Table 5 Performance metrics (mean ± standard deviation) for CTL vs PD-I and PD-I vs PD-II classifiers.
Comparison | Feature Set | Accuracy | Precision | Recall | F1 | Specificity | ROC AUC |
|---|---|---|---|---|---|---|---|
CTL vs PD-I | Metabolites Only | \(0.875 \pm 0.125\) | \(0.933 \pm 0.133\) | \(0.850 \pm 0.229\) | \(0.860 \pm 0.147\) | \(0.900 \pm 0.200\) | \(1.000 \pm 0.000\) |
Metabolites + Clinical | \(0.900 \pm 0.122\) | \(1.000 \pm 0.000\) | \(0.800 \pm 0.245\) | \(0.867 \pm 0.163\) | \(1.000 \pm 0.000\) | \(0.975 \pm 0.075\) | |
PD-I vs PD-II | Metabolites Only | \(0.800 \pm 0.187\) | \(0.750 \pm 0.310\) | \(0.800 \pm 0.332\) | \(0.757 \pm 0.298\) | \(0.800 \pm 0.245\) | \(0.775 \pm 0.284\) |
Metabolites + Clinical | \(0.900 \pm 0.122\) | \(0.933 \pm 0.133\) | \(0.900 \pm 0.200\) | \(0.893 \pm 0.137\) | \(0.900 \pm 0.200\) | \(1.000 \pm 0.000\) |
