# An Optimized Model Based on Deep Learning and Gated Recurrent Unit for COVID-19 Death Prediction

^{1}

^{2}

^{3}

^{4}

^{5}

^{6}

^{7}

^{8}

^{9}

^{10}

^{11}

^{12}

^{13}

^{14}

^{15}

^{16}

^{17}

^{18}

^{19}

^{*}

## Abstract

**:**

^{2}). ANOVA and Wilcoxon signed-rank tests were used to determine the statistical significance of the presented model. The experimental findings showed that the CNN-GRU model outperformed other models regarding COVID-19 death prediction.

## 1. Introduction

^{2}). The main contributions of this study are as follows:

- The research utilizes ML models for the death prediction of COVID-19 based on an Indian dataset from Kaggle; it includes eight features and 4692 instances.
- Data preprocessing was performed on the chosen dataset using data normalization and mean imputation.
- A novel CNN-GRU model with an IoT-based framework is suggested for predicting suspected cases of the COVID-19 pandemic.
- The proposed model (CNN-GRU) and some of the ML models (Random Forest (RF) Regressor, Support Vector Regressor (SVR), K-Nearest Neighbor (KNN) Regressor, Bayesian Ridge (BR) Regressor, Gradient Boosting (GB) Regressor, and Dummy Regressor (DR)) are examined and compared.
- The evaluation of the proposed approach was applied using MAE, MedAE, MSE, R2, and RMSE. The obtained results illustrated that the CNN-GRU model performed better than other prediction models and several studies.
- ANOVA and Wilcoxon signed-rank tests are used to determine if model performance differences are statistically significant.

## 2. Related Work

- It achieves better prediction performance on the Indian dataset, which is a challenging dataset due to its high variability.
- It is more robust to noise and outliers in the data.
- It is more interpretable, meaning that it is easier to understand how the model makes predictions.

- The model could be used to predict the number of COVID-19 deaths in different regions and countries. This information could be used to allocate resources more effectively and to develop targeted interventions to reduce mortality.
- The model could be used to identify high-risk populations for COVID-19 mortality. This information could be used to develop targeted public health campaigns and to provide support to vulnerable individuals.
- The model could be used to predict the impact of different public health measures on COVID-19 mortality. This information could be used to inform decision-making about how to best control the pandemic.

## 3. The Proposed CNN-GRU Model

^{2}), were employed. The proposed methodology is visually represented in Figure 2, outlining the steps taken to develop and evaluate the models. These rigorous testing and validation methods ensure that the models are dependable and accurate, making them valuable tools in various applications.

Algorithm 1: Proposed CNN-GRU for COVID-19 death prediction |

1. Input: COVID-19 dataset D, Number of CNNs N.2. Initialize GRU parameters.//Preprocess dataset 3. Normalize sample in dataset D.4. Divide D into 2 subsets: training and testing.5. Define the CNN layer with filters, kernel size, activation function, and padding.6. Apply the CNN layer to the input data to extract relevant features.7. Define the GRU layer with hidden units, activation function, and dropout rate.//Train CNNs 8. For i = 1 to n do9. Train CNN using the training set//Build the GRU model 10. Add the GRU layer of L1 units and set dropout = d1 and recurrent dropout = s1. 11. Compute update gate d_{t}, reset gate s_{t} using Equations (7) and (8).12. Compute the candidate state p_{t} using Equation (9)13. While stopping criteria did not met do14. While training for all instances do15. Calculate linear function as an activation function used in the output layer. 16. Update weights and bias17. End while18. End while//Test the proposed model 19. Test hyperparameters with the test dataset.20. Return evaluate result in the test dataset. |

## 4. Statistical Analysis of Dataset

## 5. Performance Indicators

^{2}[51].

#### 5.1. Mean Squared Error (MSE)

_{t}represents the predicted value at time t.

#### 5.2. Mean Absolute Error (MAE)

#### 5.3. Median Absolute Error (MedAE)

#### 5.4. Root Mean Squared Error (RMSE)

#### 5.5. Coefficient of Determination (R^{2})

^{2}mathematical equation.

## 6. Experimental Results

^{2}metrics.

