AI is not yet driving drug development
Rhetoric is no substitute for results.
TL;DR
- A peer-reviewed paper in Nature Reviews Drug Discovery states AI's clinical impact in drug discovery is "disappointingly limited."
- AI is effective at identifying drug candidates but struggles to prove their survival in human biology.
- AI-derived drugs face the same Phase II bottleneck as conventional drugs, needing to prove efficacy in large, diverse patient populations.
- Challenges include insufficient or messy cellular data, making it difficult to clean and categorize for AI/ML.
- Over $40 billion has been invested in AI biopharma this decade, but few medicines have progressed beyond Phase II.
- While some AI applications exist, like Moderna and Merck's Phase III trial for cancer treatment decision-making, they focus on specific individual choices rather than wholesale drug development.
- VCs emphasize the need for more biological data, regulatory reform, and advanced computing to achieve industry disruption.
- Current AI applications in development may be surface-level, focusing on biostatistics and data lock rather than complex trial enrollment and monitoring.