Pattern Recognition

The Complete Control product line from Coapt adds revolutionary Pattern Recognition machine learning intent decoding algorithms to benefit upper limb powered prostheses


Whenever we move our arms and hands, multiple muscles make coordinated contractions and each muscle emits its own small electrical signature (called myoelectricity), like it is its own instrument in an orchestra. Each different arm and hand movement results in a unique but repeatable set of these myoelectric patterns—like different songs from the orchestra.

Myoelectric signals are very tiny but they can be detected by electrodes on the surface of our skin. Using a full array of electrode contacts on the skin—covering the whole area of these underlying muscle contractions—lets all of the rich muscle pattern information be captured (akin to an array of microphones over our orchestra).

The patterns of information are representations of what the human is intending to do and can be decoded by algorithms calibrated to recognize them.

The complex sets of myoelectric patterns need to be “decoded” in real time and matched to the arm or hand action that is intended. Coapt’s pattern recognition is a system of finely tuned algorithms providing machine learning that does just that, specifically for the residual muscle signals of those with upper limb loss or difference. For example, the pattern of myoelectric activity recorded on the residual forearm during hand opening is different from the pattern recorded while the hand is being closed, or making a point gesutre, or turning a wrist, and so on.

The Complete Control pattern recognition-based intent decoder system Coapt listens to the myoelectric activity and uses mathematical algorithms to determine when a pattern matches the user’s intention to make an arm or hand movement. It then tells the prosthesis to move accordingly, providing intuitive control of multiple prosthetic movements.

“Pattern Recognition is the technology that will render one- or two-site conventional myoelectric control to be obsolete.”



Coapt’s Complete Control pattern recognition system is generally applicable to those with upper limb loss or upper limb difference.

A woman getting fitted to try and calibrate the Coapt system

Candidates for Coapt pattern recognition:

Clinical evaluation is always encouraged to showcase each user’s potential with Coapt’s pattern recognition intent decoding system. In general, the two major factors influencing success are 1) the ability to achieve and maintain reliable skin-electrode contact within the prosthesis interface, and 2) the willingness of the user and practitioner to adopt the intuitive and powerful mechanism of machine learning pattern recognition.

Being a recipient of Targeted Muscle Reinnervation (TMR), Regenerative Peripheral Nerve Interface (RPNI), Osseointegration (OI), or other similar surgeries are not a requirement for using Coapt Complete Control technology but they are complementary and may expand control options and overall function for some individuals.

Coapt knows that all candidates are unique, please contact us to discuss any presentation you may have questions about.


Intuitive Intent Decoding

Natural movement intents are decoded and sent as commands to the prosthesis, generally making it easier to use. For example, the feeling of opening or closing a hand is used to control the prosthetic hand open and close (instead of wrist flexion motions to control the hand common for basic myoelectric devices).

Calibration is Personalization

Function can be refined and personalized at any time using the quick on-board or app-based calibration. This makes addressing changes in socket fit, skin condition, fatigue, desired feelings for control, and much more easy to do without the need to take the prosthesis off or make extra clinical visits.

Superior Proportional Control

A wider range of prosthesis command speeds are available because myo inputs aren’t limited by thresholds and a broader range of input levels can be recognized by the algorithms.

No Mode-Switching

There is no need to use cumbersome actions like co-contractions, pulses, fast-slow gestures, etc. This makes the command of a prosthesis more seamless.

Electrode Array Flexibility

There is much less need for isolated signals or precise electrode placement. This opens the door to more freedom in placing electrodes for socket comfort or limb constraints and provides helpful forgiveness in prosthesis donning.