- Random Forest (RF): N_estimators = 20: The number of estimators determines the number of decision trees in the random forest ensemble. A higher number of estimators can improve performance, but it also increases computational complexity. The value of 20 was likely chosen as a trade-off between accuracy and computational efficiency.
- K-Nearest Neighbors (KNN): N_neighbors = 10: This parameter specifies the number of neighbors to consider for classification or regression. Choosing 10 neighbors suggests that the model should consider a relatively large neighborhood for making predictions. The weights = “distance” determines the weight assigned to each neighbor during prediction. By setting it to “distance,” the model gives higher weight to closer neighbors, which can be useful when the distribution of data points is uneven.
- Support Vector Regression (SVR): Tol = 0.01: This parameter represents the tolerance for stopping criteria. A smaller tolerance value can lead to a more precise solution at the cost of increased computation time. C = 1: The C parameter controls the trade-off between achieving a smaller training error and a larger margin. A smaller C value allows more errors in the training set but may result in a wider margin. Kernel = “rbf”: The kernel parameter specifies the type of kernel function to be used. “Rbf” stands for radial basis function, which is commonly used for non-linear regression problems.
- Gradient Boosting (GB): Learning_rate = 0.1: This parameter determines the step size at each boosting iteration. A smaller learning rate can make the model converge more slowly but may lead to better generalization. n_estimators = 200: The number of boosting stages to perform. Increasing the number of estimators can improve the model’s performance, but it also increases the computational cost. max_depth = 3: This parameter sets the maximum depth of each decision tree in the gradient boosting ensemble. Limiting the depth can prevent overfitting and promote better generalization.
- Decision Tree Regression (DR): Strategy = “mean”: This parameter specifies the strategy to use when a node in the decision tree has no samples. The “mean” strategy replaces the missing value with the mean of the target values of the samples in that node.
- Bayesian Ridge (BR): N_iter = 300: The number of iterations for the Bayesian Ridge estimator. Increasing the number of iterations allows the model to refine its estimates further. tol = 0.001: This parameter sets the tolerance for convergence. A smaller tolerance value indicates a more precise convergence criterion.

^{−5}, MAE of 0.003, MedAE of 0.0017, RMSE of 0.0057, and R

^{2}of 0.64. The K-Nearest Neighbors (KNN) model had an MSE of 3.7 × 10

^{−5}, MAE of 0.004, MedAE of 0.0023, RMSE of 0.0060, and R

^{2}of 0.60. The Support Vector Regressor (SVR) model had an MSE of 1.8 × 10

^{−6}, MAE of 0.001, MedAE of 0.0008, RMSE of 0.0013, and R

^{2}of 0.96. The Gradient Boosting (GB) model had an MSE of 4.2 × 10

^{−5,}MAE of 0.004, MedAE of 0.0028, RMSE of 0.0065, and R

^{2}of 0.53. The Dummy Regressor (DR) model had an MSE of 7.06 × 10

^{−5}, MAE of 0.006, MedAE of 0.0043, RMSE of 0.0084, and R

^{2}of 0.23. The Bayesian Ridge (BR) model had an MSE of 7.4 × 10

^{−5}, MAE of 0.006, MedAE of 0.0045, RMSE of 0.0085, and R

^{2}of 0.22. The proposed CNN-GRU model achieved the best results; it had an MSE of 1.14 × 10

^{−9}, MAE of 2.5 × 10

^{−5}, MedAE of 1.8 × 10

^{−5}, RMSE of 3.3 × 10

^{−5}, and R

^{2}of 0.99.

## 7. Conclusions and Perspectives

^{2}. Thus, the CNN-GRU model is recommended for similar predictive tasks. There are several possible avenues for future work in this area:

- Incorporate more data sources: Many potential data sources could be used to improve COVID-19 death prediction models. For example, social media data could be used to track the spread of misinformation about COVID-19, which could help identify regions at a higher risk of experiencing a surge in cases and deaths. Additionally, data on weather patterns, pollution levels, and other environmental factors could be integrated to better understand how these variables affect the spread and severity of COVID-19.
- Improve model accuracy: There are several ways to improve the accuracy of COVID-19 death prediction models. One approach is to use more sophisticated machine learning algorithms, such as reinforcement learning, which can handle more complex data and provide more accurate predictions. Another method is to improve the data quality used to train the models, for example, by incorporating more granular data on individual patients’ health status and medical history.
- Develop models for specific populations: COVID-19 has been shown to affect different populations in different ways, with some people (such as the elderly or those with underlying health conditions) at a higher risk of death than others. Therefore, developing targeted models for specific populations could be an effective way to improve the accuracy of COVID-19 death predictions and better allocate resources for prevention and treatment.
- Combine prediction models with interventions: Predicting COVID-19 death is only helpful if the information can be used to take action to prevent deaths. Therefore, future work could integrate death prediction models with intervention strategies, such as targeted vaccination campaigns, lockdown measures, or specific treatments for high-risk patients. By combining prediction models with interventions, public health officials could take a more proactive approach to managing the pandemic and reducing the number of COVID-19 deaths.

## Author Contributions

## Funding

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## Appendix A

**Table A1.**An overview of the proposed CNN-GRU model and an examination of the outcomes of alternative models, focusing on the RMSE metric.

