Mathematical model to predict the disease course of Multiple Sclerosis (MS)

Abstract

Multiple Sclerosis (MS) is an auto-immune disease of the brain and the spinal cord in which the immune system of an individual attack the protective sheath covering their neurons. MS is characterized by weakness, numbness, blurred vision, bladder dysfunction and lack of muscle coordination. Various forms of the disease exist and are categorized by how they progress from the onset. However, various test and examinations are available to monitor the course of the MS. The most widely used score by the neurologist is the expanded disability status scale. Also, evoked potentials such as visual evoked potentials(VEPs), somatosensory evoked potentials(SEPs) and motor evoked potentials(MEPs) have been used. It is of interest here to predict the course of MS by studying the EDSS score as a function of MEPs and some other clinical variables such as age, gender, and type of multiple sclerosis. The MEPs are projected firstly into the wavelet domain to produce a sparse representation of the data. Then a penalized regression model, wavelet-based logistic LASSO regression is fitted to these variables to predict the course of MS. It was observed that the MEPs predicted the EDSS score with 72% accuracy. Adding other patient characteristics improved the accuracy of the prediction by about 6%. Moreover, the results also show that only early stages of the disease were well predicted by the MEPs.