Related Experiment Video
Updated: Nov 16, 2025
10:46
A Method of Trigonometric Modelling of Seasonal Variation Demonstrated with Multiple Sclerosis Relapse Data
Published on: December 9, 2015
10.8K
Leveraging electronic health records data to predict multiple sclerosis disease activity
Yuri Ahuja1, Nicole Kim1, Liang Liang1
1Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, MA, USA.
Annals of Clinical and Translational Neurology
|February 24, 2021
Summary
A new tool uses electronic health records to predict multiple sclerosis (MS) relapse risk. This machine learning algorithm accurately forecasts future relapses, aiding treatment decisions for MS patients.
Area of Science:
- Neurology
- Medical Informatics
- Machine Learning
Background:
- Multiple Sclerosis (MS) lacks a validated relapse risk prediction tool for guiding treatment.
- Electronic Health Records (EHR) offer readily accessible data for clinical decision-making.
Purpose of the Study:
- To develop and validate a clinical tool for predicting MS relapse risk using EHR data.
- To create a practical, point-of-care tool for MS management.
Main Methods:
- Developed a two-stage machine learning algorithm using EHR data from 2006-2016.
- Utilized L1-regularized logistic regression (LASSO) for relapse phenotyping and risk prediction.
- Validated the model on an independent cohort of 186 MS patients.
Main Results:
- The final model, including age, disease duration, and imputed relapse history, achieved an AUC of 0.707.
- Model performance was superior to baseline and comparable to using actual relapse history.
- Predicted relapse risk decreased with longer disease duration and older age.
Conclusions:
- A novel EHR-based machine learning algorithm accurately predicts 1-year MS relapse risk.
- The tool demonstrates applicability at the point of care for MS treatment guidance.
- This two-stage EHR prediction approach may extend to other neurological conditions.
