Computational classifiers for predicting the short-term course of Multiple sclerosis

2011

Predicting Short-Term Outcomes in Multiple Sclerosis Using Computational Classifiers

Sample size: 51 publication 10 minutes Evidence: moderate

Author Information

Author(s): Bejarano Bartolome, Bianco Mariangela, Gonzalez-Moron Dolores, Sepulcre Jorge, Goñi Joaquin, Arcocha Juan, Soto Oscar, Carro Ubaldo Del, Comi Giancarlo, Leocani Letizia, Villoslada Pablo

Primary Institution: Department of Neuroscience, CIMA-University of Navarra, Pamplona, Spain

Hypothesis

Can clinical, imaging, and neurophysiological variables accurately predict the short-term prognosis of multiple sclerosis?

Conclusion

Using a neural network with key clinical variables, we achieved good accuracy in predicting short-term disability in multiple sclerosis patients.

Supporting Evidence

Disability at baseline, grey matter volume, and MEP were the best predictors of clinical outcomes.
Neural networks showed good performance in predicting EDSS change two years later.
Combining clinical and imaging data improved predictive accuracy.

Takeaway

Doctors can use special computer programs to help predict how multiple sclerosis will affect patients in the short term, which can help them make better treatment decisions.

Methodology

The study involved a prospective cohort of 51 MS patients and 20 matched controls, using clinical data, MRI, and MEP metrics to develop computational classifiers.