Machine learning classification of multiple sclerosis patients based on raw data from an instrumented walkway

Abstract

Background Using embedded sensors, instrumented walkways provide clinicians with important information regarding gait disturbances. However, because raw data are summarized into standard gait variables, there may be some salient features and patterns that are ignored. Multiple sclerosis (MS) is an inflammatory neurodegenerative disease which predominantly impacts young to middle-aged adults. People with MS may experience varying degrees of gait impairments, making it a reasonable model to test contemporary machine leaning algorithms. In this study, we employ machine learning techniques applied to raw walkway data to discern MS patients from healthy controls. We achieve this goal by constructing a range of new features which supplement standard parameters to improve machine learning model performance. Results Eleven variables from the standard gait feature set achieved the highest accuracy of 81%, precision of 95%, recall of 81%, and F1-score of 87%, using support vector machine (SVM). The inclusion of the novel features (toe direction, hull area, base of support area, foot length, foot width and foot area) increased classification accuracy by 7%, recall by 9%, and F1-score by 6%. Conclusions The use of an instrumented walkway can generate rich data that is generally unseen by clinicians and researchers. Machine learning applied to standard gait variables can discern MS patients from healthy controls with excellent accuracy. Noteworthy, classifications are made stronger by including novel gait features (toe direction, hull area, base of support area, foot length and foot area).

Machine learning corroborates subjective ratings of walking and balance difficulty in multiple sclerosis

“…After a rest, the participants performed a dual task walking test at their self-selected speed for four times. During this dual task walking, the participant was asked to walk while subtracting 7 from a preceding number beginning with 100, and speaking aloud the results (Kirkland et al, 2015 ; Chen et al, 2020 ; Hu et al, 2022 ).…”

“…These included 21 features, foot type/length/width/area, unsigned toe angle, step/stride length, step/stride width, base width, step/stride time, step/stride velocity, single/double support time, stance time, toe direction, hull area, base of support (BOS) area, line of progression (LOP) deviation angle. The details regarding each parameter and how they were extracted from the walkway sensor data were described previously (Hu et al, 2022 ).…”

“…The remaining data were scaled to exhibit zero mean and unit variance. Feature selection were completed using ANOVA-SVM based on previous work (Hu et al, 2022 ). However, in this paper, we used the adaptive synthetic sampling approach (ADASYN) for data balancing and managing different hyperparameters for each model.…”

“…For example, McGinnis et al estimated gait speed of MS patients from an array of accelerometers attached to different locations on the body (McGinnis et al, 2017 ). Our previous published work (Hu et al, 2022 ) applied machine learning to successfully detect MS patients from healthy volunteers using novel features combined with standard features derived from the raw walkway sensor data. Also based on instrumented walkway raw data, Trentzsch et al explored different machine learning algorithms (Gaussian Naive Bayes, Decision Tree, k-Nearest Neighbor, and Support Vector Machines) to distinguish mild and moderate MS levels defined by Expanded Disability Status Scale (EDSS) (Trentzsch et al, 2021 ).…”

“…We grouped patients by their self-reported walking and balance problems using questions in the MS Impact Scale−29 (Phillips et al, 2014 ) and applied machine learning to gait features derived from the walkway sensors (Hu et al, 2022 ). It was hypothesized that machine learning models could effectively distinguish the targets.…”