Comparative study of machine learning methods for COVID-19 transmission forecasting.

Area of Science:

• Epidemiology
• Artificial Intelligence
• Data Science

Background:

• The COVID-19 pandemic necessitates advanced technological solutions for accurate forecasting.
• Machine learning (ML) has demonstrated utility in managing past epidemics.
• Accurate short-term forecasting of COVID-19 spread is crucial for hospital resource management and optimization.

Purpose of the Study:

• To conduct a comparative analysis of various machine learning methods for forecasting COVID-19 transmission.
• To evaluate the performance of deep learning models against baseline ML models.

Main Methods:

• Investigated deep learning models: LSTM-CNN, GAN-GRU, GAN, CNN, LSTM, RBM.
• Included baseline ML models: Logistic Regression (LR) and Support Vector Regression (SVR).
• Utilized time-series data of confirmed and recovered COVID-19 cases from seven countries.

Main Results:

• Hybrid deep learning models, specifically LSTM-CNN and GAN-GRU, demonstrated efficient forecasting capabilities for COVID-19 cases.
• Deep learning models significantly outperformed baseline ML models (LR, SVR).
• LSTM-CNN achieved superior performance with an average Mean Absolute Percentage Error (MAPE) of 3.718%.

Conclusions:

• Hybrid deep learning models are effective for forecasting COVID-19 transmission trends.
• Advanced ML techniques offer superior predictive accuracy compared to traditional methods.
• Accurate forecasting aids in better pandemic management and resource allocation.