Poster #: 36A
Session/Time: A
Author: Reza Pousti, PhD, MS
Mentor: Christopher K. Rhea, PhD
Research Type: Basic Science

Abstract

Digital wearables enable large-scale, ecological gait assessment, yet many commonly used gait metrics lack mechanistic interpretability and cannot identify stride-level symmetry dynamics. To this end, we developed the Reza Dimensional Equilibrium for gait (RDE-GAIT) method and evaluated it using a public mobile brain-body dataset of 35 healthy young adults from Hanada, Kalabic, Ferris (2024). Using a left ankle IMU with a waist reference, we reconstructed the 3D leg vector at the forwardmost (anterior) and backward-most (posterior) locations during each gait cycle, and flipped/mirrored the posterior vector so that it could be directly compared to the anterior vector in 3D space. We defined the similarity between the vectors as p^((k)) ranging from 0 to 1, with values near 0 indicating less similarity and values near 1 indicating more similarity. Next, we normalized p^((k)) to a robust local baseline p^0 computed as a rolling median over a 200-cycle moving window, so each gait cycle is evaluated relative to nearby cycles (100 cycles before and 100 after) within the session. Lastly, an equilibrium value at each gait cycle termed E^((k)) was calculated as p^((k))/p^0, with 0.95≤E≤1.05 indicating a consistent gait cycle relative to the surrounding gait cycles and E<0.95 or E>1.05 indicating inconsistent gait cycles (either an inconsistent gait cycle embedded within consistent gait cycles or a consistent gait cycle embedded within inconsistent gait cycles). Outdoor walking sessions contained 1,326 5,739 gait cycles (median 4,632). The median proportion of consistent gait cycles was 78% (range 12.2 94.8%), and the median proportion of inconsistent gait cycles was 22% (range 5.2 44.8%). Twenty-eight participants exhibited consistency in >70% of gait cycles, with eight of those exhibiting a very high proportion of consistency in their gait cycles (>85%). On the contrary, three participants exhibited a high proportion of inconsistency in their gait cycles (>65%). What remains an empirical question is the extent to which the similarity metric (p) and/or the equilibrium metric (E) relate to adaptive and/or functional gait patterns. Future work should explore these questions, as well as compare RDE-GAIT to traditional symmetry gait metrics across diverse locomotor tasks (e.g., running, irregular terrain, etc.).