Table 3.
Normal and abnormal signals together of three diseases and controls.
| Classification method | TPR of LQT1 | TPR of HCM | TPR of CPVT | TPR of WT | Accuracy |
|---|---|---|---|---|---|
| kNN, cityblock metric, equal weighting, k = 5 | 93.3 | 76.1 | 70.4 | 68.4 | 74.6 |
| kNN, cityblock metric, inverse weighting, k = 5 | 93.3 | 74.6 | 71.7 | 68.4 | 75.0 |
| kNN, cityblock metric, squared inverse weighting, k = 5 | 91.1 | 76.1 | 71.7 | 69.2 | 75.0 |
| kNN, Mahalanobis metric, equal weighting, k = 1 | 87.8 | 80.3 | 71.7 | 69.2 | 75.0 |
| kNN, Mahalanobis metric, inverse weighting, k = 1 | 87.8 | 80.3 | 71.7 | 69.2 | 75.0 |
| kNN, Mahalanobis metric, squared inverse weighting, k = 11 | 94.4 | 78.9 | 71.2 | 66.9 | 75.1 |
| Random forests, 54 trees | 88.9 | 81.7 | 76.8 | 72.9 | 78.6 |
| LS-SVM RBF kernel, parameters C = 24, sigma = 2 | 85.6 | 71.8 | 70.8 | 78.2 | 75.3 |
True positive rates (TPR, %) of LQT1, HCM, CPVT diseases and controls (WT) with 90, 71, 233 and 133 signals respectively and accuracy (%) of all signals (kNN is k nearest-neighbor searching method and LS-SVM least square support vector machine). The best accuracy is bolded.