Coronavirus disease situation analysis and prediction using machine learning: a study on Bangladeshi population

TL;DR

This study develops a machine learning-based prediction system using an MLP model to forecast COVID-19 infection and death rates in Bangladesh, aiding resource allocation and treatment planning during the pandemic.

Contribution

It introduces a predictive model specifically trained on Bangladeshi COVID-19 data, demonstrating the MLP's superior accuracy over SVR and linear regression for epidemic forecasting.

Findings

MLP outperforms SVR and linear regression in prediction accuracy

Predicted COVID-19 cases range between 929-2443 with deaths between 19-57

The model provides early warning for potential future outbreaks

Abstract

During a pandemic, early prognostication of patient infected rates can reduce the death by ensuring treatment facility and proper resource allocation. In recent months, the number of death and infected rates has increased more distinguished than before in Bangladesh. The country is struggling to provide moderate medical treatment to many patients. This study distinguishes machine learning models and creates a prediction system to anticipate the infected and death rate for the coming days. Equipping a dataset with data from March 1, 2020, to August 10, 2021, a multi-layer perceptron (MLP) model was trained. The data was managed from a trusted government website and concocted manually for training purposes. Several test cases determine the model's accuracy and prediction capability. The comparison between specific models assumes that the MLP model has more reliable prediction capability than the support vector regression (SVR) and linear regression model. The model presents a report about the risky situation and impending coronavirus disease (COVID-19) attack. According to the prediction produced by the model, Bangladesh may suffer another COVID-19 attack, where the number of infected cases can be between 929 to 2443 and death cases between 19 to 57.