Despite significant advances in the multiarticulate capabilities of upper-limb myoelectric prostheses in recent decades, up to half of people with upper-limb amputation continue to abandon these advanced myoelectric prostheses in favor of more traditional body-powered devices or no prosthesis at all (Biddiss and Chau, 2007b; Østlie et al., 2012; Salminger et al., 2020). Users cite many contributors to their dissatisfaction, among them unreliable prosthetic control (Biddiss and Chau, 2007a; Espinosa and Nathan-Roberts, 2019). The most recent advances in myoelectric control are not yet commercially available and remain limited to research laboratories. Improving user acceptance of myoelectric prostheses necessitates adequate user involvement in the development process, including in a research setting. Unfortunately, recruiting a satisfactory number of research participants to assess novel myoelectric control strategies is impeded by the high fabrication cost and low electrode count of traditional, individualized myoelectric sockets. Alternatives to embedding electrodes into sockets, such as intrafascicular electrodes or targeted muscle reinnervation, are surgically invasive and even more costly to implement (Engdahl et al., 2015). Despite considerable interest in myoelectric control from individuals with upper-limb amputation (Engdahl et al., 2017), user involvement in research studies is often limited to just one or a few individuals with upper-limb amputation (Hebert and Lewicke, 2012; Jiang et al., 2014a; Amsuess et al., 2015) often working in virtual reality environments (Kumar and Todorov, 2015; Kluger et al., 2019; Teh and Hargrove, 2020) or with a robotic arm mounted apart from the user (Ninu et al., 2014; Chadwell et al., 2016; Zhuang et al., 2019). Other studies validate only with healthy participants (Englehart and Hudgins, 2003; Hahne et al., 2014; Jiang et al., 2014a,b; Vidovic et al., 2016; Parajuli et al., 2019) or offline analyses in which there is no active human-in-the-loop involvement (Ortiz-Catalan et al., 2015; Geng et al., 2017).

One approach to increasing the number of studies that validate results with individuals with upper-limb amputation is to reduce the cost per participant by using an adjustable socket (Infinite-Socket^TM-TF; Schofield et al., 2019). An adjustable myoelectric socket could accommodate dynamic changes in the residual limb that would otherwise require casting a new socket. Additionally, an adjustable myoelectric socket could be used with multiple individuals and multiple terminal devices. Recent work toward reducing the cost of custom, individualized sockets has focused on 3D printing, embedded gel liners, and expandable foams (Buis et al., 2017; Reissman et al., 2018; Hallworth et al., 2020; Ismail et al., 2020; Miller et al., 2020; Olsen et al., 2021). These sockets have the added advantage of not requiring a prosthetist for the fabrication and fitting processes. Notwithstanding, these approaches still require a lengthy fabrication process for each person, involving protracted laboratory visits before the participant can begin using the socket. While sockets that are both customizable and affordable have been explored, they have yet to be adapted for myoelectric prosthesis use (Thomas et al., 2015). It should be noted that prosthetists traditionally fabricate diagnostic check sockets to quickly assess and refine socket fit. However, these clinical check sockets are not readily available to researchers and are not intended to assess functionality for myoelectric control strategies. Clinical check sockets are used primarily to validate the length and comfort of the socket, and they do not accommodate electrodes for electromyography.

To address these needs, we developed a multi-user, 3D-printed, transradial, functional-test socket for validation of new myoelectric control strategies in research settings, hereafter referred to as simply “functional-test socket.” By “functional-test” we mean that the socket allows for functional testing of prosthetic control. Researchers can fabricate the components prior to the participants' arrival, allowing participants to don and use the customized socket within minutes. We explored its comfort and functionality with a high-count surface electromyography (sEMG) control system. Subsequent use by other participants does not typically necessitate reprinting of parts, as the 3D-printed material can be reshaped. The development of this socket constitutes an important step toward expanding the involvement of individuals with transradial amputation in state-of-the-art myoelectric-control research.

