Abstract
Closed-loop myoelectric prostheses, which combine supplementary sensory feedback and electromyography (EMG) based control, hold the potential to narrow the divide between natural and bionic hands. The use of these devices, however, requires dedicated training. Therefore, it is crucial to develop methods that quantify how users acquire skilled control over their prostheses to effectively monitor skill progression and inform the development of interfaces that optimize this process. Building on theories of skill learning in human motor control, we measured speed-accuracy tradeoff functions (SAFs) to comprehensively characterize learning-induced changes in skill—as opposed to merely tracking changes in task success across training—facilitated by a closed-loop interface that combined proportional control and EMG feedback. Sixteen able-bodied and one amputee participated in a 3-day experiment where they were instructed to perform the box-and-blocks task using a timed force-matching paradigm at four specified speeds to reach two target force levels, such that the SAF could be determined. We found that the participants’ accuracy increased in a similar way across all speeds we tested. Consequently, the shape of the SAF remained similar across days, at both force levels. Further, we observed that EMG feedback enabled participants to improve their motor execution in terms of reduced trial-by-trial variability, a hallmark of skilled behavior. We then fit a power law model of the SAF, and demonstrated how the model parameters could be used to identify and monitor changes in skill. Significance: We comprehensively characterized how an EMG feedback interface enabled skill acquisition, both at the level of task performance and movement execution. More generally, we believe that the proposed methods are effective for measuring and monitoring user skill progression in closed-loop prosthesis control.
Introduction
Learning to use a myoelectric prosthesis—a motorized hand controlled by electrical signals from residual muscles (EMG)—is a complex process often compared to learning a musical instrument. Traditionally, progress in prosthetic training is measured by simple task success metrics, such as the number of blocks moved in a minute. However, these metrics fail to distinguish between a user simply trading speed for accuracy and a genuine increase in underlying skill. This research addresses the lack of robust, longitudinal metrics for monitoring how a user’s internal "control policy" improves during training by applying classical theories of human motor control to the field of prosthetics.
Methods and Experimental Design
The study involved 16 able-bodied participants and one transradial amputee over a three-day period. Participants performed a modified "box-and-blocks" task requiring them to grasp a block with a specific target force (Level 3 or 5) within a specific time window.
The Closed-Loop Interface
Users controlled the prosthetic hand via EMG and received vibrotactile EMG feedback through an array of motors on their upper arm. This feedback represented the strength of their muscle signals, allowing for a closed-loop control system. The task required matching specific force levels, which were mapped to myoelectric commands and prosthesis inputs as detailed in the technical specifications.
Quantifying Skill via SAF
To decouple speed and accuracy, participants performed the task at four distinct speeds: Slow, Medium, Fast, and Very Fast. This allowed researchers to plot Speed-Accuracy Tradeoff Functions (SAFs). The data was fit to a power-law model: $y = 1 - a(x - 0.8)^b$, where $y$ is the success rate and $x$ is the reach duration. This model allows for the identification of skill acquisition through changes in specific parameters, specifically tracking how the curve shifts as performance improves.
Key Results and Insights
The "Upward Shift" in Performance
A primary finding was that the SAF curve did not change its fundamental slope ($b$) over the three days. Instead, it shifted "upward" as the parameter $a$ decreased. This indicates that training improves accuracy uniformly across all speeds, rather than just at slower, easier speeds. Average success rates improved from 75.4% on Day 1 to 85.3% on Day 3, with the power-law model providing a strong fit ($R^2 = 0.76 \pm 0.09$).
Reduction in Variability
As users gained skill, their trial-by-trial EMG commands became less variable and more stereotypical. This reduction in variability is a classic hallmark of motor learning. The EMG feedback allowed users to "pre-plan" contractions, reaching target forces immediately upon contact rather than relying on mid-movement corrections.
Clinical Implications
Because the slope of the tradeoff remains stable, the researchers suggest that a simplified one-parameter model can monitor a user's skill. This could allow clinicians to estimate skill levels from natural, self-chosen speeds in a clinical setting without requiring complex multi-speed experimental setups.
Discussion and Future Work
While the study successfully demonstrated the SAF model's utility, limitations include the small amputee cohort (n=1) and the relatively short three-day duration. Future research should explore whether the SAF shape changes over weeks or months of expert-level training and if these models apply to more complex, multi-degree-of-freedom tasks. Additionally, comparing these results with open-loop control (no feedback) would further quantify the specific contribution of sensory feedback to skill acquisition.
Summary updated