A Comparative Study of Classification Approaches for COVID-19 Prediction

The novel coronavirus (COVID-19), a highly infectious disease that first found at Wuhan Province of China in Dec 2019, spread worldwide in some months and already become a pandemic. Covid-19 has already changed the world economic structure, people's religious, political, social life, public health structure, people's daily life structure and also made millions of people jobless. The only way to fight this epidemic is to identify the infected person as soon as possible and separate them from a healthy person, so that they can't infect anyone again. At present, RT-PCR is currently used to detect coronavirus patients around the world. But the World Health Organization (WHO) said that RT-PCR suffers from low sensitivity and low specificity for early-stage cases. Recent research has shown that chest CT scan images play a beneficial role in identifying coronavirus cases. In this study, we compared the performances of four classification algorithms, such as Random Forest (RF), Support Vector Machine (SVM), Extra Trees (ET), and Convolutional Neural Network (CNN) for classifying COVID-19 cases and proposed a prediction model based on classification results. The result shows that our proposed CNN model outperformed the other classification algorithms and obtained an accuracy of 98.0%.