Techniques of identifying patients for enrollment in clinical trials based on prospective use of patient attributes that are intended to increase the likelihood of detecting a treatment effect (if one truly exists) compared to an unselected population.