Utilizing Aerobic Capacity Data for EDSS Score Estimation in Multiple Sclerosis: A Machine Learning Approach

Abstract

The Expanded Disability Status Scale (EDSS) is the most popular method to assess disease progression and treatment effectiveness in patients with multiple sclerosis (PwMS). One of the main problems with the EDSS method is that different results can be determined by different physicians for the same patient. In this case, it is necessary to produce autonomous solutions that will increase the reliability of the EDSS, which has a decision-making role. This study proposes a machine learning approach to predict EDSS scores using aerobic capacity data from PwMS. The primary goal is to reduce potential complications resulting from incorrect scoring procedures. Cardiovascular and aerobic capacity parameters of individuals, including aerobic capacity, ventilation, respiratory frequency, heart rate, average oxygen density, load, and energy expenditure, were evaluated. These parameters were given as input to CatBoost, gradient boosting (GBM), extreme gradient boosting (XGBoost), and decision tree (DT) machine learning methods. The most significant EDSS results were determined with the XGBoost algorithm. Mean absolute error, root mean square error, mean square error, mean absolute percent error, and R square values were obtained as 0.26, 0.4, 0.26, 16, and 0.68, respectively. The XGBoost based machine learning technique was shown to be effective in predicting EDSS based on aerobic capacity and cardiovascular data in PwMS.

Keywords

• aerobic capacity
• Expanded Disability Status Scale
• gradient boosting
• machine learning
• multiple sclerosis