Improved deep learning algorithm with innovation perspective: a prediction model of the mortality of respiratory infections

Public health is now seriously threatened by the COVID-19 outbreak. COVID-19 infections have spread to most countries, but certain regions have had more infections and casualties than others. It is yet unclear what causes these variances specifically. This motivates us to investigate, the association between air pollutants, metrological indices, and COVID-19 cases and deaths. We collected the daily air pollution, metrological and COVID-19 infected cases data and predicted the respiratory casualty. In this study, we assess the impact of air pollution and the metrological indicators on the respiratory infection casualty. First, we assessed how air pollution and metrological parameters correlate to respiratory infection transmission. Our findings highlight that temperature, wind speed, and particulate matter (PM2.5) positively correlate to respiratory virus transmission. In this study, an Enhanced Regularization Function in the Artificial neural networks (ERF-ANN) model predicts respiratory casualty. The ERF-ANN model was found to have minimal errors when predicting respiratory casualties over the rest. We conclude that respiratory infection transmission prefers low temperatures and polluted air. This system will alert chronic patients early based on their environment, and all disease groups will be notified.