A new preprint on arXiv asks whether user-specific calibration is still necessary for continuous myoelectric control. The authors note that the need for calibration is a frequently cited barrier to adoption, and they investigate whether cross-user models trained on large multi-user datasets can recognize gestures from unseen users without that step. The study is positioned as a closed-loop comparison, meaning it evaluates controllers in an interactive setting rather than only on offline classification accuracy.
The source is a single preprint, so there are no conflicting findings to compare. However, the abstract provided is truncated, so the paper's actual results and conclusions are not fully visible. What is clear is the research question: can calibration-free controllers match the performance of user-calibrated ones in continuous control? If they can, it would remove a significant hurdle for practical myoelectric prostheses and other assistive devices.
Because the outcome is not stated in the available text, readers should treat this as an early-stage investigation rather than a definitive answer. The preprint's significance lies in its focus on closed-loop, continuous control, which is more demanding than simple gesture classification and closer to real-world use.