Fisher Scientific, dives into the transformative impact of AI and machine learning in drug development. He explains how predictive modeling is eliminating traditional trial-and-error methods, significantly cutting costs and timelines. Anil shares insights on accelerated stability testing with the ASAP program and the integration of digital tools to enhance manufacturing efficiency. He also discusses the balance between human expertise and AI, and the exciting future of digital transformation in pharmaceuticals.
AI Snips
Chapters
Transcript
Episode notes
Predictive Modeling Replaces Trial And Error
- Traditional drug development relied on empirical trial-and-error that caused costly rework and delays.
- Predictive modeling aims to eliminate rework by prioritizing promising techniques to save time and money.
Quadrant 2 Predicts Formulation Choices
- Thermo Fisher built OSD Predict's Quadrant 2 to rank formulation technologies for poorly soluble small molecules.
- The model uses simple molecular properties to predict methods like spray drying or hot-melt extrusion before using expensive drug substance.
Feed Dose Estimates And Fresh Data Into Models
- Always include the estimated human dose range and relevant PK context when running formulation predictions.
- Update the model with outcomes from new molecules so it becomes more robust over time.