Gait Speed Assessment in the 10-meter Walk Test for Older Adults Using a Computer Vision-based System: A Cross-sectional Study on Validity, Reliability, and Usability

The 10-meter walk test (10MWT) is a useful field test for gait speed assessment in older adults. However, the conventional 10MWT only provides an overall average gait speed and does not monitor changes in gait speed at specific times. A computer vision-based system is a markerless motion-tracking technology that detects an individual’s instantaneous gait speed, providing valuable insights into gait control.

This cross-sectional study aimed to investigate: 1) the concurrent validity of a computer vision-based system against a motion analysis system for capturing instantaneous gait speed in the 10MWT; 2) the test-retest reliability of gait speed measures obtained from a computer vision-based system in the 10MWT; and 3) the usability of a computer vision-based system for gait speed assessment during the 10MWT in free-living environments.

In the validity and reliability testing phase, ten older adults (mean age = 67.50 (6.36) years) participated. Participants performed the 10MWT under two walking conditions: walking at a comfortable speed and walking at maximum speed. For the validity testing, the instantaneous gait speed obtained from a computer vision-based system was compared against that from a motion analysis system. After 30 minutes, the same protocol was repeated to assess the test-retest reliability of a computer vision-based system. The outcome measures were the average instantaneous gait speed for each meter (i.e., meters 0-1, 1-2, 2-3, 3-4, 4-5, and 5-6). In the usability testing phase, six physical therapists (mean age = 26.83 (0.98) years) were asked to test a computer vision-based system for 10MWT assessment in a free-living environment. After completing the test, they were asked to rate the perceived usability of a computer vision-based system using the System Usability Scale (SUS). A Spearman's rank correlation was used to determine a correlation between a computer vision-based system and a motion analysis system, with significance set at P-value < 0.05. An intraclass correlation coefficient (ICCs (3,2)) was used to test the agreement between two repeated sessions of a computer vision-based system. Descriptive statistics were employed to analyze the SUS score.

The average gait speed obtained from a computer vision-based system and a motion analysis system showed high to very high agreement in all meters across two walking conditions, with correlations ranging from 0.70 to 1.00 (P-value < 0.05). The average walking speed measured by a computer vision-based system demonstrated very high repeatability across two sessions for all walking conditions, with ICCs falling between 0.940 and 0.986. The total SUS score for all participants was 75.83 (7.36), suggesting a good perception of the system's usability.

A computer vision-based system is accurate, consistent, and acceptably user-friendly, making it a promising approach for measuring instantaneous gait speed of the 10MWT. However, the proposed system would need to be improved in terms of feasibility for use in community-based settings.