Figure 1
From: Bridging a translational gap: using machine learning to improve the prediction of PTSD
TIE* Algorithm Flow Chart. The figure outlines the successive steps used by the TIE* algorithm to, first, identify (step I) and validate (step II) compact set of maximally predictive risk indicators (MBs), calculate ROC curve AUC for the MB (step III), include the MB in a pool of MBs if AUC ≥ that for the original MB (Step IV), extract MB features from the dataset (step V) and reiterate steps I to V until all MBs in a dataset are identified.