We developed a novel, universal 3D-printed socket that rapidly provides myoelectric control to individuals with transradial amputation in research settings. The 3D-printed fabrication approach has shown to be an appropriate short-term alternative to conventional techniques, offering high functionality at a small fraction of the time and cost of traditional socket-fabrication methods (Hallworth et al., 2020; Ismail et al., 2020). Our socket can be fabricated in ~7 h at any time prior to participant arrival, custom fit within 10 mins, and donned in under a minute. The total cost of materials of our functional-test socket is ~$8.00 USD. In contrast, traditional sockets, albeit intended for long-term use, require three to four visits with a prosthetist over 3–6 weeks and result in an estimated $800 to $3,000 USD before affiliated labor costs (Types of Prosthetics and Other Prosthetic FAQs Handspring; Frossard et al., 2017). Although less expensive diagnostic check sockets are routinely created by prosthetists during the socket fabrication process, they are intended to check proper fit rather than for functional use, much less myoelectric control (Quigley, 1985; Olsen et al., 2021). Our design can greatly increase participant involvement by further reducing the price per participant, making custom myoelectric control far more accessible.

Other attempts have been made to reduce costs and promote socket accessibility for individuals with upper-limb amputation. One group used photogrammetry to create a 3D scan of the residual limb onto which a socket could be generated and then fabricated using 3D printing (Ismail et al., 2020). This approach required at least two participant visits: one to scan the limb and another post-fabrication to test the socket. Additionally, the resulting socket was inherently specific to a single individual and would not generalize to others without repeating the fabrication process. Another group explored a low-cost and time-efficient solution for low- and middle-income countries that did not require electricity and instead used a rapidly expanding foam for socket adhesion (Miller et al., 2020). Though cost-effective ($100 USD) and quick to fabricate (~90 mins), this socket was specific to one individual, and a harness was needed to maintain a proper fit. Additionally, myoelectric control was not explored with this prosthesis. In another study, this same group also developed a gel-liner system with embedded electrodes that could be used with an individual's existing traditional transhumeral socket to enable myoelectric control (Reissman et al., 2018). Don times for the liner and the prosthesis were ~30 secs, and the liner was tested at home by the users over an average of 7 weeks. In that study, cost was not a specified design requirement, and, while not explicitly discussed, it can be reasonably assumed that it might be relatively high given the materials used and the customized nature of the device. Finally, a group explored a modular, low-cost myoelectric prosthetic platform that could accommodate individuals with or without transhumeral amputation and various terminal devices (Hallworth et al., 2020). While promising, it was only tested on able-bodied individuals, costs $200 USD, and performance was assessed ad-hoc in a gesture classification task rather than with a more functional, real-time task involving a physical prosthesis.

Our present work builds on these previous works and offers great customizability that addresses the high cost of traditional sockets as well as the variation in residual limbs, terminal devices, and DAQ methods. Our post-fabrication fitting process is especially useful for rapidly accommodating different residual limbs and individuals without requiring the reprinting of parts. Moreover, the socket adapts to residual-limb volume fluctuations and can be molded to avoid painful sites (e.g., bone protrusion, neuroma, etc.). Further, the adaptability to nearly any myoelectric terminal device and DAQ method indicates the value of this socket in a wide variety of research applications, potentially even beyond electromyography to applications in magnetomyography, sonomyography, or tactile myography (Kim and Colgate, 2012; Heidari et al., 2018; Calado et al., 2019; Dhawan et al., 2019; Connan et al., 2020). We note that in the present work, a single DAQ system was tested, and although additional DAQ systems for sEMG are theoretically compatible with the socket, these other systems have not been directly tested with this design. Other methods, could be potentially sensitive to the external pressure applied by the struts. The socket presented here can provide great value clinically for implementation with individuals interested in exploring the use of myoelectric prostheses; our socket allows for rapid use with no prior authorization from insurance nor financial commitment needed. The opportunity for individuals with transradial amputation to investigate myoelectric control with physical instead of virtual prostheses grants participants a more immersive and authentic experience to better determine if a myoelectric prosthesis is ideal for them. Importantly, the socket presented here is not intended to serve as a clinical diagnostic check socket, nor is it meant for long-term use as a final socket; rather, it is best utilized briefly in a research or clinical setting.

