Table 12 Comparison of the proposed model with existing models.
Study | Domain | No. of features | Methodologies | CV | *Classification accuracy (%) | MAE | RMSE |
|---|---|---|---|---|---|---|---|
Time | 19 | K-NN, DT, and RFT | None | 80–91 | NA | NA | |
Spatiotemporal and Statistical Features | 17 | DT, SVM, ensemble classifier (EC) and Bayes classifier (BC) | 10-fold | 99.4 | NA | NA | |
Time, Frequency | 34 | ANN | One-fold & leave one out | PD detection (97.4) & (87.69) Severity assessment | NA | NA | |
Time and Frequency | 10 | K-NN, SVM, DT, and RFT | 10-fold | 98.50 for classification and 96.4 for severity assessment | NA | NA | |
Time and Frequency | NA | CNN(Classification) ResNet (Severity Assessment) | 10-fold | 97.42 for Classification 96.52%(Severity Assessment) | NA | NA | |
Statistical Features | 10 | SVM | 10-fold | 85 for detection | NA | NA | |
Time | NA | 1-D CNN | None | 92.7(multi-class classification) | NA | NA | |
Time | NA | K-means and LR | LOOCV | 98% applied only to classification | NA | NA | |
45 (works with real-time dataset) | Time and Frequency | NA | RF Regressor | CV-5 | NA | NA | 10.02 |
Proposed Model | Spatial, Time, and Frequency | 14 | RFT + ER | 10-fold | PD detection (with an accuracy of 97.5 ± 2.1) and 96.4 ± 2.3 for severity assessment | 0.065 ± 0.024 | 0.080 ± 0.06 |
