Multigesture Electromyographic Control Complexity in Upper Limb Prostheses Actuated via Single Sensor Input Contraction Magnitude: Qualitative Study for Evaluating Performance and Cognitive Load.
Source: PubMed, NCBI / U.S. National Library of Medicine
Lack of functionality is one factor that contributes to prosthetic rejection rates. Electromyographic upper limb prostheses are controlled through muscle contractions in the user's residual limb. The incorporation of multigesture controls into a novel, in-house developed upper limb prosthesis requires users to differentiate between the strengths of muscle contractions to trigger programmed gestures. Little research exists on the limitations of expanding device capabilities. This expansion may lead to a decline in accuracy and perceived usability or an increase in training time and cognitive workload. This study aimed to determine the feasibility of implementing multiple gestures when learning electromyographic controls during a single training session. Participants with full upper extremity control were fitted with a Flex Controller, a surface electromyography device that measures muscle contraction. Contractions were visualized as peaks and calibrated through an adjustable scale on a tablet. A training app was developed in-house to test novice users on an electromyography control system. Users interacted with 1, 3, or 5 zones on the screen. Each horizontal zone represented a threshold required to trigger a distinct gesture on the prosthesis. The cohorts were labeled A1 (n=9), A2 (n=10), A3 (n=9), and B1 (n=26). Every participant completed 3 trials per arm, and each trial consisted of 15 randomized cues. Each cue was represented by a green color change, with 1 point earned af
Abstract
Lack of functionality is one factor that contributes to prosthetic rejection rates. Electromyographic upper limb prostheses are controlled through muscle contractions in the user's residual limb. The incorporation of multigesture controls into a novel, in-house developed upper limb prosthesis requires users to differentiate between the strengths of muscle contractions to trigger programmed gestures. Little research exists on the limitations of expanding device capabilities. This expansion may lead to a decline in accuracy and perceived usability or an increase in training time and cognitive workload. This study aimed to determine the feasibility of implementing multiple gestures when learning electromyographic controls during a single training session. Participants with full upper extremity control were fitted with a Flex Controller, a surface electromyography device that measures muscle contraction. Contractions were visualized as peaks and calibrated through an adjustable scale on a tablet. A training app was developed in-house to test novice users on an electromyography control system. Users interacted with 1, 3, or 5 zones on the screen. Each horizontal zone represented a threshold required to trigger a distinct gesture on the prosthesis. The cohorts were labeled A1 (n=9), A2 (n=10), A3 (n=9), and B1 (n=26). Every participant completed 3 trials per arm, and each trial consisted of 15 randomized cues. Each cue was represented by a green color change, with 1 point earned after a successful peak. Collected outcomes included performance, the System Usability Scale, and the National Aeronautics and Space Administration Task Load Index. Scores decreased significantly as zones increased (Kruskal-Wallis H=24.9, P<.001). The mean scores were 15.0 (SD 0.0) for 1 zone, 9.1 (SD 1.1) for 3 zones, and 5.5 (SD 1.1) for 5 zones. Perceived usability, measured by System Usability Scale, showed modest omnibus difference across cohorts (Kruskal-Wallis H=5.22, P=.16); however, a pairwise comparison showed the 5-gesture cohort rated usability lower than the progressive cohort (2-tailed Welch t=-2.19, P=.05). The 5-gesture cohort rated the system lowest (mean 63.3, SD 16.2). Cognitive workload, assessed through the National Aeronautics and Space Administration Task Load Index, increased with the number of gestures. The performance subscale showed a significant omnibus difference across cohorts (Kruskal-Wallis H=21.4, P<.001). Mean performance subscale scores were 84.4 (SD 14.7) for the single-gesture condition, 30.6 (SD 21.7) for the 5-gesture condition, and 44.2 (SD 21.2) for the progressive cohort, reflecting increasing perceived difficulty with more gestures. The sample size for quantitative analysis was 54. These findings support the implementation of progressive training for 3 gestures. Usability perceptions were the highest among the more complicated progressive cohort, which is likely related to perceived improvement. Progressively learning 3 gestures enables a balance between device capability, user intention, perceived usability, and cognitive workload.