Classification of Rheumatoid Arthritis using Machine Learning Algorithms

Rheumatoid Arthritis (RA) is a persistent provocative ailment that effects and decimates the joints of wrist, finger, and feet. If left untreated, one can lose their ability to lead a normal life. RA is the most typical fiery joint inflammation, influencing around 1-2 of the total populace. Throughout the years, soft computing played an important part in helping ailment analysis in doctor's decision process. The main aim of this study is to investigate the possibility of applying machine learning techniques to the analysis of RA characteristics. As a preliminary work, a credible database has been identified to be used for this research. The database has outputs of array temperature values from thermal imaging for the joints of hand. Furthermore, this database which consists of 8 attributes and 32 instances, are used to determine the performance in terms of accuracy for the classification of different algorithms. In this preliminary work, ensemble algorithms such as bagging, AdaBoost and random subspace with base classifier such as random forest and SVM were trained and tested using the assessment criteria such as accuracy, precision, sensitivity and AUC using Weka tool. From the preliminary finding of this paper, it can be concluded that with base classifier SVM, bagging has better classification accuracy over the others and with base classifier random forest Adaboost slightly outperformed other models for rheumatoid arthritis dataset. © 2019 IEEE.