COVIDetect-DESVM: Explainable framework using Differential Evolution Algorithm with SVM classifier for the diagnosis of COVID-19

Abstract

The SARS-CoV-2 (COVID-19) epidemic has a huge impact on the health and daily life of people. More than 200 countries are impacted due to this pandemic. To light the COVID-19 virus we need a powerful monitoring system to identify the patients and isolate them. The current detection tests are either done by measuring the body temperature or spotting the genetic material of the SARS-CoV-2. These techniques are time-consuming and have a poor detection rate. Radiological images like chest X-rays are also highlighted and help in the diagnosis of COVID-19 patients. Initial studies suggest that COVID-19 patients have abnormalities in their chest X-rays and can be used in the diagnosis of COVID-19. Based on this literature research, various solutions have been proposed utilizing chest X-rays to detect the SARS-CoV-2. Most of these solutions use non-public datasets and complicated structures with fewer accurate results. In our study, we propose a self-learning, interpretable model for real-time detection of COVID-19. This model utilizes a Differential evolution algorithm for feature selection and Support Vector Machine (SVM) as a classifier. The aim is to obtain higher accuracy in detecting COVID-19 infected patients using X-ray images. We have also used the LIME explanation algorithm to explain the predictability of our model and this makes our design very robust and sustainable. This fully transparent, Interpretable, and explainable model can be used in hospitals where there is a huge demand for rapid tests and radiologists are busy.