Automating provision of feedback to stroke patients with and without information on compensatory movements: A pilot study

Abstract

Providing effective feedback to patients in a rehabilitation training program is essential. As technologies are being developed to support patient training, they need to be able to provide the users with feedback on their performance. As there are various aspects on which feedback can be given (e.g., task success and presence of compensatory movements), it is important to ensure that users are not overwhelmed by too much information given too frequently by the assistive technology. We created a rule-based set of guidelines for the desired hierarchy, timing, and content of feedback to be used when stroke patients train with an upper-limb exercise platform which we developed. The feedback applies to both success on task completion and to the execution of compensatory movements, and is based on input collected from clinicians in a previous study. We recruited 11 stroke patients 1-72 months from injury onset. Ten participants completed the training; each trained with the rehabilitation platform in two configurations: with motor feedback (MF) and with no motor feedback (control condition) (CT). The two conditions were identical, except for the feedback content provided: in both conditions they received feedback on task success; in the MF condition they also received feedback on making undesired compensatory movements during the task. Participants preferred the configuration that provided feedback on both task success and quality of movement (MF). This pilot experiment demonstrates the feasibility of a system providing both task-success and movement-quality feedback to patients based on a decision tree which we developed.

Keywords: compensatory movements; exergames; human-computer interface; human-machine interface; patient-centered design; serious games; stroke; user experience.

FIGURE 1

The experimental setup. Left: The participant was seated in front of a height-adjustable table, behind which was a 27-inch computer screen, on which instructions and feedback were presented. The clinician sat in the same room, next to a second computer, used to monitor the experimental progression and input information about movement compensations that the participant performed. In each of 12 trials, a set of colored circles was displayed on the computer screen, arranged in a circle around a central location, similar to a bullseye arrangement. The participant had to place a corresponding set of colored cups on the table according to the picture shown on the computer’s screen. Right: The Target Exercise Game with all seven cup locations occupied, arranged according to the on-screen instructions.

FIGURE 2

The decision tree used in the experiment. It is depicted as a flow chart, and is based on the hierarchical structure suggested by clinicians in a previous study (Fruchter et al., 2022): feedback on task success precedes feedback on compensations; within compensations, order is set by the clinician (here, set to a default order: trunk flexion, scapular elevation and elbow flexion).

FIGURE 3

Numeric scales to indicate levels of pain and fatigue. Used following both the CT and the MF conditions.

FIGURE 4

Results of the 11-item modified System Usability Scale (SUS) questionnaire (mean ± SD). A score of 1 reflects “Strongly disagree” and 5 reflects “Strongly agree.” Results from the CT condition are marked in gray and results from the Motor-Feedback condition are marked in green; The “need for support” category is marked with a different color, as it was negatively worded: a lower score corresponds to higher usability of the platform. The rightmost category (“Feedback on compensations”) shows a single bar, since feedback on compensations was provided only in the MF condition.

FIGURE 5

Responses to the five statements that were only presented in the MF condition (mean ± SD). A score of 1 reflects “Strongly disagree” and 5 reflects “Strongly agree.” The question numbers correspond to the questions listed in Table 2. The rightmost bar is marked with a different color, as it was negatively worded: a lower score corresponds to higher usability of the platform.