Development of a machine learning model to predict the expanded disability status scale in multiple sclerosis...

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Informatics

Background:

• Accurate assessment of disability in multiple sclerosis (MS) is vital for treatment and prognosis.
• The Expanded Disability Status Scale (EDSS) is a standard measure, but prediction is challenging due to disease heterogeneity.
• Machine learning (ML) offers a novel approach to predict EDSS scores using patient data.

Purpose of the Study:

• To evaluate the feasibility of using ML techniques for predicting EDSS scores in MS patients.
• To compare the performance of different ML models in EDSS score prediction.
• To identify the most effective ML model for enhancing clinical decision-making in MS.

Main Methods:

• A dataset of 231 MS patients with physical, psychosocial, and cognitive assessments at three time points was utilized.
• Feature selection was performed based on saliency and correlation analysis from 126 initial features.
• XGBoost, Random Forest, and Linear Regression models were trained and hyperparameter-tuned, with performance evaluated using MAE, MSE, and R².

Main Results:

• The XGBoost ML model achieved the highest performance in predicting EDSS scores at T2.
• XGBoost yielded an MAE of 0.2361, MSE of 0.2408, and R² of 0.9705.
• These results demonstrate the strong predictive capability of the XGBoost model for EDSS scores in this dataset.

Conclusions:

• ML techniques are feasible for predicting EDSS scores in MS patients.
• The developed ML models, particularly XGBoost, show promising performance.
• These predictive models have the potential to improve clinical decision-making and patient management in MS care.