Comparability of Methods for Remotely Assessing Gait Quality.

Area of Science:

• Neurology
• Biomedical Engineering
• Digital Health

Background:

• Remote gait analysis offers efficient, cost-effective, and personalized real-time assessments for individuals with Parkinson's disease (PD).
• Traditional gait assessment methods can be limited in accessibility and real-time feedback capabilities.

Purpose of the Study:

• To compare the gait quality metrics of three remote gait assessment methods in individuals with PD: observation, wearable sensors, and pose estimation.
• To provide evidence on the comparability of these technologies for remote PD gait monitoring.

Main Methods:

• A cross-sectional, multiple case series study involving 20 participants with PD remotely assessing gait.
• Participants performed a modified Timed Up and Go (TUG) test using the Heel2Toe^TM wearable sensor.
• Videos were analyzed using a PD-specific observational checklist and the MediaPipe Pose Landmarker task estimation library.

Main Results:

• Observational ratings agreed with the Heel2Toe^TM wearable on detecting heel strike 64% of the time and push-off 28.5% of the time.
• Significant differences in gait parameter ranks were found between methods, except for push-off when comparing MediaPipe pose estimation to observational checklist ratings (p = 0.498).

Conclusions:

• A combination of digital technologies, including wearable sensors and pose estimation, can detect subtle gait impairments in PD.
• These advanced methods offer potential for more comprehensive and sensitive remote gait analysis compared to human observation alone.