Learning Bayesian Statistics BITESIZE | The Trial Design That Learns in Real Time
8 snips
Jan 7, 2026 Scott Berry, a biostatistician and co-founder of Berry Consultants, discusses the revolutionary shift from frequentist to Bayesian approaches in clinical trials. He highlights the limitations of traditional trial designs and emphasizes the efficiency of adaptive and platform trials, especially in the rapid response to COVID-19. Berry shares insights on designing impactful trials that save lives, using real-time data to adapt strategies. This engaging conversation reveals how innovative methodologies are reshaping the future of medical research.
AI Snips
Chapters
Transcript
Episode notes
Frequentist Trials Limit Learning
- Clinical trials historically favored frequentist methods and fixed designs that limit learning during the trial.
- Scott Berry argues adaptive, Bayesian-driven designs allow trials to change as data arrives and answer more questions efficiently.
Design Trials To Answer Multiple Questions
- Work with stakeholders to design trials that answer multiple questions in one experiment.
- Use simulations and Bayesian models to drive dose, sample size, and adaptive decisions during the trial.
Bayes Handles Complexity Better
- Frequentist inference breaks down as trial designs gain complexity and include many moving parts.
- Bayesian methods scale naturally to complex, adaptive designs and external data by conditioning only on observed data.