Biotech Values

[iwfal]

adaptive, and not necessarily bayesian, can also be helpful for dropping a dose after the stage1 portion of an adaptive design. Or in some of the newer onc and AD multi-company designs, it can be used to drop a drug. Sample size re-estimation through adaptive design is very useful, especially if it turns out you have a pretty good effect in stage 1 of the phase 2 portion. phase 1/2 adaptive rather straightforward.

Agree completely. As long as you are highly confident what decisions you will want to make and they are simple (e.g. Dose adjustment or size adjustment) they work well. But they get oversold dramatically.

And of course good ITT effect size always makes things easier - e.g. No need for difficult decisions (difficult to preplan) about protocol changes you can only make using part 1 data. (With strong effect size tuning becomes simpler or non-existent.)

phase 2/3 takes more simulations and modelling and discussions with agency, but if you have a great statistician and have run a useful phase 1 in pts (not HV), you can save lots of time. and the risk is small because you can always finsih the trial as a phase2 if you didnt met your predetermined threshhold.

Agree that with enough data you have a much better chance of predicting a few simple protocol changes you might want to make. Suggest that that is fairly rare in practice and when it does happen it means you took more time in ph 1. All a trade - and would fully agree that under the right conditions it can be a useful tool. But not the panacea that oft gets touted. And oft not useful at all (ii.e. If you aren't highly confident you know what decisions you want to make and it is a simple set - e.g. Because you have too little data to have any confidence of potential protocol changes)