The intent of this section is to briefly introduce a range of frontier directions for improving the outcomes of patients with amputations. A focus is placed on breadth as opposed to depth, such that a wide range of topics can be covered, ranging from the conceptual (TRL 1-5) to those nearing practical use (TRL 5-8). Specifically, this section addresses ongoing work that aims to further close the loop between a person and the person’s robotic prosthesis, the deployment of implantable technologies, the use of machine learning and intelligent systems for advanced control, and groundbreaking new robotic prostheses that significantly extend the function of currently available devices.
Bidirectional Control and Feedback Targeted Motor and Sensory Reinnervation
Surgical reconstruction of the amputated limb plays an essential role in maximizing outcomes for prosthetic applications. In addition to advances in bone management, residual muscle management, and skin coverage, advanced nerve procedures have been developed to improve the ability to extract the rich control signals that are lost after upper limb amputation. Targeted reinnervation (TR) surgically redirects the amputated nerve endings that used to innervate the hand and wrist muscles to new muscle sites to provide physiologically natural motor command signals for myoelectric control (Kuiken, Schultz Feuser, and Barlow 2013). The surgically redirected nerves reinnervate purposely denervated remaining muscles, which then act as biological amplifiers for the neural signals that are still under voluntary (brain) control. These muscle responses, which are intuitively activated, are then linked to the action of the prosthesis. After reinnervation, patients are able to operate multiple degrees of freedom of advanced prosthetic devices with increased ease. Combining newer surface EMG recording techniques (such as pattern recognition) with TR may allow even more signals to be extracted for prosthetic control. Recently, in subjects with upper limb amputation having undergone TR, simultaneous pattern recognition control was found to be superior in preference and performance to both sequential pattern recognition and conventional myoelectric control (Young et al. 2013).
In addition to improved motor control, TR provides a potential avenue for sensory feedback. Redirection of the amputated sensory nerves to denervated skin restores the sensation of the hand and fingers on the new target area of skin (Hebert, Elzinga et al. 2014; Marasco, Schultz, and Kuiken 2009). This “transfer sensation” from reinnervation of the sensory afferents is a possible access point to provide physiologically natural and appropriate avenues of cutaneous touch and proprioceptive feedback through robotic devices (Hebert, Olson et al. 2014). Ongoing research in this area is linking haptic feedback to tactor devices that stimulate the skin and muscle in proportion to sensors on the prosthesis, thereby providing real-time bidirectional feedback in a noninvasive socket system.