Covid-19 Patient Health Prediction Using Boosted Random Forest Algorithm

Abstract

COVID-19, also known as 2019-nCoV, is no longer a pandemic but an endemic disease that has killed many people worldwide. COVID-19 has no precise treatment or remedy at this time, but it is unavoidable to live with the disease and its implications. By quickly and efficiently screening for covid, one may determine whether or not one has COVID-19 and thus limit the financial and administrative burdens on healthcare systems. This reality puts a huge demand on these countries' healthcare systems, especially in emerging nations, due to the poor healthcare systems around the world. Although the COVID-19 pandemic cannot be stopped by any licenced vaccine or antiviral medicine, there are other possible solutions that could lighten the burden of the virus on healthcare systems and the economy. The most promising approaches for usage outside of a clinical environment include non-clinical approaches like machine learning, data mining, deep learning, and other artificial intelligence technologies. Artificial intelligence (AI) approaches are increasingly being integrated into wireless infrastructure, real-time data collection, and end-user device processing. A positive and negative COVID-19 case dataset is used to validate artificial intelligence (AI) systems such decision trees, support vector machines, artificial neural networks, and naive Bayesian models. The correlation coefficients between various dependent and independent variables were examined to determine the strength of the relationship between the dependent features. The model was tested 20% of the time while being trained 80% of the time during the preparation phase. The Random Forest had the highest precision (94.99%), according to the evaluation of success.