Covid-19 Prediction Analysis Using Machine Learning Approach

ABSTRACT

The unforeseen outbreak of Covid-19 which resulted in a global pandemic posed a threat to the human civilization. The entire world is trying their best to combat against the outspread of the disease. The rapid spread of the disease has put governing bodies under pressure and made it difficult to confront the situation. The RT-PCR which is the test confirming if a person has Covid-19 infection, is restricted by the shortfall of reagents, time taking, high cost and need for dedicated labs with trained pathologists. With the sudden rise in daily cases, there were large queues for Covid-19 tests, stressing the medical laboratories with many such laboratories facing shortage of kits for testing. Hence, there is a requirement for cost effective and quick diagnostic model to determine positive and negative cases of Covid-19. This paper aims to predict Covid-19 infection in an individual person from initial symptoms and information like fever, cough, sore throat using machine learning algorithms. The study includes working with six predicting models, MLP, GBC, Decision tree, SVM, Logistic Regression and Random forest with highest accuracy of 92.94% achieved in logical regression. The results can help in the initial diagnosis of Covid-19, especially when there is a shortage of RT-PCR kits, specialized laboratories and to screen large number of patients. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.