The Applications of Machine Learning Algorithms in Multiple Sclerosis: A Systematic Review

Multiple Sclerosis (MS) is a common chronic disease that affects society, especially young people. In recent years, data sciences have been used extensively to deal with the disease. Machine learning is one of the main data sciences types which has been used to deal with chronic diseases such as MS. This study aimed to identify the applications of machine learning algorithms in MS disease. This study is a systematic review that conducted in 2020. The searches were done in PubMed, Scopus, ISI Web of Sciences, Ovid, Science Direct, Embase, and Proquest scientific databases, by combining related keywords. Data extraction was done by using a data extraction form to follow the trends of this field of study. The results of the study showed that diagnosis of MS was the main application of machine learning in MS (33.3 %); also, assessment (24.24%) and prediction (18.18 %) of the disease were other main applications. The most used data type was medical images such as MRI and CT scans (55.17 %). The most used machine learning algorithm type was Support Vector Machine (SVM) (30 %) as a classification algorithm. The most optimized algorithm for the diagnosis and prediction of MS was KNN. It’s suggested to use machine learning algorithms to diagnose, assess, predict lesion classification, treatment, and severity determining of MS disease. Although the most common form of data used for MS is medical images, it is suggested that other types of data are generated to be used in machine learning algorithms. Considering the optimization rate of the algorithms used, it is suggested to pay more attention to the type of data and study objectives in data analysis using machine learning.