Table 9.
A comparison of recent works developed for CVD detection with machine learning and portable devices.
| Author (Year) | Features | Approach | Modality | Accuracy (%) | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|---|---|
| This work | Temporal HRV | Convolutional Neural Networks | ECG; PPG | ECG = 95.50 | ECG = 94.50 | ECG = 96.00 |
| PPG = 95.10 | PPG = 94.60 | PPG = 95.20 | ||||
| Zhou et al. [17] (2015) | R-R intervals | Shannon Entropy | ECG | 97.89 | 97.37 | 98.44 |
| Cui et al. [18] (2017) | R-R intervals | Ensemble Model | ECG | 97.78 | 97.04 | 96.97 |
| Shashikumar et al. [74] (2018) | R-R Intervals and waveform features | Bidirectional Recurrent Neural Networks | ECG; PPG | ECG = 94.00 PPG = 95.00 |
- | ECG = 95.00 PPG = 100.00 |
| Bashar et al. [75] (2018) | R-R intervals and waveform features | Support Vector Machines | PPG | 91.16 | - | - |
| Tarniceriu et al. [76] (2018) | R-R Intervals | Markov Model | PPG | - | 98.45 | 99.13 |
| Aliamiri et al. [77] (2018) | Waveform features | Convolutional Recurrent Neural Networks | PPG | 98.19 | - | - |
| Tison et al. [78] (2018) | R-R Intervals | Neural Network | PPG | - | 98.00 | 90.20 |
| Fallet et al. [79] (2019) | R-R intervals and waveform features | Decision Trees | PPG | 95.00 | 92.90 | 96.20 |
| Kwon et al. [80] (2019) | R-R intervals | Convolutional Neural Network | PPG | 97.58 | 99.32 | 95.85 |