Artificial neural network-based prediction of multiple sclerosis using blood-based metabolomics data.

Area of Science:

• Neurology
• Biochemistry
• Computational Biology

Background:

• Multiple sclerosis (MS) diagnosis and management are challenging, often leading to delayed treatment due to late detection.
• Metabolomics data offers a promising avenue for applying artificial intelligence (AI) in predicting and understanding MS.
• Current diagnostic and management strategies for MS require improvement, particularly in early detection.

Purpose of the Study:

• To develop and validate an AI model using machine learning (ML) for analyzing blood-based metabolomics data to predict MS presence and severity.
• To identify unique alterations in biochemical metabolites and their correlation with MS disease severity parameters.
• To enhance the efficiency of using metabolic profiles for early MS detection and management.

Main Methods:

• A machine learning (ML) approach, specifically an artificial neural network (ANN) with perceptrons, was employed.
• Blood metabolomics datasets from MS patients and healthy controls (HC) were utilized.
• Data was partitioned into training, validation, and testing sets to rigorously assess model performance.

Main Results:

• The developed AI model achieved an accuracy of 87%, sensitivity of 82.5%, specificity of 89%, and precision of 77.3%.
• Unique alterations in biochemical metabolites were identified and correlated with MS disease severity.
• The model demonstrated robustness, generalizability, and capacity to handle large datasets.

Conclusions:

• The AI model shows significant potential for assisting neurologists in the accurate diagnosis and improved management of multiple sclerosis.
• Further validation through large-scale, multicenter cohort studies is essential for clinical integration.
• AI-driven analysis of metabolomics data represents a promising frontier for advancing MS care.