Introducing Decision-Making Under Uncertainty to Medical Research:...
Talkat Optimization and Decision-Making Under Uncertainty (Simons Institute, September 2016) · 2016-09-19..09-23·
summary
Argues that patient-level Bayesian adaptive and multi-arm bandit designs raise real-world success rates (example: 63 expected successes vs 50 under 50/50 randomization for N=100 with a seven-patient adaptive learning window), replace fixed trials with continuous learning, and shorten confirmatory trials. Describes implementation: I-SPY 2: 11 experimental arms, 5 graduated to phase III, plus a global platform GBM‑Agile being built as an ongoing adaptive registration pathway. Simulations and longitudinal models drive decisions.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:12New York Academy of Sciences, March 2017: "The Need to Accelerate Therapeutic Development - Must Randomized Controlled Trials Give Way"
- 3:09What Are Randomized Clinical Trials?
- 10:24The Belmont Report (1979)
- 17:55Decisions depend on scenario
- 35:11Robustness of Bayes adaptive vs RCT vs play-the-winner (Robbins)
- 36:11Decision Analysis of Trials
- 51:16Consequence of inadequate simulations ...
related talks
21mLeveraging Multi-Armed Bandit Algorithms for Dynamic Decision Making
29mBalancing Personalization & A/B Testing in a Customized World
32mMulti-Armed Bandits in Recommendations
01h 51mMorning Session: Genomics, Medicine, Epidemiology and Data Interpretation
01h 30mBayBE: A Bayesian Back End for Experimental Planning in the Low-To-No-Data Regime
01h 03mExperimentation in Data Science
25mScaling Bayesian Experimentation
43mToward Human-Centered Machine Learning