Patient-recorded home video feasible method to assess gait kinematics in MS

Key takeaways:

• Adults with MS took home videos of themselves walking and uploaded them to a digital platform.
• Video taken in home were a feasible, accurate, low-cost method of acquiring data.

WEST PALM BEACH, Fla. — In-home video taken by adults with multiple sclerosis is a feasible and valid method to determine the mechanism and motion of their gait, according to a poster from ACTRIMS 2024.

“Last year we looked at dexterity tasks and what we wanted to do this year was ... looking at gait tasks, which are inherently a lot more complex,” Riley M. Bove, MD, a neurologist at the Weill Institute for Neurosciences at the University of California, San Francisco, told Healio.

Having previously examined dexterity in upper extremities in those with MS, Bove and colleagues sought to expand the body of knowledge through evaluation of the feasibility and validity of measuring gait quality and range of motion. They accomplished this through analyzing patient-recorded videos in their home environment using human pose estimation, a set of algorithms trained on images that detect body landmarks.

Their ongoing, transdiagnostic digital phenotyping study included 30 adults with MS who spent both time within a clinic and time at home.

In the clinic, participants walked at a preferred speed across an instrumented, pressure sensor walkway to generate precise gait kinematics including step length for each limb, as well as velocity readings. Standard analysis measures such as the Expanded Disability Status Scale score and Timed 25 Foot Walk for each were included with the data set. Then, within the home and utilizing a personal smartphone or device, participants recorded videos of themselves walking 20 feet and uploaded them to an electronic study platform. Position and velocity data were extracted from the videos using open access MediaPipe Holistic pose estimation software, with the big toe on each foot as a key landmark.

According to results from the first 18 participants analyzed (mean age 48.5 years; 70% female), feasibility of video tracking was good. From 68 total requested home videos, researchers were able to analyze 62 (91%).

Extracted velocity analyzed from the home videos was associated with key domains of gait measured by the instrumented walkway including step length [right foot (r -0.41, P = 0.01), left foot (r -0.42, P = 0.02) and velocity (r 0.39, P = 0.05).

Additionally, total displacement of the big toe in the videos was correlated with the same measured by the instrumented walkway (r -0.45, P = 0).

“We were able to extract measures that correlated very well with our clinical measures of dexterity but also with patients’ self-reported differences,” Bove told Healio. “Further, we were able to show over 6 months through the videos that we were better able to pick up some images than in the clinic.”