COVID-19 Mortality Rate Prediction for India Using Statistical Neural Network Models

Area of Science:

• Computational epidemiology
• Machine learning applications in public health
• Predictive modeling for infectious diseases

Background:

• Accurate COVID-19 mortality prediction is crucial for public health resource allocation.
• Statistical Neural Networks (SNNs) offer potential for complex time-series forecasting.
• Hybrid models can enhance the accuracy of predictive algorithms.

Purpose of the Study:

• To investigate and compare the efficacy of various SNN models for COVID-19 mortality prediction in India.
• To develop hybrid SNN models to improve prediction accuracy.
• To estimate future COVID-19 death cases in the Indian population.

Main Methods:

• Application of SNN models: Probabilistic Neural Network (PNN), Radial Basis Function Neural Network (RBFNN), and Generalized Regression Neural Network (GRNN).
• Development of hybrid models by combining SNNs with Non-linear Autoregressive Neural Network (NAR-NN) to predict and correct errors.
• Performance evaluation using Root Mean Square Error (RMSE) and correlation coefficient (R) on two datasets (D1 and D2).

Main Results:

• The Probabilistic Neural Network (PNN)-based model demonstrated superior performance on dataset D2.
• The Radial Basis Function Neural Network (RBFNN)-based model achieved the best results on dataset D1.
• Hybrid models showed potential for improving the accuracy of mortality rate prediction.

Conclusions:

• Specific SNN models, namely PNN and RBFNN, are effective for COVID-19 mortality prediction in the Indian context.
• Hybridization of SNN models with NAR-NN can enhance predictive accuracy for mortality rates.
• The findings contribute to better forecasting of COVID-19 deaths in India.