Genetics

This blog post examines a large multicentric study that applies machine learning techniques to predict disease severity and short-term outcomes in multiple sclerosis by integrating clinical assessments, neuroimaging, and molecular profiling. Using Random Forest models trained on longitudinal data from well-characterized patient cohorts, the study demonstrates that routinely collected clinical and imaging variables can achieve moderate to high predictive accuracy for disability progression, disease activity, and treatment escalation, while omics data provide only incremental benefit in selected contexts. Together, these findings highlight both the promise and current limitations of multimodal machine learning approaches in advancing precision medicine for multiple sclerosis.