Table 5.
Comparison between the Conv1D MF model and state-of-the-art inter-patient ECG classification methods. The best results per column (metric) are highlighted in red.
| ID | Model | Classes | Features | Classifier | ACC % |
N | SVEB | VEB | Macro Average Scores | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PREC | SEN | PREC | SEN | PREC | SEN | PREC | SEN | F1 | ||||||
| 1 | Proposed MF 1 | 3 | ECG Segments + RR | Conv1D MF | 96.48 | 99.38 | 96.96 | 58.06 | 85.30 | 90.20 | 96.34 | 82.55 | 92.87 | 86.81 |
| 2 | Proposed MF 6 | 3 | ECG Segments + RR | Conv1D MF | 98.18 | 99.00 | 99.10 | 82.68 | 81.60 | 95.63 | 95.00 | 92.44 | 91.90 | 92.17 |
| 3 | Zhang et al. [13] | 3 | ECG Segments + RR | Conv1D +ADNN | 94.70 | 98.00 | 96.20 | 90.80 | 78.80 | 94.30 | 92.50 | 94.37 | 89.17 | 91.62 |
| 4 | Wang et al. [14] | 4 | CWT + RR | Conv2D | 99.27 | 98.17 | 99.42 | 89.54 | 74.56 | 93.25 | 95.65 | 70.75 | 67.47 | 68.76 |
| 5 | Mondéjar-Guerra et al. [15] | 4 | Wavelets + HOS + LBP + RR | Ensemble SVM | 94.50 | 98.20 | 95.90 | 49.70 | 78.10 | 93.90 | 94.70 | 66.35 | 70.28 | 67.08 |
| 6 | Raj and Ray [16] | 5 | DOST | ABC + LSTSVM | 96.08 | 88.50 | 98.54 | 72.29 | 52.06 | 81.59 | 62.35 | 54.89 | 43.15 | 45.88 |
| 7 | Garcia et al. [17] | 3 | TVGG + PSO | SVM | 92.40 | 98.00 | 94.00 | 53.00 | 62.00 | 59.40 | 87.30 | 70.13 | 81.10 | 74.60 |
| 8 | Chen et al. [18] | 3 | DCT + Projection + RR | SVM | 93.10 | 95.40 | 98.40 | 38.40 | 29.50 | 85.10 | 70.80 | 72.97 | 66.23 | 69.18 |
| 9 | Zhang et al. [19] | 5 | Morph + RR | SVM | 88.34 | 98.98 | 88.94 | 35.98 | 79.06 | 92.75 | 85.48 | 65.65 | 81.53 | 65.46 |
| 10 | Lin and Yang [20] | 3 | Morph + WT + RR | LDC | 93.00 | 99.30 | 91.60 | 31.60 | 81.40 | 73.70 | 86.20 | 68.20 | 86.40 | 73.43 |