Table 12.
Comparison with related works in relation to dataset, methodology, and estimation error.
| Author | Method Used | Number of Subjects | Performance Criteria | Systolic Blood Pressure | Diastolic Blood Pressure |
|---|---|---|---|---|---|
| Kachuee et al. [24] | SVM | MIMIC II (1000 subjects) | MAE MSE RMSE R |
12.38 - - - |
6.34 - - - |
| Kim et al. [23] | Multiple nonlinear regression (MLP) | 180 recordings, 45 subjects | MAE MSE RMSE R |
5.67 - - - |
- - - - |
| Kim et al. [23] | Artificial neural network (ANN) | 180 recordings, 45 subjects | MAE MSE RMSE R |
4.53 - - - |
- - - - |
| Cattivelli et al. [25] | Proprietary algorithm | MIMIC database (34 recordings, 25 subjects) | MAE MSE RMSE R |
- 70.05 - - |
- 35.08 - - |
| Zhang et al. [27] | Support vector machine (SVM) | 7000 samples from 32 patients | MAE MSE RMSE R |
11.64 - - - |
7.62 - - - |
| Zhang et al. [27] | Neural network (nine input neurons) | 7000 samples from 32 patients | MAE MSE RMSE R |
11.89 - - - |
8.83 - - - |
| Zadi et al. [59] | Autoregressive moving average (ARMA) models | 15 subjects | MAE MSE RMSE R |
- - 6.49 - |
- - 4.33 - |
| Slapničar et al. [30] | Deep learning (spectro-temporal ResNet) |
MIMIC III database (510 subjects) | MAE MSE RMSE R |
9.43 - - - |
6.88 - - - |
| Su et al. [28] * | Deep learning (long short-term memory (LSTM)) |
84 subjects | MAE MSE RMSE R |
- - 3.73 - |
- - 2.43 - |
| This work | Gaussian process regression (GPR) | 222 recordings, 126 subjects | MAE MSE RMSE R |
3.02
45.49 6.74 0.95 |
1.74
12.89 3.59 0.96 |
* Deep learning algorithm on a small database.