Table 3.
Between condition performance metrics
| IV ACC | IV sensitivity | IV specificity | IV MCC | IV AUROC | |
|---|---|---|---|---|---|
| Compounds with and without 24-hour time-of-day patterns included | |||||
| Logistic Regression: 5H and BL (66) | 71% | 47% | 100% | 0.54 | 0.82 ± 0.08 |
| Logistic Regression: 5H and 9H (66) | 61% | 47% | 71% | 0.19 | 0.60 ± 0.10 |
| Linear SVM: 5H and BL (47) | 71% | 47% | 100% | 0.54 | 0.91 ± 0.06 |
| Linear SVM: 5H and 9H (47) | 56% | 47% | 62% | 0.09 | 0.63 ± 0.10 |
| Compounds with 24-hour time-of-day patterns excluded | |||||
| Logistic Regression: 5H and BL (65) | 71% | 47% | 100% | 0.54 | 0.83 ± 0.08 |
| Logistic Regression: 5H and 9H (65) | 64% | 47% | 76% | 0.24 | 0.63 ± 0.10 |
| Linear SVM: 5H and BL (65) | 71% | 60% | 85% | 0.46 | 0.83 ± 0.08 |
| Linear SVM: 5H and 9H (65) | 69% | 60% | 76% | 0.37 | 0.64 ± 0.10 |
5H and BL indicates samples from the 9H condition were excluded from the calculation of performance metrics. 5H and 9H indicates samples from the BL condition were excluded from the calculation of performance metrics. Numbers in parenthesis indicate total number of compounds in each biomarker model. BL, baseline condition; 9H, control condition; 5H, insufficient sleep condition; ACC, accuracy.