Analysis and Prediction of COVID-19 Multivariate Data Using Deep Ensemble Learning Methods.

Area of Science:

• Epidemiology
• Machine Learning
• Public Health

Background:

• The COVID-19 pandemic significantly impacted the global economy and healthcare systems.
• Accurate predictive models are crucial for effective healthcare resource management and disease spread prevention.
• Developing a universal method for predicting COVID-19 cases is essential for pandemic preparedness.

Purpose of the Study:

• To develop a robust and universal method for predicting COVID-19 positive cases.
• To assist collaborators in developing and refining pandemic response strategies.
• To enhance the accuracy and reliability of disease spread prediction models.

Main Methods:

• Proposed an adaptive gradient Long Short-Term Memory (AGLSTM) model utilizing multivariate time series data.
• Employed Convolutional Neural Networks (CNN) for feature extraction and adaptive LSTM for case prediction.
• Evaluated the model using case studies from India and incorporated data fusion and transfer-learning techniques.

Main Results:

• The AGLSTM model demonstrated superior performance with an accuracy of 99.81%.
• The model requires minimal training and prediction time.
• Comparative analysis included Recurrent Neural Networks (RNN), LSTM, LASSO regression, Ada-Boost, Light Gradient Boosting, and KNN models.

Conclusions:

• The AGLSTM model offers a highly accurate and efficient solution for predicting COVID-19 cases.
• The developed methodology can be adapted for predicting the onset of future infectious diseases.
• The findings support the use of advanced machine learning techniques in public health crisis management.