Abstract:
Introduction: Learning to operate a myoelectric prosthesis involves complex visuomotor adaptation processes that extend beyond isolated muscle activation, engaging both multi-joint motor coordination and visual feedback systems. Yet, the temporal dynamics of these adaptations in novice prosthesis users remain poorly understood, especially in ecologically valid contexts that reflect everyday functions and training progression.
Methods: This study investigated the co-adaptation of gaze behavior and upper-limb kinematics in seven able-bodied participants using a bypass myoelectric prosthesis over a two-week training period. The prosthesis was controlled sequentially via two surface electromyography (EMG) sensors placed over the forearm flexor and extensor muscles, with muscle activations commanding hand closure and wrist rotation velocity—a method commonly used in commercial devices. Participants underwent structured training and pre-/post-assessments, including the End-State Comfort Effect (ESCE) task, a subset of the Jebsen Hand Function Test, and standardized surveys. As a control, participants repeated the protocol using their non-dominant (left) biological arm, on the same side where the bypass was later attached. Upper-body motion capture and eye-tracking were recorded during ESCE trials to assess visuomotor strategies, and progress in myoelectric capacity was monitored using the Assessment of Capacity for Myoelectric Control (ACMC) scale during four training sessions.
Results: Participants exhibited measurable improvements in functional assessment, clinical myoelectric capacity, and subjective ratings of usability and trust across sessions. Gaze patterns and upper-body kinematics remained largely different from controls, reflecting immediate changes in prosthesis adoption. These measures, however, evolved more slowly, with small adaptations that indicate gradual yet effective training. Motor improvements were particularly evident in trajectory performance, with smoother movements and partially reduced compensatory strategies post-training. Nevertheless, trunk-dominant compensations increased, suggesting a trade-off in adaptation rather than a full return to normative motor patterns. Eye-tracking data revealed shifts in visual attention during prosthetic use: participants showed reduced gaze duration across broad task-related areas and increased fixation on the grasping area. These patterns indicate persisting compensatory visual strategies, likely due to limited sensory feedback. Significant correlations between movement-derived metrics and functional outcomes highlight that individual improvements in motor efficiency and predictive gaze behavior are associated with better clinical evaluations.
Conclusions: Gaze and kinematic metrics jointly reflect early signs of adaptation in prosthesis use and provide valuable insight into the intuitiveness of the system. While short-term training fosters immediate functional gains and gradual evolution of visuomotor strategies, persistent compensations and reliance on visual feedback in the grasping area highlight limitations in intuitive control. These findings underscore the potential of multimodal data analysis to inform adaptive prosthetic design and refine training protocols by addressing aspects of motor learning that may be overlooked in standard assessments. Such approaches could enhance user experience and support a more seamless integration of prosthetic technologies into daily life.
Please note that access to the campus in the Olympiapark from 4:30 p.m. is only possible with a work ID card/ZHS ID card. If you would like to take part in the event, please register at kerstin.laimgruber(at)tum.de