An estimated 15,900 upper limb amputations are performed in the USA each year, and ~6,300 in the EU1. This procedure leads to loss of both motor and sensory function in the affected limb. The aim of an upper limb prosthetic is to restore these functions, and substantial advances have been made in the development of such prosthetics, particularly as regards their mechanical construction and methods to control them. However, providing sensory feedback from prosthetic hands remains one of the greatest challenges; work in this area has led to several research solutions, but virtually none has reached clinical fruition or the prosthetic market.

“providing sensory feedback from prosthetic hands remains one of the greatest challenges”

A prosthetic hand with sensory feedback that closely imitates that of an intact hand provides the user with several benefits: a natural perception of touch, the possibility of the prosthetic hand becoming included in the body image, and reinforcement of the control of the prosthesis. Indeed, the addition of sensory feedback to upper limb prostheses has been shown to improve control, increase embodiment and reduce phantom limb pain1. Restoring sensation to the level of an intact limb, however, is challenging owing not only to the overall number of sensors required — the hand is innervated by more than 60,000 sensory axons that emanate from the median and ulnar nerves2 — but also the size and variety needed to mimic the different types of sensory receptor. A typical sensor would be a strain gauge that measures the compound force generated by a finger, but new avenues of research are being pursued in the area of electronic skin3 that could enable complete sensory mimicry of the human hand.

An optimal sensory feedback device would evoke sensations that are similar in quality and location to those perceived from the intact limb. Ideally, sensory restoration would reproduce normal patterns of neuronal activation, making the resulting sensations natural and intuitive. Several attempts have been made to provide such comprehensive sensory feedback from prosthetic devices, and the approaches used fall into two categories: invasive and non-invasive.

Invasive approaches

Several invasive approaches to creating sensory feedback from upper limb prostheses have been tested during the past 50 years, involving various types of neural interface. The quality of studies has varied — studies with more participants and longer durations have provided the most important contributions.

One approach is the use of longitudinal intrafascicular electrodes, which are flexible wires that are used to stimulate individual fascicles within a nerve. One of the most notable studies of this technology, published about 20 years ago, involved six people with upper limb amputations at the elbow or below4. In this study, the interface enabled force and position feedback to be provided by modulating the frequency and amplitude of the stimulation current applied to the nerves.

An alternative approach is the use of flat interface electrodes that wrap around the nerve and flatten it to increase the proximity to the fascicles. In a study of this approach published around 10 years ago, an interface consisting of flat interface electrodes was implanted in the median and ulnar nerves of two people with upper limb amputations for a period of more than 1 year5. Electrical stimulation via this interface improved the participants’ ability to control the gras** strength of the prosthesis and enabled them to better manipulate delicate objects, and this benefit remained stable during the course of the study. Other studies have demonstrated the use of similar technology, but the duration of this study makes it particularly notable.

Multielectrode arrays can also be used to stimulate nerves. In a study published in 2016, a 96-channel electrode array was implanted for 30 days in the median or ulnar nerve of two people with upper limb amputation, and stimulation evoked more than 80 sensory percepts at different locations6. Modulation of stimulation parameters varied the quality of these percepts (for example, tingling, pressure or vibration).

An alternative approach is a technique called targeted reinnervation, in which the remaining arm nerves are transferred to residual chest areas after amputation. This procedure was originally intended to aid motor control but leads to touch on the chest area being perceived as touch on the missing limb7, meaning this area can be used as the target of sensory stimulation.

Non-invasive approaches

Non-invasive sensory feedback approaches typically involve information being conveyed to the skin without the need for surgical intervention. Feedback provided by non-invasive methods is broadly categorized into electrotactile, vibrotactile and mechanotactile, and each of these types of feedback have been studied for their ability to benefit control of upper limb prostheses. The examples described below are among the key studies demonstrating the efficacy of these methods.

In a study that involved nine people with upper limb amputation, the use of surface electrotactile stimulation with an array of electrodes demonstrated that electrotactile feedback can be used for the training of force control with myoelectric prostheses8. In a study of vibrotactile feedback, six people with upper limb amputation at the elbow or below received feedback via a dual tactor system, which consists of two vibrating units (tactors) that deliver vibrations to the skin to simulate different intensities and patterns of touch. The feedback improved grip force accuracy for light and medium forces9. Use of mechanotactile feedback was evaluated in seven people with upper limb amputations at the elbow or below10. A completely passive system based on pneumatic air bubbles, with no active electronics, was used for 4 weeks, and the users’ experience was evaluated. The feedback contributed to a feeling of completeness for the users, but did not provide objective improvements in performance as assessed by functional tests, questionnaires and structured interviews.

Future prospects

Several hurdles remain in the development of upper limb prostheses that provide sensory feedback for clinical use11, and translation of the benefits observed in controlled laboratory settings to everyday use is still early in development. First, the effectiveness of sensory feedback technologies varies greatly between users, which complicates quantitative comparisons and necessitates the development of personalized technology.

“Several hurdles remain in the development of upper limb prostheses that provide sensory feedback”

Second, invasive and non-invasive technologies offer different benefits, and a balance needs to be found between providing detailed feedback, which is easier with invasive approaches, and maximizing usability and safety, which are benefits of non-invasive approaches. Implantable systems can also suffer from mechanical failures and fibrous encapsulation, which decreases functionality over time. The development of new biocompatible materials and electronics could create interfaces with better long-term stability, but until this technology fully matures, non-invasive approaches remain more viable options.

Third, the necessity for natural sensation is under scrutiny — functionality might ultimately outweigh the need for natural-feeling feedback, as users can adapt to useful artificial sensations. Finally, improving sensor coverage in prosthetic hands is also crucial; develo** high-resolution sensors and artificial skins could substantially improve the quality of sensory feedback, with the potential to mimic human sensory capabilities and improve user dexterity.

Overall, the field requires innovative technological development and a deep understanding of user-specific needs to make advanced prosthetic solutions both sophisticated and practical for everyday use.