Suppose you have a registry or cohort of volunteers on whom you are gathering longitudinal data and can hence characterize their recent past history, and suppose you do so with ratings that are used in prevention trials. Then you have a cohort that you can rapidly recruit from because you know them and they know you. They are following you on Twitter and Facebook. That’s what I mean by recruitment in a nanosecond. Then, because you know their pre-randomization “trajectories” or characteristics, you could better estimate how long a trial might be (taking into consideration how you expect the drug to work), then randomize them into an appropriate strata, and you can customize each individual’s outcome.

View all comments by Lon Schneider