Non-linear Measures of Gait Adaptability in Multiple Sclerosis - Sensitivity and Neurological Correlates

ABSTRACT Multiple sclerosis (MS) is the leading cause of non-traumatic disability among young and middle-aged adults. People with MS (pwMS) rate walking and mobility as their highest priority for both research and symptom management. Given the importance of early initiation of disease-modifying therapeutics (DMTs) to minimize long-term disability, tools to identify early disease activity are needed. Traditional measures of disability, the Expanded Disability Status Scale (EDSS) and gait speed tests demonstrate poor reliability and responsiveness in cases with minimal disability. Nonlinear measures of gait, Local dynamic stability (LDS), Complexity (sample entropy) and Regularity (autocorrelation), measured in laboratory settings is sensitive to subclinical gait deterioration in people with MS (pwMS). These measures have not been tested in a clinical setting using wearable sensors. Methods Gait metrics were calculated in MatLab from inertial data collected from 59 pwMS (EDSS 0-4) and 23 age- and sex-matched healthy controls (HC) during a 5-minute walk. We aimed to provide known-groups validation of non-linear gait measures and compare sensitivity of LDS from sensors placed at sternum and sacrum in pwMS (ROC analysis). Associations of gait metrics with disability, Kurtzke Functional System scores and 3T MRI segmental brain volumes were examined. Results Most sternum-derived LDS measures detected significant differences between HC and pwMS (EDSS0-1) with moderate to large effect sizes (η 2 =.100-305), while the effect sizes for sacrum-derived LDS were lower (η 2 =.104-.166). Sternum 3D best distinuished EDSS 0-1 from HC whereas the effect size was lower for gait speed (η 2 =.104). Sternum Instability-3D showed strongest correlation with pyramidal dysfunction (r s =.455, p<.001). Sensory dysfunction correlated significantly with Regularity in the vertical plane from both sensors, while cerebellar dysfunction was significantly associatesd with sacrum Regularity in the saggital plance (r=-.343, p=.008) and brainstem dysfunction with Complexity in the frontal plane (r=-.343, p=.008). Conclusion Sternum-derived LDS measures were more sensitive than Sacrum-derived measures. Correlations with clinical and morphological brain measures support the validity of walking deterioration as reflective of neurodegeneration in subcortical grey matter. The current findings of high sensitivity in non-disabled cases, as well as the clinical feasibility and relatively low costs, support the utility of these measures as a supplementary clinical assessment tool.