| | Authors | Dataset | Signals | Features | Feature selection | AI algorithm | Result (mmHg) | | SBP | DBP |
| | Haddad et al. [60] | MIMIC I
28 subjects | PPG | 27 | — | MLR | MAE: 6.10
STD: 8.01 | MAE: 4.65
STD: 6.22 | | El-Hajj et al. [23] | MIMIC II
942 subjects | PPG | 52 | Pearson’s coefficient, mutual information, recursive elimination | Deep learning recurrent model | MAE: 4.51
STD: 7.81 | MAE: 2.6
STD: 4.41 | | Li et al. [24] | MIMIC
50 subjects | PPG, ECG | 7 | — | Deep LSTM | MAE: 6.726
STD: 14.505 | MAE: 2.516
STD: 6.442 | | Farki et al. [59] | MIMIC II | PPG, ECG | 3 | — | means+ (RFR, gradient boosting regression, multilayer perception) | MAE: 2.56 | MAE: 2.23 | | Senturk et al. [70] | MIMIC II | PPG, ECG | 19 | — | RNN, NARX-NN, LSTM | ME: 0.0224
STD: 2.211 | ME: 0.0417
STD: 1.2193 | | Thambiraj et al. [55] | UCI_BP
3801 records | PPG, ECG | 43 | GA | RFR | MAE: 9.54 | MAE: 5.48 | | Tiloca et al. [20] | MIMIC II | PPG, ECG | 11 | — | RFR | RMSE: 13.01 | RMSE: 12.89 | | Manamperi et al. [61] | MIMIC II, self-collected
50 subjects | PPG | 53 | — | ANN | MAE: 4.8 | MAE: 2.5 | | Hasanzadeh et al. [25] | UCI_BP | PPG | 19 | — | LR, decision tree, RFR, Adaboosting | MAE: 8.22
STD: 10.38 | MAE: 4.17
STD: 4.22 | | El-Hajj et al. [69] | MIMIC II | PPG | 22 | Pearson’s correlation, random forest feature importance, RFE, sequential forward search | Feedforward neural networks, LSTM, GRU | MAE: 3.23
STD: 4.74 | MAE: 1.59
STD: 1.77 | | Khalid et al. [58] | Queensland, MIMIC
18010segments | PPG | 16 | VIF | KNN+RT | ME: 0.07
STD: 7.1 | ME: -0.08
STD: 6.0 | | Yang et al. [64] | Self-collected
14 subjects | PPG, ECG | 90 | | SVR, Lasso, ANN | MAE: 7.33
STD: 9.53 | MAE: 5.15
STD: 6.46 | | Attarpour et al. [62] | Self-collect
111 subjects | PPG | 34 | Moving backword algorithm, GA | Multilayer neural network | MAE: 5.59
STD: 0.30 | MAE: 4.45
STD: 0.16 | | Liu et al. [56] | Self-collected
35 subjects | PPG, ECG | 15 | — | DTR, SVR, Adaboosting, RFR | ME: 0.04
STD: 6.11 | ME: 0.11
STD: 3.62 | | Chakraborty et al. [63] | MIMIC II
50 subjects | PPG | 15 | NCA, RLF | Modified ANN | ME: 0.461
STD: 2.62 | ME: 0.15
STD: 4.482 | | Chen et al. [65] | MIMIC III | PPG, ECG | 14 | MIV | GA-SVR | MAE: 3.27
STD: 5.52 | MAE: 1.16
STD: 1.97 | | Chowdhury et al. [67] | Figshare_BP
219 subjects | PPG | 107 | Correlation, RLF, minimum redundancy maximum correlation | LR, RT, Gaussian process regression, SVR, integration tree regression | RSME: 6.74 | RSME: 3.59 | | Khalid et al. [57] | Queensland
8133 segments | PPG | 5 | VIF | MLR, SVR, RT | ME: -0.1
STD: 6.5 | ME: -0.6
STD: 5.2 | | Dey et al. [68] | Self-collected
206 subjects | PPG | 233 | — | Lasso | MAE: 6.9 | MAE: 5 | | Tan et al. [66] | Self-collect
10 subjects | PPG, ECG | 17 | MIV | GA-BP | RMSE: 2.114 | RMSE: 1.30 |
|
|
