Figure 1.

Method visualization and description. (A) Workflow for implementing the proposed method in a new clinical trial. B (continuous) and C (survival outcome): Schematic diagram for individualized predictions that are generated for each person in the current trial, where the solid red lines indicate observed outcome for the participants of the current trial and the dashed blue lines indicate predicted outcome for those participants had they not been treated. (B) illustrates the method for continuous outcomes, where the left side represents the outcomes of those randomized to the control arm and the right side represents the outcomes of treated participants. The predicted outcome values (dashed blue lines) for the control units had they not been treated would be exactly what they are observed to be (solid red lines), while the predicted outcome values (dashed blue lines) for the treated units had they not been treated are different from the observed outcome (solid red lines). (C) illustrates the analogous mechanism for survival outcomes, where the predicted survival times for the control units (dashed blue lines) are the same as the observed survival times (solid red lines), whereas the predicted survival for the treated individuals are lower than the observed survival times. Our method capitalizes on these differences to augment statistical power.