AI in Drug Discovery
summary
Argues that AI is reducing the biggest bottlenecks in drug R&D: target selection, screening and design. Gives concrete results: predicted 1,000 gene targets from gene–disease data, screened 12 million compounds then bought 200 and advanced three candidates to preclinical work in one year (vs six), and uses GANs and reinforcement learning for de novo molecules. Describes CNNs for high‑throughput microscopy, cites quantum molecular machine learning work, and flags hurdles: data access, explainability and regulatory risk.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:25Natarajs background
- 3:36Healthcare AI
- 5:29Agenda
- 5:54Drug Discovery Life Cycle
- 6:37New Drug Development Costs
- 8:27What it takes to create a drug
- 10:04Compounds
- 11:14Drug Efficacy
- 12:44Virtual Screening
- 14:16High throughput microscopy
- 15:28Denovo drug design
- 17:38Cloud Pharmaceuticals
- 18:16Challenges
- 22:34AI Culture
- 25:24Where can we apply AI
- 27:54Conclusion
related talks
23mAI in the Life Science Sector
08mEngineering the Future of Pharma: AI-Driven Drug Discovery
35mCan AI solve the biggest problem in drug discovery? Time!
16mBeyond Patterns: The Case for Physics-Grounded AI in Drug Discovery
17mWill AI Help Solve the Biggest Challenges Facing Humanity?
30mMachine Learning: A New Approach to Drug Discovery
37mAI in medicine: where do we stand?
21mOpening