Table 3. Classifier performance.
Model | Sensitivity [95% CI] | Specificity [95% CI] | Best CVA [95% CI] | AUC | Features Used | Noise/Not Noise | ||
Noise | M1 | All Noise | 0.910.88 0.93 | 0.820.78 0.86 | 0.870.84 0.89 | 0.93 | 147 | 475/880 |
M2 | Eyeballs | 0.460.25 0.61 | 1.01.0 1.0 | 0.980.97 0.99 | 0.93 | 40 | 30/880 | |
M3 | HeadMotion | 0.390.21 0.57 | 0.990.99 1.0 | 0.970.97 0.98 | 0.99 | 16 | 28/880 | |
M4 | Ventricles | 0.430.29 0.62 | 0.990.99 1.0 | 0.970.96 0.98 | 0.93 | 5 | 37/880 |
Performance metrics (sensitivity, specificity, best cross validation accuracy (CVA), area under the curve (AUC)), number of features selected, and proportion of noise components in data for four successful models, including comprehensive noise (All Noise, M1) and three noise subtypes (M2)(M3)(M4), built with and tested with ten -fold cross validation on Data A (healthy control).