Learning Bayesian Statistics #148 Adaptive Trials, Bayesian Thinking, and Learning from Data, with Scott Berry
25 snips
Dec 30, 2025 Scott Berry, a statistician and co-founder of Berry Consultants, dives into the world of Bayesian adaptive trials. He highlights how these methods revolutionize clinical design, especially during urgent situations like the COVID-19 pandemic. Scott shares insights into the practical hurdles of implementing these designs and the importance of simulation tools for decision-making. He passionately discusses effective communication strategies for stakeholders and predicts a future with more enriched data, learning healthcare systems, and the skills needed for emerging statisticians.
AI Snips
Chapters
Transcript
Episode notes
Bayesian Roots Of Berry Consultants
- Scott Berry founded Berry Consultants to design smarter, more efficient clinical trials using Bayesian ideas and computation.
- He saw adaptive designs as a way to learn during trials instead of waiting years to see results.
Why Bayesian Fits Complex Trials
- Frequentist methods work well for simple, fully prospective experiments but struggle with complex adaptive designs and many external data sources.
- Bayes conditions on observed data and handles moving parts and external information more naturally.
Platform Trials In COVID Response
- Platform trials pooled multiple experimental arms against shared controls during COVID and produced rapid, high-impact answers.
- REMAP‑CAP and RECOVERY showed adaptive platforms could stop ineffective arms quickly and save lives.