Integrating Reproductive and Clinical Variables to Predict Postpartum Disability Outcomes in Multiple Sclerosis Using...

Area of Science:

• Neurology
• Immunology
• Data Science

Background:

• Pregnancy alters the immune system in women with multiple sclerosis (MS).
• Postpartum disease reactivation is a significant concern for MS patients.
• Long-term effects of pregnancy on MS disability progression require further investigation.

Purpose of the Study:

• To predict postpartum disability changes (EDSS-based) in women with MS.
• Utilize machine learning models incorporating clinical and demographic pregnancy variables.
• Identify key predictors of postpartum disability progression in MS.

Main Methods:

• Retrospective study of 662 women and 909 pregnancies.
• Engineered features: EDSS scores, disease duration, age, postpartum relapse, obstetric factors.
• Employed Random Forest, XGBoost, Elastic Net, and Support Vector Classifier models with cross-validation.

Main Results:

• Classification models demonstrated high accuracy (85-88%) and F1 scores (0.84-0.87).
• Postpartum relapse, disease duration, and maternal age were primary predictors.
• Multiple pregnancies and obstetric factors (delivery type, breastfeeding) provided additional prognostic value.

Conclusions:

• Machine learning models effectively predict postpartum disability changes in MS.
• Postpartum relapse is the dominant factor, with reproductive and obstetric variables offering supplementary insights.
• The postpartum period is crucial for MS management, necessitating individualized prognostic frameworks.