RF | KNN | SVR | GB | DR | BR | CNN-GRU | |
---|---|---|---|---|---|---|---|

Number of values | 10 | 10 | 10 | 10 | 10 | 10 | 10 |

Minimum | 0.002004 | 0.002132 | 0.000128 | 0.001919 | 0.002729 | 0.00339 | 1.41 × 10^{−6} |

25% Percentile | 0.002431 | 0.002559 | 0.000554 | 0.002665 | 0.003582 | 0.003625 | 1.41 × 10^{−6} |

Median | 0.002431 | 0.002559 | 0.000554 | 0.002772 | 0.003582 | 0.003625 | 1.41 × 10^{−6} |

75% Percentile | 0.002431 | 0.002559 | 0.000653 | 0.002772 | 0.003582 | 0.003625 | 1.42 × 10^{−6} |

Maximum | 0.002844 | 0.002985 | 0.000981 | 0.003198 | 0.004009 | 0.003902 | 1.49 × 10^{−6} |

Range | 0.00084 | 0.000853 | 0.000853 | 0.001279 | 0.00128 | 0.000512 | 8 × 10^{−8} |

Mean | 0.00243 | 0.002559 | 0.000594 | 0.002687 | 0.003539 | 0.003629 | 1.42 × 10^{−6} |

Std. Deviation | 0.000198 | 0.000201 | 0.000237 | 0.000336 | 0.000315 | 0.000121 | 2.7 × 10^{−8} |

Std. Error of Mean | 6.26 × 10^{−5} | 6.36 × 10^{−5} | 7.5 × 10^{−5} | 0.000106 | 9.95 × 10^{−5} | 3.83 × 10^{−5} | 8.54 × 10^{−9} |

Sum | 0.0243 | 0.02559 | 0.005938 | 0.02687 | 0.03539 | 0.03629 | 1.42 × 10^{−5} |

ANOVA Table | SS | DF | MS | F (DFn, DFd) | p Value |
---|---|---|---|---|---|

Treatment (between columns) | 0.000117 | 6 | 1.95 × 10^{−5} | F (6, 63) = 375.3 | p < 0.0001 |

Residual (within columns) | 3.26 × 10^{−6} | 63 | 5.18 × 10^{−8} | ||

Total | 0.00012 | 69 |

**Table A3.**The Wilcoxon Signed-Rank test applied to the showcased CNN-GRU model and alternative models.

RF | KNN | SVR | GB | DR | BR | CNN-GRU | |
---|---|---|---|---|---|---|---|

Theoretical median | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

Actual median | 0.002431 | 0.002559 | 0.000554 | 0.002772 | 0.003582 | 0.003625 | 1.41 × 10^{−6} |

Number of values | 10 | 10 | 10 | 10 | 10 | 10 | 10 |

Wilcoxon Signed Rank Test | |||||||

Sum of signed ranks (W) | 55 | 55 | 55 | 55 | 55 | 55 | 55 |

Sum of positive ranks | 55 | 55 | 55 | 55 | 55 | 55 | 55 |

Sum of negative ranks | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

p value (two tailed) | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 |

Exact or estimate? | Exact | Exact | Exact | Exact | Exact | Exact | Exact |

Significant (alpha = 0.05)? | Yes | Yes | Yes | Yes | Yes | Yes | Yes |

How big is the discrepancy? | |||||||

Discrepancy | 0.002431 | 0.002559 | 0.000554 | 0.002772 | 0.003582 | 0.003625 | 1.41 × 10^{−6} |

