Royal Hallamshire Hospital
Neurology

Author Of 1 Presentation

Biomarkers and Bioinformatics Poster Presentation

P0035 - Can gait modelling predict disease progression in MS? A study using small body worn sensors in a clincial setting.  (ID 1377)

Speakers
Presentation Number
P0035
Presentation Topic
Biomarkers and Bioinformatics

Abstract

Background

Accurate assessment of mobility is critical for the clinical management of people with MS (pwMS), and as a biomarker in clinical trials. Small, body worn sensors hold the possibility to provide greater reliability and accuracy than existing clinical tools. Since these sensors can provide a variety of metrics, they have the potential to provide a richer and more holistic assessment of gait impairment than existing clinical tools. Paradoxically however, the sheer number and partial overlap between the metrics provided by these sensors has led to confusion and impeded their clinical translation and acceptability.

Objectives

This study in the first to establish a data driven conceptual model of factors contributing to gait disturbance in pwMS using data obtained from body worn sensors. We then tested the model for its ability to quantify gait differences across different levels of disability and clinical courses of MS.

Methods

We studied 114 pwMS, divided in three groups according to their Expanded Disability Status Scale (EDSS) score. (mild, EDSS ≤ 3.5, moderate, 4.0 ≤ EDSS ≤ 5.5, and severe EDSS ≥ 6), as well as the clinical course of their illness (relapsing remitting or progressive), and 24 healthy controls. Gait was assessed with inertial sensors (OPAL, APDM), located on the lower shanks and on the lower back while they walked for 6 minutes at their self-selected speed along a 10-m path in a hospital corridor.

Results

Thirty-six metrics were initially computed from the sensor data. Twenty of these met quality criteria for exploratory factor analysis, which revealed a gait model consisting of five factors: rhythm/variability, pace, asymmetry, and forward and lateral dynamic balance. After confirming overall goodness with a confirmatory factor analysis, the model was used to investigate differences in gait features across pwMS with different levels of disability. We found significant alterations in rhythm/variability, asymmetry, and pace domains in the mild disability group, which further progressed in the moderate and severe disability group. Dynamic balance, conversely, appeared to be conserved in mild and moderate disability groups, only deteriorating in the severe disability group.

Conclusions

This model of gait in pwMS highlights clinically relevant and differential gait impairment across different clinical disease course and disability levels. The data can be obtained from small body worn sensors in a clinical setting. This approach has potential as an accurate and responsive clinical biomarker in clinical trials and more widely in clinical practise.

Collapse