Extended Data Table 1 Comparison of associations between the traditional and MTAG-based coronary artery disease (CAD) polygenic risk scores and CAD in the UKBB. The UKBB contained 343,676 samples (17,986 CAD cases). The middle quintile of the PRS (40th to 60th percentiles) was used as the reference group. The AUCs and Nagelkerke pseudo-R² were calculated using PRSs as continuous variables in the models including the PRS and all covariates or only the PRS. 75% of the full sample was used for training and 25% for validation

From: Genetic predictors of lifelong medication-use patterns in cardiometabolic diseases