**Figure A1.**Graph depicting the showcased CNN-GRU model and alternative models with respect to the RMSE.

**Figure A3.**Visualize and compare the supplied CNN-GRU model to others. (

**a**) residual plot, (

**b**) homoscedasticity plot, (

**c**) QQ plot, and (

**d**) heat maps.

## References

- Elshewey, A.M.; Shams, M.Y.; El-Rashidy, N.; Elhady, A.M.; Shohieb, S.M.; Tarek, Z. Bayesian Optimization with Support Vector Machine Model for Parkinson Disease Classification. Sensors
**2023**, 23, 2085. [Google Scholar] [CrossRef] [PubMed] - Mohammed, M.N.; Syamsudin, H.; Al-Zubaidi, S.; Sairah, A.; Ramli, R.; Yusuf, E. Novel COVID-19 Detection and Diagnosis System Using IOT Based Smart Helmet. Int. J. Psychosoc. Rehabil.
**2020**, 24, 2296–2303. [Google Scholar] - Alsaeedy, A.A.R.; Chong, E.K.P. Detecting Regions at Risk for Spreading COVID-19 Using Existing Cellular Wireless Network Functionalities. IEEE Open J. Eng. Med. Biol.
**2020**, 1, 187–189. [Google Scholar] [CrossRef] [PubMed] - Didi, Y.; Walha, A.; Wali, A. COVID-19 Tweets Classification Based on a Hybrid Word Embedding Method. Big Data Cogn. Comput.
**2022**, 6, 58. [Google Scholar] [CrossRef] - Arun, M.; Baraneetharan, E.; Kanchana, A.; Prabu, S. Detection and Monitoring of the Asymptotic COVID-19 Patients Using IoT Devices and Sensors. Int. J. Pervasive Comput. Commun.
**2020**, 18, 407–418. [Google Scholar] - Kollu, P.K.; Kumar, K.; Kshirsagar, P.R.; Islam, S.; Naveed, Q.N.; Hussain, M.R.; Sundramurthy, V.P. Development of Advanced Artificial Intelligence and IoT Automation in the Crisis of COVID-19 Detection. J. Healthc. Eng.
**2022**, 2022, 1987917. [Google Scholar] [CrossRef] - Adeniyi, E.A.; Ogundokun, R.O.; Awotunde, J.B. IoMT-Based Wearable Body Sensors Network Healthcare Monitoring System. IoT Healthc. Ambient Assist. Living
**2021**, 933, 103–121. [Google Scholar] - Yang, T.; Gentile, M.; Shen, C.-F.; Cheng, C.-M. Combining Point-of-Care Diagnostics and Internet of Medical Things (IoMT) to Combat the COVID-19 Pandemic. Diagnostics
**2020**, 10, 224. [Google Scholar] [CrossRef] - Alotaibi, S.; Al-Rasheed, A.; Althahabi, S.; Hamza, M.; Mohamed, A.; Zamani, A.; Motwakel, A.; Eldesouki, M. Optimal Kernel Extreme Learning Machine for COVID-19 Classification on Epidemiology Dataset. CMC–Comput. Mater. Contin.
**2022**, 73, 3305–3318. [Google Scholar] [CrossRef] - Elzeki, O.M.; Shams, M.; Sarhan, S.; Abd Elfattah, M.; Hassanien, A.E. COVID-19: A New Deep Learning Computer-Aided Model for Classification. PeerJ Comput. Sci.
**2021**, 7, e358. [Google Scholar] [CrossRef] - Elzeki, O.M.; Abd Elfattah, M.; Salem, H.; Hassanien, A.E.; Shams, M. A Novel Perceptual Two Layer Image Fusion Using Deep Learning for Imbalanced COVID-19 Dataset. PeerJ Comput. Sci.
**2021**, 7, e364. [Google Scholar] [CrossRef] [PubMed] - Singh, D.; Kumar, V.; Kaur, M.; Jabarulla, M.Y.; Lee, H.-N. Screening of COVID-19 Suspected Subjects Using Multi-Crossover Genetic Algorithm Based Dense Convolutional Neural Network. IEEE Access
**2021**, 9, 142566–142580. [Google Scholar] [CrossRef] - Woźniak, M.; Wieczorek, M.; Siłka, J. BiLSTM Deep Neural Network Model for Imbalanced Medical Data of IoT Systems. Future Gener. Comput. Syst.
**2023**, 141, 489–499. [Google Scholar] [CrossRef] - Abdelhamid, A.A.; El-Kenawy, E.-S.M.; Khodadadi, N.; Mirjalili, S.; Khafaga, D.S.; Alharbi, A.H.; Ibrahim, A.; Eid, M.M.; Saber, M. Classification of Monkeypox Images Based on Transfer Learning and the Al-Biruni Earth Radius Optimization Algorithm. Mathematics
**2022**, 10, 3614. [Google Scholar] [CrossRef] - Eid, M.M.; El-Kenawy, E.-S.M.; Khodadadi, N.; Mirjalili, S.; Khodadadi, E.; Abotaleb, M.; Alharbi, A.H.; Abdelhamid, A.A.; Ibrahim, A.; Amer, G.M. Meta-Heuristic Optimization of LSTM-Based Deep Network for Boosting the Prediction of Monkeypox Cases. Mathematics
**2022**, 10, 3845. [Google Scholar] [CrossRef] - Khafaga, D.S.; Ibrahim, A.; El-Kenawy, E.-S.M.; Abdelhamid, A.A.; Karim, F.K.; Mirjalili, S.; Khodadadi, N.; Lim, W.H.; Eid, M.M.; Ghoneim, M.E. An Al-Biruni Earth Radius Optimization-Based Deep Convolutional Neural Network for Classifying Monkeypox Disease. Diagnostics
**2022**, 12, 2892. [Google Scholar] [CrossRef] - Khafaga, D. Meta-Heuristics for Feature Selection and Classification in Diagnostic Breast Cancer. Comput. Mater. Contin.
