Machine Learning Algorithms Introduce Evoked Potentials As Alternative Biomarkers for the Expanded Disability Status...

Area of Science:

• Medical Artificial Intelligence
• Neurology
• Biomarker Discovery

Background:

• Machine learning (ML) is increasingly vital in healthcare for predictive modeling and operational efficiency.
• Early detection of diseases like multiple sclerosis (MS) is crucial for effective management and treatment.
• Evoked potentials (EPs) correlate with Expanded Disability Status Scale (EDSS) in MS, indicating prognostic potential.

Purpose of the Study:

• To apply artificial intelligence (AI) techniques to identify predictors of MS disability progression (EDSS).
• To clarify the role of EPs in the prognostication of MS.
• To develop an expert AI system for predicting EDSS using knowledge-mining algorithms.

Main Methods:

• Analysis of empirical data from a medical database (125 records).
• Development of intelligent systems using ML algorithms: decision trees and neural networks.
• Experimental evaluation comparing EPs-based predictions with MRI data.

Main Results:

• ML models for EPs achieved high accuracies: Decision Trees (88.9%), Neural Networks (92.9%), Bayes (88.2%).
• EPs-based prediction accuracies were comparable to MRI: MRI (88.2%, 96.0%, 85.0%).
• EPs demonstrate efficacy analogous to MRI findings in predicting MS progression.

Conclusions:

• EPs can be established as effective predictors for MS, with accuracy comparable to MRI.
• Further validation is needed to establish EPs as equally effective biomarkers as imaging or biochemical methods.
• EPs offer advantages of lower cost, portability, and simpler administration compared to MRI.