Skip to main content
Fig. 5 | BMC Medical Research Methodology

Fig. 5

From: Comparison of State-of-the-Art Neural Network Survival Models with the Pooled Cohort Equations for Cardiovascular Disease Risk Prediction

Fig. 5

Kaplan–Meier Observed Event Rate and Recalibrated Predicted Event Rate for the ASCVD Outcome in the MESA Cohort. For each model, we divided participants into 10 groups (decile) based on their sorted recalibrated event rate estimates. Then, for each decile, mean observed event rate (Kaplan–Meier method) was plotted against mean recalibrated event rate estimates. In a perfectly calibrated model, the recalibrated event rate estimates would be the same as the observed event rate in each decile. This means that all points would be clustered around the dotted identity line

Back to article page