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

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Epidemiology

Background:

• The COVID-19 pandemic presents a global health crisis requiring innovative monitoring and prediction solutions.
• Internet of Things (IoT) offers capabilities for real-time patient monitoring and data collection in healthcare.
• Artificial Intelligence (AI) is crucial for analyzing complex health data to predict disease spread and outcomes.

Purpose of the Study:

• To develop and evaluate an enhanced AI model for predicting COVID-19 mortality.
• To leverage Internet of Things (IoT) technology for improved COVID-19 patient monitoring and risk assessment.
• To utilize an Indian dataset for training and validating a novel predictive model.

Main Methods:

• An enhanced Convolutional Neural Network with Gated Recurrent Unit (CNN-GRU) model was developed.
• Data preprocessing included normalization and imputation on an Indian COVID-19 dataset (4692 cases, 8 features).
• Model performance was evaluated using metrics like MAE, MSE, RMSE, MedAE, and R-squared, with statistical significance tested via ANOVA and Wilcoxon signed-rank tests.

Main Results:

• The proposed CNN-GRU model demonstrated superior performance in predicting COVID-19 deaths compared to other models.
• The model effectively utilized IoT data for patient monitoring and risk assessment.
• Statistical tests confirmed the significance of the model's predictive capabilities.

Conclusions:

• The AI-powered CNN-GRU model, integrated with IoT, provides an effective tool for COVID-19 death prediction.
• This approach can enhance healthcare systems' ability to manage and mitigate the impact of infectious disease outbreaks.
• The study highlights the potential of advanced machine learning techniques in public health surveillance and forecasting.