Table 3.
Performance of models for glycated hemoglobin elevation prediction.
| Model | Variables used | Number of RCSa knots | AUR ROCb | 95% CI | Recall | Precision | F1 |
| PMc1 | Completed | 3 | 73.67 | 74.71-77.51 | 85.24 | 77.58 | 81.23 |
| PM2 | Complete | N/Ae | 74.04 | 74.35-77.16 | 82.18 | 78.76 | 80.43 |
| PM3 | Reducedf | 5 | 74.73 | 75.38-78.15 | 84.40 | 78.80 | 81.50 |
aRCS: restricted cubic splines.
bAUR ROC: area under the receiver operating characteristic.
cPM: predictive model.
dAll variables (age, random blood sugar, cholesterol, non–high-density lipoproteins, estimated glomerular filtration rate, and BMI).
eN/A: not applicable.
fReduced variables (age, random blood sugar, cholesterol, non–high-density lipoproteins, and estimated glomerular filtration rate).