Table 3. .
Classifier types, inputs, and metrics for evaluation during classification
| Classifier | SVM | RF | KNN | LOG_R | LDA | Deep learning | Multiple | Other |
|---|---|---|---|---|---|---|---|---|
| Frequency* | 171 (68.4%) | 20 (8.0%) | 17 (6.8%) | 22 (8.8%) | 22 (8.8%) | 20 (8.0%) | 46 (18.0%) | 52 (20.8%) |
| Inputs into classifier | Brain network metrics | Injury/disease actor | Demographic | Behavior/cognitive data | Medical Hx | Meds | Genes/blood biomarkers | Other |
| Frequency | 100% | 13.5% | 10.1% | 5.9% | 2.5% | 1.7% | 0% | 1.6% |
| Metric for evaluation | Accuracy | Sensitivity | Specificity | AUC (AUROC) | Predictive power | Regression outputs | Other (e.g., F1) | |
| Frequency | 87% | 70.4% | 69% | 40% | 12% | 12% | 20% | |
Note: SVM, support vector machine; RF, random forest; KNN, k nearest-neighbor; LOG_R, logistic regression; LDA, linear discriminant analysis. *Total >100%, including studies with more than one classification approach.