A Sensor-based Data Capture Solution for Home-based Gait Assessment: The GAIT-HUB System
Timothy Ko1, Erica Kreisberg1, Luis Fernandez1, Josef Gutman1, Carrie Sammarco1, Tyler Smith1, Leigh Charvet1, Giuseppina Pilloni1
1Neurology, New York University Grossman School of Medicine
Objective:
To validate a sensor-based system for conducting mobility assessments in home settings for people with multiple sclerosis (MS).
Background:
Gait deviations are common manifestations of MS. Currently, clinical mobility assessments predominantly rely on in-person observational evaluations. The "Gait Assessment with Innovative Technologies – Home Use Benefit" (GAIT-HUB) system utilizes off-the-shelf sensor technology to measure gait, offering more sensitive and accurate metrics than traditional clinical measures, with the potential to be used by patients at home.
Design/Methods:
In this ongoing validation study, individuals with MS were recruited and initially underwent a baseline in-person gait assessment. During the baseline evaluation, participants wore a gold-standard wearable tri-axial accelerometer on the lower back (G-sensor, BTS Bioengineering) and two commercially available shoelace sensors (Runscribe) simultaneously when completing a 2-minute walk test. Afterwards, participants, provided with the GAIT-HUB kit (iPad and shoelace sensors), conducted three weekly at-home assessments under live videoconference guidance. Each session involved a sensor-based 2-minute walk test. Gait velocity data from both in-clinic and at-home visits were compared to validate off-the-shelf sensors' accuracy and reliability for self-administered data capture.
Results:
We enrolled n=20 participants (46.5±10.8 years; 75% female; Median Patient Determined Disease Steps (PDDS) score: 3.0). 94.7% of the participants rated GAIT-HUB system’s usability as excellent, with System Usability Score exceeding 80. At baseline, we found a strong agreement between gait velocities assessed by the shoelace and the gold-standard sensor (interclass correlation coefficient (ICC) of 0.989; p<0.001). Reliability across three remote sessions was substantial, with ICC=0.990 (p<0.001). When compared to in-clinic, participants showed a slower gait velocity in home setting (1.22±0.20 vs. 1.18±0.17 m/s, p<0.001), potentially reflecting variations in environments and assessment conditions.
Conclusions:
The GAIT-HUB system demonstrates feasibility for home-based data capture, exhibiting strong consistency with in-clinic assessments and reliability across repeated measures. This study substantiates the possibility of utilizing wearable sensors for at-home gait assessments.