There's data from clinical trials. Before doing experiments with humans a company has to make the experiments in mice (or other animals). Any clinical trial of a drug that fails can be seen as a trial where a hypothesis failed to generalize. If I remember right the FDA does release some of the relevant data.