Sample Entropy Improves Assessment of Postural Control in Early-Stage Multiple Sclerosis

Abstract

Postural impairment in people with multiple sclerosis (pwMS) is an early indicator of disease progression. Common measures of disease assessment are not sensitive to early-stage MS. Sample entropy (SE) may better identify early impairments. We compared the sensitivity and specificity of SE with linear measurements, differentiating pwMS (EDSS 0–4) from healthy controls (HC). 58 pwMS (EDSS ≤ 4) and 23 HC performed quiet standing tasks, combining a hard or foam surface with eyes open or eyes closed as a condition. Sway was recorded at the sternum and lumbar spine. Linear measures, mediolateral acceleration range with eyes open, mediolateral jerk with eyes closed, and SE in the anteroposterior and mediolateral directions were calculated. A multivariate ANOVA and AUC-ROC were used to determine between-groups differences and discriminative ability, respectively. Mild MS (EDSS ≤ 2.0) discriminability was secondarily assessed. Significantly lower SE was observed under most conditions in pwMS compared to HC, except for lumbar and sternum SE when on a hard surface with eyes closed and in the anteroposterior direction, which also offered the strongest discriminability (AUC = 0.747), even for mild MS. Overall, between-groups differences were task-dependent, and SE (anteroposterior, hard surface, eyes closed) was the best pwMS classifier. SE may prove a useful tool to detect subtle MS progression and intervention effectiveness.

Keywords

• stability
• posturography
• balance
• sample entropy
• regularity
• jerk