JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI)-based gait analysis systems are increasingly used in rehabilitation settings due to their ability to provide objective and quantitative assessments of gait function. However, clinical adoption of these systems remains limited, primarily due to insufficient attention to usability and user experience (UX), particularly in real-world clinical environments.

Objective:

This study aimed to evaluate the usability of an AI-based gait analysis system (MediStep M) through a formative usability evaluation, focusing on human-technology interaction and user-centered design improvements.

Methods:

A mixed-methods usability evaluation was conducted involving five licensed physical therapists. Focus group interviews (FGI) were used to collect qualitative data on user experiences, and the System Usability Scale (SUS) was administered to quantify perceived usability. The evaluation took place at a national usability testbed and followed a structured scenario-based procedure.

Results:

The FGI revealed key usability issues, including poor accessibility of the power button, lack of battery status indicators, burdensome manual calibration, and insufficient feedback in the gait analysis report. Participants also highlighted the need for wireless operation, enhanced portability, better patient data management, and integration with hospital electronic medical record (EMR) systems. The SUS score averaged 57, corresponding to a grade of “D,” indicating suboptimal usability, especially in the domains of Utility, Integration, and Overall Satisfaction.

Conclusions:

Despite the clinical potential of AI-based gait analysis systems, critical usability barriers remain. Improvements in hardware design, automated calibration, clinical report specificity, and interface integration are necessary to support broader clinical adoption. User-centered design approaches are essential for enhancing system effectiveness and ensuring alignment with rehabilitation workflows.