---- thinking it seems nice to train a system to predict behavior and directly communicate with user or subject around checking what their wellness is or how something it might do could impact wellness (both checks could correct for user mistake)