Fig 1.
Flow chart demonstrating the exclusion criteria, and the development process of the CAD model, and subsequent validation.
The process was the same for the other CMDs.
Table 1.
Baseline characteristics of the study samples for each dataset used to derive CMD-specific prediction models.
Table 2.
Discriminative ability of CMD risk prediction models among different subpopulations.
Fig 2.
Calibration plots for CMD risk prediction models.
Five-year absolute risks for CAD (A), DM2 (B), stroke (C), DVT (D), and AAA (E) were split into deciles, and mean risk probability for each decile was plotted versus the portion of positive CMD cases in the decile for a time horizon of five years.
Fig 3.
CMD risk prediction models demonstrate the relation between risk level and survival.
Risk probabilities for five diseases, CAD (A), DM2 (B), stroke (C), DVT (D), and AAA (E), were split into deciles, and mean risk probability for each decile was plotted versus mean survival time in the decile.