Applications of Datamining Techniques for Predicting the Post - Covid 19 Symptoms in Saudi Arabia, Jazan

Abstract

Background The entire world was combating COVID-19; however, a significant proportion of patients demonstrate the persistence of some COVID-19 symptoms, new symptom development, or exaggeration of pre-existing disease after a negative viral load. They are referred to as a post-COVID-19 syndrome. According to various researches, COVID-19 has a wide range of long-term effects on virtually all systems, including the respiratory, cardiovascular, gastrointestinal, neurological, mental, and dermatological systems. Finding the various symptoms of post-acute and chronic is critical since they might have a significant impact on the patients' everyday functioning. As a result, we aimed to distinguish the symptoms immediately after the initial phase in which the symptoms affected them for more than three weeks using data mining techniques. Methodology: Post-COVID conditions do not affect everyone the same way. They can cause various types and combinations of symptoms in different people. The purpose of this research is to analyse the complications of post covid-19 syndrome. The purpose of Data mining is for discovering the knowledge from vast amount of database. To classify the symptoms of post covid-19, data mining techniques is used. In this study, ranking method was used in preprocessing to select subset of attributes for strengthening the rate of accuracy of classifiers. The data were collected through Google form of 384 household of students from Public Health College in Jazan University. The WEKA open-source software is used for this research work under Windows7 environment. An experimental study is carried out using data mining technique such as J48 and Random Forest tree. The data records are classified as six categories such as General symptoms, Nervous symptoms, Respiratory symptoms, Heart symptoms, Digestive symptoms and normal. Result: The performances of classifiers are evaluated through the confusion matrix in terms of accuracy, time taken to build the Model and error rate. It has been concluded that Random Forest Tree gives better accuracy, minimum time taken to build the model and less error rate than the J48 classifier.