Table 8 Comparison of predictive performance of the three models
From: Development and internal validation of risk prediction model of metabolic syndrome in oil workers
Evaluation index | Logistic regression model | Random forest model | CNN |
|---|---|---|---|
Accuracy rate(%) | 82.49 | 95.98 | 92.03 |
Sensitivity(%) | 87.94 | 95.52 | 90.59 |
Specificity(%) | 74.54 | 96.65 | 94.14 |
F1 Score | 0.86 | 0.97 | 0.93 |
AUC | 0.88 | 0.96 | 0.92 |
Brier score | 0.15 | 0.08 | 0.12 |
observed-expected ratio | 0.83 | 0.97 | 1.13 |
calibration-in-the-large | 0.109 | 0.099 | 0.098 |
Integrated Calibration Index | 0.075 | 0.073 | 0.074 |