Paper Title
COVID-19 Prediction Using an Efficient Symptom Question Mining Method

Abstract
Abstract - Questionnaires on COVID-19 pandemic are becoming increasingly important for measuring COVID-19 cases that was designed to measure the symptoms to which individuals appraise situations in their lives as COVID-19 diagnosis. To avoid as much as several waves of infection cases and hospitalization cases identified, efficient tools to facilitate the diagnosis of COVID-19 are needed. In this study, we propose an efficient method for designing predictors of COVID-19 cases using a small set of symptoms question items obtained by using novel machine learning models from existing dataset which is composed90,839 adultsCovid-19 negative and 8,393 adultsCovid-19 positive who responses to the nationwide data publicly reported by the Israeli Ministry of Health. For the independent-adult prediction, the training dataset contains 51,831 individuals (of whom 4769Covid-19 positive)and 47,401 tested individuals (of whom 3624Covid-19 positive) serves as the test dataset. A predictor, PreCOVID19, composed of 8 features selected using orthogonal experimental design bases on an orthogonal array is created for predicting COVID-19 cases with 95.81 % training accuracy and the test accuracy is 95.87%. The web server is available at http://changcw.pythonanywhere.com. Keywords - Covid-19 Questionnaire, Symptoms Questions, Orthogonal Experimental Design, Prediction, XGBoot.