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

Abstract

BackgroundThe aim of this study was to assess the diagnostic accuracy (sensitivity and specificity) of clinical, imaging and motor evoked potentials (MEP) for predicting the short-term prognosis of multiple sclerosis (MS).MethodsWe obtained clinical data, MRI and MEP from a prospective cohort of 51 patients and 20 matched controls followed for two years. Clinical end-points recorded were: 1) expanded disability status scale (EDSS), 2) disability progression, and 3) new relapses. We constructed computational classifiers (Bayesian, random decision-trees, simple logistic-linear regression-and neural networks) and calculated their accuracy by means of a 10-fold cross-validation method. We also validated our findings with a second cohort of 96 MS patients from a second center.ResultsWe found that disability at baseline, grey matter volume and MEP were the variables that better correlated with clinical end-points, although their diagnostic accuracy was low. However, classifiers combining the most informative variables, namely baseline disability (EDSS), MRI lesion load and central motor conduction time (CMCT), were much more accurate in predicting future disability. Using the most informative variables (especially EDSS and CMCT) we developed a neural network (NNet) that attained a good performance for predicting the EDSS change. The predictive ability of the neural network was validated in an independent cohort obtaining similar accuracy (80%) for predicting the change in the EDSS two years later.ConclusionsThe usefulness of clinical variables for predicting the course of MS on an individual basis is limited, despite being associated with the disease course. By training a NNet with the most informative variables we achieved a good accuracy for predicting short-term disability.

Central Motor Pathways in Children With Multiple Sclerosis

“…В педиатрической практике более надежным параме-тром считается ВЦМП [3]. Показатель ВЦМП весьма чувствителен к демиелинизации моторных путей, равно как и к аксональным нарушениям [8].…”

“…Снижение амплитуды ВМО отражает аксональные нарушения. Асимметрия ВМО у взрослых между сторо-нами более 50% расценивается как патологическая [8,13]. Часть авторов считают, что асимметрия амплитуд ВМО на первых этапах РС является наиболее чувствительным показателем и первым параметром, изменение которого возможно обнаружить нейрофизиологическими мето-дами [13].…”

Evoked potentials are of little use in the diagnosis or monitoring of MS: No

“…MEPs correlate very well with clinical disability expressed by the EDSS. 3 As in the other EP modalities, the pathological substrates of MS account for all the abnormalities detected by MEPs; however, and more especially to detect axonal loss, which may result in loss of amplitude and other abnormalities, MEPs should be done in a well-standardized way in order to provide accurate results.…”

“…Oscar Fernández 1 and Victoria Fernández 2 Multimodal EPs have been tested in prospective and retrospective studies in MS, [3][4][5] showing a very good correlation between EP measures and scales with EDSS and evolution time. Multimodal EPs were used prospectively in a homogeneous population of relapsing-remitting MS patients with early disease and found to be a robust biomarker of disability development.…”

Gait Features and their Relationships to Physiological Attributes of Multiple Sclerosis

“…Ramani and Sivagami [64] conduct a survey of data mining techniques for the classification of persons with Parkinson's disease from healthy subjects. Bejarano et al [65] assess the diagnostic accuracy of clinical variables, including expanded disability status scale (EDSS), using various computational classifiers and employ data mining for feature selections, for predicting the MS course. Li and Schwartz [12] use a partitioning tree analysis technique to identify response shift patterns in persons with MS. Other applications of data mining in bioinformatics include probe selection for gene-expression arrays [66], plant genotype discrimination [67], experiments with automatic cancer diagnosis [68], protein annotation [69], and identification and prediction of drug-induced nausea [70].…”