Abstract
Objective. 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. Approach. 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 healthy participants and one individual with a transradial limb loss participated in a three-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. Main results. 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 device controlled by electrical signals from residual muscles—is a complex motor task that often requires months of dedicated training. A significant challenge for researchers and clinicians is accurately measuring this progress. Traditional metrics often rely on success rates at a single, comfortable speed. However, such measures can be misleading; a user might appear more accurate simply by slowing down, which does not necessarily reflect an increase in underlying skill. To address this, we propose a framework based on Speed-Accuracy Tradeoff Functions (SAFs) to quantify the "skill frontier" in prosthetic use.
Methods and Experimental Design
The study involved 16 healthy participants and one individual with transradial limb loss. Participants used a proportional control scheme where muscle contraction strength dictated the closing force of the prosthesis. This was paired with a closed-loop vibrotactile feedback system providing "EMG biofeedback" via an array of five motors on the upper arm.
The experimental task was a modified "box-and-blocks" paradigm. Users were required to reach specific target force levels within four distinct time windows: Slow (4–8s), Medium (3–5s), Fast (2–4s), and Very Fast (1–3s).
To model the learning process, we utilized a power-law description of the SAF: $$y = 1 - a(x - 0.8)^b$$ where $y$ represents the success rate and $x$ represents the reach time. This model allows for the characterization of skill through specific parameters, potentially simplifying clinical monitoring to the tracking of a single value (the intercept) rather than re-mapping the entire curve daily.
Key Insights and Results
Parallel Learning Across Speeds
A primary finding was that learning was not restricted to specific speeds. Instead, the entire SAF curve shifted upward as training progressed. This "parallel learning" suggests that practicing at various speeds enhances overall control, rather than just performance at the trained speeds.
Reduction in Variability
The acquisition of skill was characterized by a significant reduction in trial-by-trial variability. This decrease in "noise" within the user's muscle commands is a classic hallmark of motor learning, indicating more consistent and precise control over the interface.
The Smoothness Paradox
Movement smoothness, measured via Spectral Arc Length (SPARC), surprisingly decreased as users improved. This phenomenon is likely due to the discretized nature of the feedback system. As users became more precise, they learned to employ "stuttered" commands to fine-tune their force output to match the discrete levels provided by the feedback.
Conclusion
This research demonstrates that SAFs and power-law modeling provide a robust framework for monitoring skill acquisition in myoelectric prosthesis users. By shifting the focus from simple success rates to the evolution of the speed-accuracy relationship, clinicians can better track user progress and optimize training protocols. Future work will focus on validating these models in multi-functional, real-world tasks and exploring the use of observational data for skill estimation.
Summary updated