Table 3 AUCs for patients aged 40–79 years including those who were missing one or more PCE variables.

From: Machine learning and atherosclerotic cardiovascular disease risk prediction in a multi-ethnic population

N

Variables

GBM AUC (95% CI)

PCE AUC (95% CI)

157,231

PCE

0.784 (0.759–0.808)

0.774 (0.757–0.790)

157,231

PCE+

0.790 (0.765–0.814)

0.774 (0.757–0.790)

  1. The model with the highest AUC for each subgroup is given in bold.
  2. AUC area under receiver-operating characteristic curve, N number, PCE pooled cohort equation, PCE+ pooled cohort equation variables with additional electronic health record variables, GBM gradient boosting machine.
Back to article page