well, there are also 2 different ways to look at thise. there's seemless adaptive, which we are talking about where all the pts (including those in the phase 2 portion ) are included in final analysis. in this case, you need to know endpoints and inclusion very well because you cant really change that.
But you can also do operationally adaptive, whereby, you adapt to phase 3 within a single trial, but only those pts in the phase 3 portion are included toward analysis. The benefits here are not as much as seemless, but you do save time. this allows you to do adaptation as an amendment, using the same centers (dont ahve to restart a new trial), and you can change criteria and endpoint based on . Not enough companies are utilizing this approach, where you can learn and adapt.