Fig. 4: 12-month MDS-UPDRS Total meta-predictive performance and feature class contributions.
From: Genetically-informed prediction of short-term Parkinson’s disease progression
Performance of 12-month MD-UPDRS Total progression meta-prediction as measured by area under (AUC) receiver operating characteristic (ROC; left) and Precision-Recall (PR; right) curves. PPMI train-test split (top) and independent testing on PDBP (bottom) are presented. Curves are presented for full models (red), without genetic features (blue), without physician examination (MDS-UPDRS III) (gray), without survey-based features (MoCA, MDS-UPDRS I and II) (green), and without imaging features (magenta, PPMI only). See Table 4 for ROC AUC and PR AUC values.
