Prognostic models in multiple sclerosis: progress and challenges in clinical integration.

Area of Science:

• Neurology
• Inflammatory Diseases
• Biostatistics

Background:

• Multiple sclerosis (MS) is a chronic central nervous system inflammatory disease with significant health and economic impact.
• Current clinical practice lacks validated prognostic models to predict the heterogeneous disease course of MS.
• Existing prognostic models, developed using various statistical and AI methods, show limited clinical utility due to poor performance, transferability, and validation.

Discussion:

• A comprehensive Cochrane review systematically assessed 75 published MS prediction models.
• Evaluation utilized established checklists (CHARMS, PROBAST, TRIPOD) to gauge model quality and applicability.
• Findings highlight critical deficiencies in current prognostic models for MS.

Key Insights:

• No existing prognostic model for multiple sclerosis is reliably used in routine clinical care.
• Key barriers include inadequate model performance, poor transferability across patient populations, and insufficient validation.
• Significant improvements are needed for models to support clinical decision-making in MS therapy.

Outlook:

• Future research must focus on developing and validating robust prognostic models for MS.
• Enhanced model performance and transferability are crucial for successful clinical implementation.
• The goal is to enable informed therapy decisions and improve patient outcomes in multiple sclerosis management.