**2022**, 73, 749–765. [Google Scholar] - Alhussan, A.A.; Abdelhamid, A.A.; Towfek, S.K.; Ibrahim, A.; Eid, M.M.; Khafaga, D.S.; Saraya, M.S. Classification of Diabetes Using Feature Selection and Hybrid Al-Biruni Earth Radius and Dipper Throated Optimization. Diagnostics
**2023**, 13, 2038. [Google Scholar] [CrossRef] - Chaki, J.; Woźniak, M. Deep Learning for Neurodegenerative Disorder (2016 to 2022): A Systematic Review. Biomed. Signal Process. Control
**2023**, 80, 104223. [Google Scholar] [CrossRef] - Samee, N.A.; El-Kenawy, E.-S.M.; Atteia, G.; Jamjoom, M.M.; Ibrahim, A.; Abdelhamid, A.A.; El-Attar, N.E.; Gaber, T.; Slowik, A.; Shams, M.Y. Metaheuristic Optimization through Deep Learning Classification of COVID-19 in Chest X-Ray Images. Comput. Mater. Contin.
**2022**, 73, 4193–4210. [Google Scholar] - Almalki, J.; Al Shehri, W.; Mehmood, R.; Alsaif, K.; Alshahrani, S.M.; Jannah, N.; Khan, N.A. Enabling Blockchain with IoMT Devices for Healthcare. Information
**2022**, 13, 448. [Google Scholar] [CrossRef] - Awotunde, J.B.; Ajagbe, S.A.; Idowu, I.R.; Ndunagu, J.N. An Enhanced Cloud-IoMT-Based and Machine Learning for Effective COVID-19 Diagnosis System. Intell. Things Ai-Iot Based Crit.-Appl. Innov.
**2021**, 55–76. [Google Scholar] - Jain, S.; Nehra, M.; Kumar, R.; Dilbaghi, N.; Hu, T.; Kumar, S.; Kaushik, A.; Li, C.-Z. Internet of Medical Things (IoMT)-Integrated Biosensors for Point-of-Care Testing of Infectious Diseases. Biosens. Bioelectron.
**2021**, 179, 113074. [Google Scholar] [CrossRef] [PubMed] - Irkham, I.; Ibrahim, A.U.; Nwekwo, C.W.; Al-Turjman, F.; Hartati, Y.W. Current Technologies for Detection of COVID-19: Biosensors, Artificial Intelligence and Internet of Medical Things (IoMT). Sensors
**2022**, 23, 426. [Google Scholar] [CrossRef] - Un, K.-C.; Wong, C.-K.; Lau, Y.-M.; Lee, J.C.-Y.; Tam, F.C.-C.; Lai, W.-H.; Lau, Y.-M.; Chen, H.; Wibowo, S.; Zhang, X. Observational Study on Wearable Biosensors and Machine Learning-Based Remote Monitoring of COVID-19 Patients. Sci. Rep.
**2021**, 11, 4388. [Google Scholar] [CrossRef] - Al Bassam, N.; Hussain, S.A.; Al Qaraghuli, A.; Khan, J.; Sumesh, E.P.; Lavanya, V. IoT Based Wearable Device to Monitor the Signs of Quarantined Remote Patients of COVID-19. Inform. Med. Unlocked
**2021**, 24, 100588. [Google Scholar] [CrossRef] - Subramanian, M.; Shanmuga Vadivel, K.; Hatamleh, W.A.; Alnuaim, A.A.; Abdelhady, M.; VE, S. The Role of Contemporary Digital Tools and Technologies in COVID-19 Crisis: An Exploratory Analysis. Expert Syst.
**2022**, 39, e12834. [Google Scholar] [CrossRef] - Wieczorek, M.; Siłka, J.; Woźniak, M. Neural Network Powered COVID-19 Spread Forecasting Model. Chaos Solitons Fractals
**2020**, 140, 110203. [Google Scholar] [CrossRef] - Ajaz, F.; Naseem, M.; Sharma, S.; Shabaz, M.; Dhiman, G. COVID-19: Challenges and Its Technological Solutions Using IoT. Curr. Med. Imaging
**2022**, 18, 113–123. [Google Scholar] [CrossRef] - Mir, M.H.; Jamwal, S.; Mehbodniya, A.; Garg, T.; Iqbal, U.; Samori, I.A. IoT-Enabled Framework for Early Detection and Prediction of COVID-19 Suspects by Leveraging Machine Learning in Cloud. J. Healthc. Eng.
**2022**, 2022, 7713939. [Google Scholar] [CrossRef] - Kini, A.S.; Gopal Reddy, A.N.; Kaur, M.; Satheesh, S.; Singh, J.; Martinetz, T.; Alshazly, H. Ensemble Deep Learning and Internet of Things-Based Automated COVID-19 Diagnosis Framework. Contrast Media Mol. Imaging
**2022**, 2022, 7377502. [Google Scholar] [CrossRef] [PubMed] - Al-Dhaen, F.; Hou, J.; Rana, N.P.; Weerakkody, V. Advancing the Understanding of the Role of Responsible AI in the Continued Use of IoMT in Healthcare. Inf. Syst. Front.
**2021**, 1–20. [Google Scholar] [CrossRef] [PubMed] - Asghari, P. A Diagnostic Prediction Model for Colorectal Cancer in Elderlies via Internet of Medical Things. Int. J. Inf. Technol.
**2021**, 13, 1423–1429. [Google Scholar] [CrossRef] [PubMed] - Elhoseny, M.; Tarek, Z.; El-Hasnony, I.M. Advanced Cognitive Algorithm for Biomedical Data Processing: COVID-19 Pattern Recognition as a Case Study. J. Healthc. Eng.
