Computational Intelligence Technique for Prediction of Multiple Sclerosis Based on Serum Cytokines

Area of Science:

• Neuroimmunology
• Computational Biology

Background:

• Multiple sclerosis (MS) is a neurodegenerative CNS disease driven by inflammation and demyelination.
• Immune reactivity, involving cytokines, Th cells, and auto-reactive lymphocytes, is crucial in MS pathogenesis.
• Previous research links specific cytokines to MS development.

Purpose of the Study:

• To identify serum cytokine predictors for multiple sclerosis (MS) in US and Russian cohorts.
• To develop a machine learning model for classifying MS into remitting and non-remitting forms.
• To evaluate the diagnostic and prognostic potential of serum cytokines in MS.

Main Methods:

• Serum levels of eight cytokines (IL-1β, IL-2, IL-4, IL-8, IL-10, IL-13, IFN-γ, TNF-α) were analyzed.
• Machine learning models (SVM, decision tree, random forest, neural networks) were trained and tested on cytokine and clinical data.
• The random forest model was selected for its superior performance in diagnosis and classification.

Main Results:

• The random forest model demonstrated high accuracy, specificity, AUC, and sensitivity for MS diagnosis.
• Serum cytokine concentrations were identified as significant predictors for MS.
• The random forest model effectively predicted remitting versus non-remitting MS subtypes.

Conclusions:

• Serum cytokine levels serve as valuable prognostic markers for predicting MS.
• Machine learning models, especially random forest, offer robust tools for MS diagnosis and subtyping.
• This study highlights the potential of cytokine profiling for personalized MS management.