Table 9.
Summarize the performance of the DL models reviewed and the proposed model.
| Study | Year | Dataset | Algorithm | Activation Function | Data Preprocessing Technique | Accuracy Metrics |
|---|---|---|---|---|---|---|
| 9 | 2023 | PID | CNN, DNN, and MLP | Sigmoid or Soft max | - Data and feature augmentation by replacing missing values with the mean value | 98.1% |
| 10 | 2022 | PID | SVM, LR, ANN, CNN, RNN, LSTM |
Sigmoid and RELU |
- Replace missing values with the mean value | 81% |
| 11 | 2021 | PID | DT, NN KNN, RF, NB, AB, LR and SVM | RELU |
-WEKA Analysis Tool - Replace the missing values with the mean value - Pearson’s correlation technique |
88.6% |
| 12 | 2021 | PID | VAE, SAE, MLP and CNN | Sigmoid |
- Normalization using Max–Min, Mean, and Logarithmic - Data augmentation using a VAE and feature augmentation using the SAE |
92.31% |
| 13 | 2020 | PID | MLFFNN | SELU and ELU | - Imputing the missing values with the Mean value | 84.17% |
| 14 | 2021 | PID | MLP and SVM | Gaussian RBF | Imputing the missing values with the Mean value | 77.474% |
| 15 | 2020 |
PID and DT |
DNN | Softmax and linear | Bach normalization using the mean value | 99.4112% |
| The Proposed Model | MUCHD | DNN and MLP | ReLU and Sigmoid | Imputing the missing values with the Mean value | 99.8% | |