**2022**, 2022, 1773259. [Google Scholar] [CrossRef] [PubMed] - Khan, S.H.; Sohail, A.; Khan, A.; Hassan, M.; Lee, Y.S.; Alam, J.; Basit, A.; Zubair, S. COVID-19 Detection in Chest X-Ray Images Using Deep Boosted Hybrid Learning. Comput. Biol. Med.
**2021**, 137, 104816. [Google Scholar] [CrossRef] [PubMed] - Dutta, S.; Bandyopadhyay, S.K. Machine Learning Approach for Confirmation of COVID-19 Cases: Positive, Negative, Death and Release. MedRxiv
**2020**. [Google Scholar] [CrossRef] - Ghafouri-Fard, S.; Mohammad-Rahimi, H.; Motie, P.; Minabi, M.A.S.; Taheri, M.; Nateghinia, S. Application of Machine Learning in the Prediction of COVID-19 Daily New Cases: A Scoping Review. Heliyon
**2021**, 7, e08143. [Google Scholar] [CrossRef] - Li, Y.; Zou, C.; Berecibar, M.; Nanini-Maury, E.; Chan, J.C.-W.; Van den Bossche, P.; Van Mierlo, J.; Omar, N. Random Forest Regression for Online Capacity Estimation of Lithium-Ion Batteries. Appl. Energy
**2018**, 232, 197–210. [Google Scholar] [CrossRef] - Deng, Z.; Zhu, X.; Cheng, D.; Zong, M.; Zhang, S. Efficient kNN Classification Algorithm for Big Data. Neurocomputing
**2016**, 195, 143–148. [Google Scholar] [CrossRef] - Panahi, M.; Sadhasivam, N.; Pourghasemi, H.R.; Rezaie, F.; Lee, S. Spatial Prediction of Groundwater Potential Mapping Based on Convolutional Neural Network (CNN) and Support Vector Regression (SVR). J. Hydrol.
**2020**, 588, 125033. [Google Scholar] [CrossRef] - Feng, J.; Yu, Y.; Zhou, Z.-H. Multi-Layered Gradient Boosting Decision Trees. In Proceedings of the Advances in Neural Information Processing Systems; Curran Associates, Inc.: Red Hook, NY, USA, 2018; Volume 31. [Google Scholar]
- Elshewey, A.M.; Shams, M.Y.; Elhady, A.M.; Shohieb, S.M.; Abdelhamid, A.A.; Ibrahim, A.; Tarek, Z. A Novel WD-SARIMAX Model for Temperature Forecasting Using Daily Delhi Climate Dataset. Sustainability
**2022**, 15, 757. [Google Scholar] [CrossRef] - Shambour, M.K. Analyzing Perceptions of a Global Event Using CNN-LSTM Deep Learning Approach: The Case of Hajj 1442 (2021). PeerJ Comput. Sci.
**2022**, 8, e1087. [Google Scholar] [CrossRef] - ArunKumar, K.E.; Kalaga, D.V.; Kumar, C.M.S.; Kawaji, M.; Brenza, T.M. Comparative Analysis of Gated Recurrent Units (GRU), Long Short-Term Memory (LSTM) Cells, Autoregressive Integrated Moving Average (ARIMA), Seasonal Autoregressive Integrated Moving Average (SARIMA) for Forecasting COVID-19 Trends. Alex. Eng. J.
**2022**, 61, 7585–7603. [Google Scholar] [CrossRef] - Elshewey, A.; Shams, M.; Tarek, Z.; Megahed, M.; El-kenawy, E.-S.; El-dosuky, M. Weight Prediction Using the Hybrid Stacked-LSTM Food Selection Model. CSSE
**2023**, 46, 765–781. [Google Scholar] [CrossRef] - Jiang, D.; Hu, G.; Qi, G.; Mazur, N. A Fully Convolutional Neural Network-Based Regression Approach for Effective Chemical Composition Analysis Using near-Infrared Spectroscopy in Cloud. J. Artif. Intell. Technol.
**2021**, 1, 74–82. [Google Scholar] [CrossRef] - Reshi, A.A.; Rustam, F.; Mehmood, A.; Alhossan, A.; Alrabiah, Z.; Ahmad, A.; Alsuwailem, H.; Choi, G.S. An Efficient CNN Model for COVID-19 Disease Detection Based on X-ray Image Classification. Complexity
**2021**, 2021, 6621607. [Google Scholar] [CrossRef] - Niu, Z.; Yu, Z.; Tang, W.; Wu, Q.; Reformat, M. Wind Power Forecasting Using Attention-Based Gated Recurrent Unit Network. Energy
**2020**, 196, 117081. [Google Scholar] [CrossRef] - Wang, Y.; Liao, W.; Chang, Y. Gated Recurrent Unit Network-Based Short-Term Photovoltaic Forecasting. Energies
**2018**, 11, 2163. [Google Scholar] [CrossRef] - Xu, B.; Gutierrez, B.; Mekaru, S.; Sewalk, K.; Goodwin, L.; Loskill, A.; Cohn, E.L.; Hswen, Y.; Hill, S.C.; Cobo, M.M. Epidemiological Data from the COVID-19 Outbreak, Real-Time Case Information. Sci. Data
**2020**, 7, 106. [Google Scholar] [CrossRef] - Khan, M.A.; Khan, R.; Algarni, F.; Kumar, I.; Choudhary, A.; Srivastava, A. Performance Evaluation of Regression Models for COVID-19: A Statistical and Predictive Perspective. Ain Shams Eng. J.
**2022**, 13, 101574. [Google Scholar] [CrossRef]

