Related Experiment Video
Updated: Jun 3, 2025
10:46
A Method of Trigonometric Modelling of Seasonal Variation Demonstrated with Multiple Sclerosis Relapse Data
Published on: December 9, 2015
10.6K
Machine learning-based prediction of disease progression in primary progressive multiple sclerosis
Michael Gurevich1,2, Rina Zilkha-Falb1, Jia Sherman3
1Multiple Sclerosis Center, Sheba Medical Center, Ramat-Gan 5262, Israel.
Brain Communications
|January 9, 2025
Summary
Peripheral blood cell transcriptome analysis can predict disease progression in primary progressive multiple sclerosis (PPMS) patients. Machine learning models using gene expression data accurately forecast disability and brain volume loss, aiding early intervention.
Area of Science:
- Neuroscience
- Genomics
- Medical Imaging
Background:
- Primary progressive multiple sclerosis (PPMS) exhibits variable disability progression, necessitating predictive biomarkers.
- Peripheral blood transcriptome analysis offers a non-invasive approach to assess disease activity and patient outcomes.
Purpose of the Study:
- To develop machine learning models for predicting disability progression and brain volume loss in PPMS patients.
- To identify key blood transcriptional profiles associated with rapid disease advancement.
- To enable early identification of high-risk PPMS patients for timely therapeutic intervention.
Main Methods:
- RNA-sequence analysis of peripheral blood mononuclear cells from PPMS patients in the ORATORIO trial.
- Application of machine learning and cross-validation algorithms to predict clinical and radiological outcomes over 120 weeks.
- Utilized baseline blood transcriptional profiles and brain MRI data for prognostic model development.
Main Results:
- A 10-gene predictor achieved 70.9% accuracy in identifying patients with disability progression within 120 weeks.
- A 4-gene classifier distinguished fast vs. slow disability progression with 74.1% accuracy.
- A 12-gene classifier predicted brain volume loss with 70.2% accuracy, improving to 74.1% when combined with MRI data.
Conclusions:
- Blood transcriptome data is a valuable tool for predicting disability progression and brain volume loss in PPMS.
- Machine learning models integrating gene expression and MRI data enhance prognostic accuracy.
- These findings support the use of peripheral blood biomarkers for personalized treatment strategies in PPMS.
