Machine learning for phase I and II clinical data to identify biomarkers and subgroups of patients likely to respond to therapy in phase III studies, identify early biomarkers of response using preclinical and clinical data, to reduce sample size, and increase confidence intervals, and innovative stopping and proceeding rules in clinical trials.