Table 4. Multivariable logistic regression models.
| Feature 1 | Feature 2 | AUC (95% CI) | p | LR | AUCLOOCV |
|---|---|---|---|---|---|
| SBP α1 | DBP μ | 0.843 (0.68, 1.01) | 0.009 | 2.86 | 0.721 (0.50, 0.94)* |
| PI α1 | DBP μ | 0.843 (0.69, 1.00) | 0.014 | 2.86 | 0.643 (0.40, 0.89) |
| DBP α1 | DBP μ | 0.864 (0.72, 1.00) | 0.022 | 2.86 | 0.750 (0.56, 0.94) |
| PI α2 | MAP μ | 0.864 (0.72, 1.01) | 0.068 | 3.43 | 0.679 (0.46, 0.89) |
| MAP α1 | DBP μ | 0.871 (0.74, 1.01) | 0.027 | 3.43 | 0.743 (0.55, 0.94) |
| PI α2 | DBP μ | 0.921 (0.82, 1.02) | 0.035 | 4.00 | 0.821 (0.66, 0.99) |
Features included mean (μ), short- and long-term scaling exponents (α1 and α2, respectively) for mean arterial (MAP), systolic (SBP) and diastolic (DBP) blood pressure, as well as pulse interval (PI) time series. LR denotes the positive likelihood ratios, the 95% confidence intervals (CI) are reported for the AUC. p values are derived from the Delong comparison14 with the non-detrended mean MAP model. The corresponding AUCs from leave-one-out cross-validation (AUCLOOCV) are also reported, where *denotes a statistically-significant (p < 0.05) difference from the Delong comparison of the LOOCV mean DBP model14.