Written and maintained by CASRAI Editorial Board
Last updated
General information only, not medical advice. Decisions about any prosthesis belong to the user and their clinical team.
Short answer
A myoelectric prosthesis is an artificial limb that reads electrical signals produced by the user’s own muscles and converts them into commands for motors in the device. The published control schemes range from a simple on/off threshold to pattern recognition. Accuracy measured on a bench is often much higher than performance in daily life.
Definition and scope
Myoelectric signals (MES) are the electrical activity of muscle, used to identify the user’s intention to control assistive devices. A 2016 review in Medical Devices lists amputee prostheses, orthoses and exoskeletons as devices that use them (Geethanjali, PMC4968852, read 2026-11-08). This page covers the control side of a myoelectric prosthesis: how a muscle signal becomes a movement. For hand types, components and the wider device landscape, see the guide to the prosthetic hand; lower-limb components are covered in parts of a prosthetic leg and the prosthetic foot guide.
How a muscle signal becomes a command
The review describes five decoding schemes for turning myoelectric signals into device commands:
- On-off (threshold) control. The amplitude of the EMG signal, expressed as RMS or mean absolute value, is compared with a preset threshold. The comparison decides whether the motor is switched on or off; the hand runs at constant speed in either direction, with a full stop.
- Proportional control. The voltage applied to the motor is proportional to the contraction level or intensity of the EMG signal, so a stronger contraction produces faster movement.
- Finite state machine control. Hand postures are predefined as states, and the user’s EMG commands move the controller from one state to the next.
- Pattern recognition. The EMG is segmented, features are extracted from each segment, and a classifier maps those features to an intended movement.
- Postural control. EMG signals are mapped to control parameters in the principal component domain, which the review describes as allowing simultaneous control of the hand.
The review also describes targeted muscle reinnervation and neural interfaces as research approaches aimed at restoring function through the nerves. They are mentioned here as context only and are surgical matters for clinical teams.
Lab accuracy versus daily-life performance
A review of confounding factors in myoelectric control reports that classification accuracy exceeds 90% in controlled laboratory settings, yet devices face robustness problems in daily living (PMC7146367, read 2026-11-08). Across six datasets, performance was 13% lower when the limb position changed and 20% lower when contraction intensity changed, compared with the controlled setting. The authors recommend data-collection protocols that cover at least six limb positions (four limb positions across three forearm orientations).
| Condition tested | Reported effect on performance versus controlled setting |
|---|---|
| Change in limb position | 13% lower |
| Change in contraction intensity | 20% lower |
The practical reading for anyone comparing studies: a headline accuracy figure from a fixed-posture laboratory protocol does not describe how the same control scheme behaves when the arm moves through everyday positions and efforts.
Where it matters for labs, clinics and supply decisions
Research and evaluation. When appraising a control-scheme paper, check whether limb position and contraction intensity were varied, since those two factors are the ones the review quantifies.
Regulation in the United States. FDA describes one class of powered upper-limb devices in 21 CFR 890.3450: a prescription upper extremity prosthesis including a simultaneously powered elbow and/or shoulder with more than two simultaneous powered degrees of freedom, controlled by non-implanted electrical components. It is Class II with special controls, which include electronic compatibility, electrical safety, thermal safety, mechanical safety, battery and software verification and validation (Cornell LII copy of the CFR, read 2026-11-08). The product code PAE, “Upper extremity prosthesis with multiple simultaneous degrees of freedom and controlled via cutaneous electromyography”, is Class 2 in the openFDA classification database (checked 2026-11-08). There is no FDA device category named “myoelectric”, so “myoelectric” describes a control method, not a regulatory class.
Common misunderstandings
- “Myoelectric” means the limb is wired to the brain. The signals described in the sources above are muscle (EMG) signals; nerve-based approaches are a separate research area.
- High reported accuracy means reliable everyday use. The sources report accuracy above 90% in the lab and substantial drops when limb position or contraction intensity changes.
- Every powered prosthesis is regulated the same way. 21 CFR 890.3450 applies to a defined class of devices; other prostheses fall under other regulations in Part 890.
FAQ
What does myoelectric mean?
It refers to the electrical signals generated by muscle, which a myoelectric device uses to infer the user’s intended movement.
How does a myoelectric prosthesis know when to move?
In the simplest published scheme, the amplitude of the muscle signal is compared with a preset threshold, which switches the motor on or off. More complex schemes such as proportional control and pattern recognition use more of the signal.
Is a myoelectric prosthesis regulated by the FDA?
Prostheses are medical devices. Powered upper-limb systems meeting the description in 21 CFR 890.3450 are Class II with special controls.
Sources
- Geethanjali P. Myoelectric control of prosthetic hands: state-of-the-art review. Med Devices. 2016. PMC4968852. Read 2026-11-08.
- Current Trends and Confounding Factors in Myoelectric Control: Limb Position and Contraction Intensity. PMC7146367. Read 2026-11-08.
- 21 CFR 890.3450, Upper extremity prosthesis (via Cornell LII). Read 2026-11-08.
- openFDA device classification API, product code PAE. Checked 2026-11-08.