**Figure 1.**General structure of the proposed framework for the death prediction of the COVID-19 epidemic.

**Figure 17.**Compare and display the presented CNN-GRU model and other models. (

**a**) residual plot, (

**b**) homoscedasticity plot, (

**c**) QQ plot, and (

**d**) heat maps.

Count | Mean | Std. | Min. | 50% | Max. | |
---|---|---|---|---|---|---|

Latitude | 4692 | 23.185 | 6.6359 | 0.0 | 23.9408 | 34.2996 |

Longitude | 4692 | 81.451 | 6.9594 | 0.0 | 79.0193 | 94.7278 |

Total confirmed cases | 4692 | 11,393 | 37,208 | 1.0 | 619.0000 | 468,265 |

Cured/Discharged/Migrated | 4692 | 6908 | 23,390 | 0.0 | 197.5000 | 305,521 |

New cases | 4692 | 418.6 | 1259 | 0.0 | 26.0000 | 18,366 |

New deaths | 4692 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |

New recovered | 4692 | 283.06 | 947.9 | −1.0 | 8.0000 | 13,401 |

Death | 4692 | 291.28 | 1213 | 0.0 | 5.0000 | 16,476 |

Models | Hyperparameters |
---|---|

RF | N_estimators = 20. |

KNN | N_neighbors = 10, weights = “distance”. |

SVR | Tol = 0.01, C = 1, kernel = “rbf”. |

GB | Learning_rate = 0.1, n_estimators = 200, max_depth = 3. |

DR | Strategy = “mean”. |

BR | N_iter = 300, tol = 0.001. |

CNN | learnRate = 0.001, hiddenLayerTwo = 256, hiddenLayerOne = 256, epochs = 40, dropout = 0.4, batch_size = 32 |

Model | MSE | MAE | MedAE | RMSE | R^{2} |
---|---|---|---|---|---|

RF | 3.2 × 10^{−5} | 0.003 | 0.0017 | 0.0057 | 0.64 |

KNN | 3.7 × 10^{−5} | 0.004 | 0.0023 | 0.0060 | 0.60 |

SVR | 1.8 × 10^{−6} | 0.001 | 0.0008 | 0.0013 | 0.96 |

GB | 4.2 × 10^{−5} | 0.004 | 0.0028 | 0.0065 | 0.53 |

DR | 7.06 × 10^{−5} | 0.006 | 0.0043 | 0.0084 | 0.23 |

BR | 7.4 × 10^{−5} | 0.006 | 0.0045 | 0.0085 | 0.22 |

CNN-GRU | 1.14 × 10^{−9} | 2.5 × 10^{−5} | 1.8 × 10^{−5} | 3.3 × 10^{−5} | 0.99 |

**Table 4.**A description of the proposed CNN-GRU model as well as the results of other models based on the RMSE factor.

RF | KNN | SVR | GB | DR | BR | CNN-GRU | |
---|---|---|---|---|---|---|---|

Number of values | 10 | 10 | 10 | 10 | 10 | 10 | 10 |

Minimum | 0.0047 | 0.005 | 0.0003 | 0.0045 | 0.0064 | 0.00795 | 3.3 × 10^{−6} |

25% Percentile | 0.0057 | 0.006 | 0.0013 | 0.00625 | 0.0084 | 0.0085 | 3.3 × 10^{−6} |

