Integrative Machine Learning Approaches for Differentiating Pediatric Multiple Sclerosis from Non-Inflammatory Disorders via Optical Coherence Tomography
By
Chaojun Chen
Sahar Soltanieh
Sajith Rajapaksa
Farzad Khalvati
E. Ann Yeh
April 21, 2026
Clinical Scorecard: Integrative Machine Learning Approaches for Differentiating Pediatric Multiple Sclerosis from Non-Inflammatory Disorders via Optical Coherence Tomography
At a Glance
Category Detail
Condition Pediatric Multiple Sclerosis (POMS) Key Mechanisms Utilizes Optical Coherence Tomography (OCT) for high-resolution retinal imaging to assess structural integrity of retinal layers. Target Population Children and youth with suspected POMS and non-inflammatory neurological conditions. Care Setting Neuroinflammatory Registry at the Hospital for Sick Children, Toronto, Canada.
Key Highlights
Early fusion model achieved highest performance (AUC: 0.90, accuracy: 87%). POMS accounts for 3-5% of all MS cases, emphasizing the need for early diagnosis. OCT-derived features correlate with disease severity and visual dysfunction. Multimodal ML approaches enhance diagnostic performance by integrating diverse data sources. Study includes children with non-inflammatory neurological conditions as a comparator group. Guideline-Based Recommendations
Diagnosis
Utilize imaging modalities such as MRI, OCT, and VEP to establish dissemination in space. Management
Timely initiation of disease-modifying therapies is critical for improving long-term outcomes. Monitoring & Follow-up
OCT can be used to monitor disease progression and visual dysfunction. Risks
Diagnostic ambiguity due to overlapping symptoms with non-inflammatory neurological conditions. Patient & Prescribing Data
Children with suspected POMS and non-inflammatory neurological disorders.
Early therapeutic intervention can significantly improve outcomes.
Clinical Best Practices
Incorporate multimodal ML approaches for enhanced diagnostic accuracy. Focus on the anterior visual pathway as a key diagnostic target for MS. Utilize OCT as a non-invasive tool for assessing retinal structural integrity. References
