KNN approach for COVID-19 diagnosis with the use of Chi-square for Feature selection

Abstract

The coronavirus disease 2019 epidemic (COVID-19) that arose has led to a key community medical hazard in Wuhan, China. Indicators that could be utilized as best prognosticators for COVID-19 patients' experimental results are crucial to be recognized. Machine learning (ML) is effective in helping to make decisions and forecasts based on the vast volume of data generated by the healthcare sector. The most imaginable approach to improving the probability of patient recovery is early detection of COVID-19 illness. In this article, a novel approach named K-Nearest-Neighbour technique (CHI-KNN) was used to diagnose the stage of patients' illness, for which a chi-square algorithm was used to effectively select features to minimize the dimensions of the dataset and increase the classifier speed. The greatest value for k was calculated utilizing an investigational method to maximize the precision of the projected algorithm. The system was evaluated using the confusion matrix in terms of accuracy, precision, and sensitivity. The system had a performance accuracy of 99.5%. The study was also compared the proposed CHI+KNN algorithm with other previous researches