Entropy analysis of tri-axial leg acceleration signal waveforms for measurement of decrease of physiological variability in human gait.

Abstract

Disease-related and senescent decrease of physiological variability in biological time-series outputs (e.g., heart rate) has drawn increasing attention as a potential new type of biomarker. In this paradigm, measurement of variability in periodic motion may enable quantitative evaluation of functional limitation in people with musculoskeletal disorders. A novel technique to measure variability of leg motion patterns during level walking was used to study 52 adults with symptomatic knee osteoarthritis (OA), and 57 asymptomatic control subjects over a wide range of age (20-79 years). The hypothesis was that cycle-to-cycle variability in leg motion patterns, indexed by tri-axial acceleration signal entropy, would be lower in those with greater age or with knee symptoms. Leg motions were assessed using portable inertial monitors attached bilaterally just above each ankle. The tri-axial acceleration data were analyzed using a nonlinear variability measurement tool designated as Sample Entropy (SampEn). SampEn data for asymptomatic subjects exhibited a significant negative correlation (r = -0.287, p = 0.0306) with greater age. OA subjects had significantly lower SampEn values (p = 0.0002) than did age-matched asymptomatic subjects who walked at equivalent velocity. This approach holds promise as a basis for valid, inexpensive, and convenient objective evaluation of limitations in human gait function.