Classifier Model for COVID-19 Diagnosis using Hybrid Algorithms in Data Mining

Abstract

The outbreak of Covid-19 has created a difficult situation for all people around the world and the number of deaths is increasing daily. Diagnosis of Covid-19 by blood test (RT-PCR) is time consuming and experts are looking for a faster solution to control and counteract the further spread of the virus worldwide using non-clinical methods such as data technology. Mining, machine learning and artificial intelligence. Because the healthcare industry generates large amounts of data, we can use data mining to find hidden and understandable patterns that may help in rapid diagnosis and effective and efficient decision making. Prediction and diagnostic algorithms can reduce the pressure on health care systems by accurately and quickly identifying diseases. In this study, a proposed model for more accurate and faster diagnosis of patients with Covid-19 and healthy individuals using basic and combined data mining algorithms is presented. The data set includes: electronic medical and laboratory records of patients in Imam Reza (AS) Hospital in Mashhad, which has been implemented with the help of Python software version 3.7 and Veka version 3.9. We used basic algorithms such as: Naive Bayes, Decission Tree, K- nearest neighborhood, Support Vector Machine, Random Forest, Ada-Boost, Bagging, Majority Voting, XGBoost and Stacking. The results of the present study showed that the proposed model achieved an accuracy of 83% by using a combination of basic algorithms in the stacking classification, which used the gradient boosting algorithm in the meta part.