SHAP Interpretation of Machine Learning Model for Subcortical White Matter Biomarkers in Cognitive Impairment in Multiple Sclerosis Patients

Motivation: Multiple Sclerosis patients present cognitive decline at the early stages of the disease. Current neurocognitive batteries may not identify early changes. FA evaluates microstructural changes in white matter. To consider clinicopathological correlation remains complex and needs to be understood. Goal(s): A biomarker that could detect patients with cognitive deficits might benefit from early diagnosis and treatment. Approach: ML to identify subcortical white matter biomarkers between Healthy Controls with Cognitive Preserved and Relapsing-Remitting Multiple Sclerosis patients with or without cognitive impairment in verbal episodic memory. Results: We found six FA biomarkers, all located in the frontal lobes. These features maximized the accuracy, obtained: 62.22±17.33%. Impact: Since the MRI is the gold standard for MS diagnosis, we can obtain new insights about not only the patient's condition but also detect early changes in patients with cognitive impairment.