Table 5.
Confusion matrix and per-class performance (%) achieved by the proposed method across all ten-folds using the CNN model and based on the MIT-BIH arrhythmia database.
| Predicted |
Per-class Performance (%) |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | S | V | F | Q | acc | ppv | sen | spec | ||
| Actual | N | 89774 | 203 | 357 | 37 | 91 | 97.35 | 97.60 | 99.24 | 88.30 |
| S | 757 | 1945 | 56 | 1 | 18 | 98.99 | 87.89 | 70.04 | 99.75 | |
| V | 632 | 51 | 6449 | 44 | 47 | 98.77 | 91.88 | 89.28 | 99.94 | |
| F | 175 | 3 | 95 | 527 | 2 | 99.67 | 86.39 | 65.71 | 99.92 | |
| Q | 639 | 11 | 62 | 1 | 7314 | 99.20 | 97.89 | 91.12 | 99.84 | |
acc: accuracy; ppv: positive predictive value; sen: sensitivity; spec: specificity