Though commercial powered prosthetic arm systems hold potential in restoring function for people with upper-limb loss, effective use of such devices remains limited by conventional (direct) control methods, which rely on electromyographic signals produced from a limited set of muscles.
In a study published October 23 in Scientific Reports, researchers sought to compare pattern recognition and direct control in eight men with transhumeral amputations who had targeted muscle reinnervation (TMR). Each of the participants, who were competent myoelectric prosthesis users prior to enrolling in the study, performed a six- to eight-week home trial using direct and pattern recognition control with a custom prosthesis made from commercially available parts.
The participants took the devices home for a minimum of 42 days in each configuration, and were instructed to use the prostheses to complete activities of daily living and to keep a journal detailing how often they used the prostheses and for what activities it was being used. If a prosthesis needed to be returned for repair or if the user had a valid and documented reason for not wearing a myoelectric prosthesis, then additional time was added to the home trial to ensure six weeks of usage.
Each participant chose between using a powered split-hook (electric terminal device or electric Ottobock Greifer terminal device) or a hand with a single degree of freedom, depending on his preference and experience. All prostheses had a motorized elbow, wrist, and terminal device. Seven of the participants used a powered hook terminal device and one used a myoelectric hand for his terminal device.
Before and after the home trial the following tests were performed: the Southampton Hand Assessment Procedure (SHAP), in which the subject grabbed a variety of objects; the clothespin relocation task, where subjects moved three clothespins from a horizontal to a vertical pole; and the Box and Blocks Test, where subjects moved one block from a box over a four-inch wall to another box. The Assessment for Capacity for Myoelectric Control (ACMC), a validated observational test of the subject performing functional tasks, was only conducted post-trial.
The study determined that the subjects showed statistically better performance in the SHAP and the clothespin relocation task when using pattern recognition compared to using direct control. These tests required movements along three degrees of freedom. Seven of eight subjects preferred pattern recognition control over direct control. There were no significant differences for the Box and Blocks Test, nor were there any changes between pre- and post-home trial testing.
Subjects were able to use both types of control to complete activities that included preparing food, eating, dressing, shopping, and performing household chores. The frequency with which they reported in their journal that they wore the prosthesis correlated with the wear time recorded by the embedded control system. The results indicated that regardless of the type of control system being used, a variety of factors prevented the participants from using their prostheses more often, such as it being too heavy or too warm to wear on hot days.
By the end of the trial, seven of the eight subjects reported that they preferred using pattern recognition control over direct control. Most subjects found performing direct control mode-switching to be inconvenient. Subjects also noted that sometimes the prosthesis moved unintentionally while using pattern recognition control. However, subjects also noticed that it was often hard to control only one degree of freedom when desired using direct control, as simultaneous control of two degrees of freedom was always available. These unintentional movements, regardless of whether it occurred using pattern recognition or direct control, made the control less reliable for the subject.
Study results demonstrated that pattern recognition is a viable option and has functional advantages over direct control. The effects of TMR may be enhanced when paired with pattern recognition technology, the study showed. This study was the first home trial large enough to establish clinical and statistical significance in comparing pattern recognition with direct control.