Median | 0.0057 | 0.006 | 0.0013 | 0.0065 | 0.0084 | 0.0085 | 3.3 × 10^{−6} |

75% Percentile | 0.0057 | 0.006 | 0.001533 | 0.0065 | 0.0084 | 0.0085 | 3.33 × 10^{−6} |

Maximum | 0.00667 | 0.007 | 0.0023 | 0.0075 | 0.0094 | 0.00915 | 3.5 × 10^{−6} |

Range | 0.00197 | 0.002 | 0.002 | 0.003 | 0.003 | 0.0012 | 2 × 10^{−7} |

Mean | 0.005697 | 0.006 | 0.001393 | 0.0063 | 0.0083 | 0.00851 | 3.33 × 10^{−6} |

Std. Deviation | 0.000464 | 0.000471 | 0.000556 | 0.000789 | 0.000738 | 0.000284 | 6.75 × 10^{−8} |

Std. Error of Mean | 0.000147 | 0.000149 | 0.000176 | 0.000249 | 0.000233 | 8.97 × 10^{−5} | 2.13 × 10^{−8} |

Sum | 0.05697 | 0.06 | 0.01393 | 0.063 | 0.083 | 0.0851 | 3.33 × 10^{−5} |

SS | DF | MS | F (DFn, DFd) | p Value | |
---|---|---|---|---|---|

Treatment (between columns) | 0.000642 | 6 | 0.000107 | F (6, 63) = 375.4 | p < 0.0001 |

Residual (within columns) | 1.79 × 10^{−5} | 63 | 2.85 × 10^{−7} | - | - |

Total | 0.00066 | 69 | - | - | - |

RF | KNN | SVR | GB | DR | BR | CNN-GRU | |
---|---|---|---|---|---|---|---|

Theoretical median | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

Actual median | 0.0057 | 0.006 | 0.0013 | 0.0065 | 0.0084 | 0.0085 | 3.3 × 10^{−6} |

Number of values | 10 | 10 | 10 | 10 | 10 | 10 | 10 |

Wilcoxon signed-rank test | |||||||

Sum of signed ranks (W) | 55 | 55 | 55 | 55 | 55 | 55 | 55 |

Sum of positive ranks | 55 | 55 | 55 | 55 | 55 | 55 | 55 |

Sum of negative ranks | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

p value (two-tailed) | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 |

Exact or estimate? | Exact | Exact | Exact | Exact | Exact | Exact | Exact |

Significant (alpha = 0.05)? | Yes | Yes | Yes | Yes | Yes | Yes | Yes |

How big is the discrepancy? | |||||||

Discrepancy | 0.0057 | 0.006 | 0.0013 | 0.0065 | 0.0084 | 0.0085 | 3.3 × 10^{−6} |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Tarek, Z.; Shams, M.Y.; Towfek, S.K.; Alkahtani, H.K.; Ibrahim, A.; Abdelhamid, A.A.; Eid, M.M.; Khodadadi, N.; Abualigah, L.; Khafaga, D.S.;
et al. An Optimized Model Based on Deep Learning and Gated Recurrent Unit for COVID-19 Death Prediction. *Biomimetics* **2023**, *8*, 552.
https://doi.org/10.3390/biomimetics8070552

**AMA Style**

Tarek Z, Shams MY, Towfek SK, Alkahtani HK, Ibrahim A, Abdelhamid AA, Eid MM, Khodadadi N, Abualigah L, Khafaga DS,
et al. An Optimized Model Based on Deep Learning and Gated Recurrent Unit for COVID-19 Death Prediction. *Biomimetics*. 2023; 8(7):552.
https://doi.org/10.3390/biomimetics8070552

**Chicago/Turabian Style**

Tarek, Zahraa, Mahmoud Y. Shams, S. K. Towfek, Hend K. Alkahtani, Abdelhameed Ibrahim, Abdelaziz A. Abdelhamid, Marwa M. Eid, Nima Khodadadi, Laith Abualigah, Doaa Sami Khafaga,
and et al. 2023. "An Optimized Model Based on Deep Learning and Gated Recurrent Unit for COVID-19 Death Prediction" *Biomimetics* 8, no. 7: 552.
https://doi.org/10.3390/biomimetics8070552