|
| Author | Year | Method | Class | Performance |
|
| Conventional machine learning approaches |
| Inan et al. [35] | 2006 | Feature extraction: classifier | WT and timing interval
Neural network | 3 | ACC: 95.16% |
| Sayadi et al. [36] | 2010 | Feature extraction: classifier | Innovation sequence of EKF
Bayesian filtering | 2 | ACC: 99.10% |
| SEN: 98.77% |
| SPEC: 97.47% |
| Martis et al. [32] | 2012 | Feature extraction: classifier | PCA
SVM with RBF kernel | 5 | ACC: 98.11% |
| SEN: 99.90% |
| SPEC: 99.10% |
| Prasad et al. [37] | 2013 | Feature extraction: classifier | HOS+ICA
KNN | 3 | ACC: 97.65% |
| SEN: 98.75% |
| SPEC: 99.53% |
| Martis et al. [38] | 2013 | Feature extraction: classifier | Cumulant+ICA
KNN | 3 | ACC: 99.5% |
| SEN: 100% |
| SPEC: 99.22% |
| Martis et al. [7] | 2013 | Feature extraction: classifier | HOS+PCA
LS-SVM | 3 | ACC: 93.48% |
| SEN: 99.27% |
| SPEC: 98.31% |
| Martis et al. [39] | 2013 | Feature extraction: classifier | Cumulant+PCA
LS-SVM | 5 | ACC: 94.52% |
| SEN: 98.61% |
| SPEC: 98.41% |
| Martis et al. [32] | 2012 | Feature extraction: classifier | DCT+PCA
SVM with RBF kernel | 5 | ACC: 99.52% |
| SEN: 98.69% |
| SPEC: 99.91% |
| Martis et al. [40] | 2014 | Feature extraction: classifier | ICA+DCT
KNN | 3 | ACC: 99.45% |
| SEN: 99.61% |
| SPEC: 100% |
| Kaya and Pehlivan [41] | 2015 | Feature extraction: classifier | Genetic algorithms
KNN | 5 | ACC: 99.69% |
| SEN: 99.46% |
| SPEC: 99.91% |
| Kaya and Pehlivan [8] | 2015 | Feature extraction: classifier | Time series+PCA
KNN | 5 | ACC: 99.63% |
| SEN: 99.29% |
| SPEC: 99.89% |
| Li and Zhou [33] | 2016 | Feature extraction: classifier | WPE+RR
RF | 5 | ACC: 94.61% |
| Mondjar-Guerra et al. [42] | 2018 | Feature extraction: classifier | Wavelets+LBP+HOS+several amplitude values
RF | 5 | ACC: 94.5% |
| SEN: 66.4% |
| SPEC: 70.3% |
| Yang and Wei [6] | 2020 | Feature extraction: classifier | Combined parameter and visual pattern features of ECG morphology
KNN | 5 | ACC: 97.70% |
| This work | 2020 | Feature extraction: classifier | WSN+the 4th time window
PNN | 4 | ACC: 99.3% |
| SEN: 99.5% |
| SPEC: 98.8% |
|
| Deep learning approaches |
| Martis et al. [40] | 2014 | 9-layer deep convolution neural network | 5 | ACC: 93.47% |
| SEN: 96.01% |
| SPEC: 91.64% |
|
