Statistical consulting specialties
Bayesian adaptive trial design
When should a small trial continue enrolling, stop for futility, or declare success?
I work with small biotech sponsors and their clinical and biostatistics teams to compare candidate trial designs before the protocol is finalized.
Discuss a trial designWhat the work covers
- Define the decision rules
- Specify the endpoint, treatment effect of interest, interim looks, and decision thresholds. Make prior assumptions explicit, including any borrowing from historical controls.
- Simulate the trial
- Under null and plausible treatment effects, estimate false-positive probability, power, expected enrollment, and the chance of stopping early for success or futility.
- Stress-test the design
- Examine what happens when historical data conflict with new results, outcomes arrive late, or recruitment is slower than planned. Compare candidate rules with a fixed design.