Related Experiment Video
Updated: May 6, 2026
The Multiple Sclerosis Performance Test MSPT: An iPad-Based Disability Assessment Tool
Published on: June 30, 2014
Integrating Reproductive and Clinical Variables to Predict Postpartum Disability Outcomes in Multiple Sclerosis Using
Murat Emec1, Said Alizada2, Yasemin Simsek3
1Istanbul Nisantasi University, Department of Aviation Electrical and Electronics, Sarıyer, Istanbul, 34398, Turkey.
Background:
Pregnancy represents a unique immunological state in women with multiple sclerosis (MS), and postpartum disease reactivation is a major concern. While pregnancy outcomes have been extensively described, the long-term effects of reproductive and obstetric variables on disability progression remain poorly elucidated.
Objective:
To predict postpartum EDSS-based disability change based on pregnancy-related clinical and demographic variables in women with MS, using validated machine learning models.
Methods:
This retrospective real-world study included 662 women contributing 909 pregnancies. Engineered features included pre- and post-pregnancy Expanded Disability Status Scale (EDSS) scores, disease duration, maternal age, postpartum relapse, and obstetric variables. Regression and classification models (Random Forest, XGBoost, Elastic Net, Support Vector Classifier) were trained on an 80/20 train-test split with five-fold cross-validation. Model performance was assessed using R², mean absolute error (MAE), accuracy, and F1 score.
Results:
Classification models achieved superior generalization performance (test accuracy 85-88%, F1 0.84-0.87) compared to regression models (test R² 0.31-0.39, MAE 0.41-0.48). Postpartum relapse was the strongest predictor of disability change, followed by disease duration and age at pregnancy. Predictive performance was highest among women with multiple pregnancies, suggesting that cumulative reproductive history carries prognostic value. Obstetric variables such as delivery type and breastfeeding contributed secondary but clinically relevant effects.
Conclusion:
Machine learning models integrating pregnancy-related variables can provide clinically informative predictions regarding postpartum EDSS-based disability change in women with MS. Postpartum relapse remains the dominant driver of disability change, while reproductive and obstetric factors provide additional prognostic information. These findings highlight the postpartum period as a critical therapeutic window and support incorporating reproductive variables into individualized prognostic frameworks for women with MS.