The functionality of the socket was not compromised in our design considerations to minimize cost and donning time and to promote adaptability. The mechanical testing demonstrated that, despite its low cost, the socket is robust under expected use conditions of at least 8 kg without detectable slipping or mechanical failure. We note that we did not test the socket to the point of mechanical failure in the present work but rather ensured reliability within use cases common in a research environment. Further testing would be necessary to ensure that the socket does not impose limitations for certain prostheses that are rated for loads beyond 8 kg (e.g., TASKA). We found that the socket presented here does not impede performance with the target-touching task and the BBT. Although neither test fully extrapolates to functionality with prosthesis use in ADLs, they both offer value in the investigation of performance with this socket as well as in comparison to traditional sockets (TASKA^TM; i-Limb® Ultra; Roche et al., 2014; Alkhatib et al., 2019). It is interesting to note that performance during the target-touching task did not necessarily correlate with aptitude with the modified BBT. For example, P1 had a low mean PPT and high RMSE compared to other participants but fared equally well during the modified BBT. This is because the target-touching task requires precise control at ~50% flexion or extension in order to stay in target, which was not a position that the mKF was explicitly trained for. Thus, the decoding algorithm was required to generalize to a novel endpoint for both flexion and extension movements. On the other hand, the modified BBT is much more forgiving with the level of required contraction and could reasonably be accomplished by participants simply contracting beyond the level needed to make sufficient contact with a given block, without a penalty for grasping “too hard”. Additionally, the modified BBT does not test precision with extension movements and simply requires participants to relax in order to open the prosthesis sufficiently to drop each block. While we established that SNR as a group was not compromised by donning the socket, it is unclear why it varied considerably at the individual participant level. These differences could be attributed to changes in electrode adhesion to the residual limb with the added pressure from the socket. However, additional experiments would be needed to investigate why these discrepancies were more pronounced with SNR than with the other functional assessments. The comfort of our socket was acceptable throughout the entirety of the experiment, although it scored lower than participants' traditional sockets immediately following the fitting and donning process. For all participants, comfort degradation throughout their visit did not exceed the minimum detectable change, indicating adequate comfort for multiple hours of use.

Only three participants with specific levels of upper-limb amputation were recruited for this study. While our socket encountered no difficulties accommodating the variance in arm length or circumference across these three individuals, there is a great deal of variability in limb presentations; a larger group should be included in future studies to verify that the socket accommodates individuals who may present with significantly shorter residual limbs. Although the literature indicates that repeated heating and shaping of PLA will not degrade its mechanical properties (Chalgham et al., 2021), we did not methodically evaluate this claim with the present socket, and explicit tests would need to be performed to confirm this for our device.

Future work should focus on expanded implementation and improved comfort. Design and material considerations may accommodate other amputation levels (e.g., transhumeral or lower-limb) to further increase research participation. Greater comfort is of especial importance for the use of heavier terminal devices as to still allow for ADLs and functional tasks beyond controlling and lifting the prosthesis.

The socket design has been open-sourced and is available for other researchers at https://github.com/utahneurorobotics/u-of-u-functional-test-socket. The repository contains .stl files for the 3D-printed parts and a parts list with links to other required off-the-shelf hardware. Additionally, an assembly video is provided in the repository.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the University of Utah Institutional Review Board (Protocol No. 98851). The patients/participants provided their written informed consent to participate in this study.

TH wrote the manuscript, designed experiments, recruited participants, and collected and analyzed the data. AC wrote the manuscript, investigated performance metrics, and collected and analyzed the data. ES refined the socket design by CB, collected the data, and assisted in drafting the manuscript. TT assisted in designing the experiments and collecting data. CB developed the initial version of the design and provided clinical support and expertise. CD provided clinical support and expertise throughout. JG oversaw the development of all aspects of the research. All authors contributed to the article and approved the submitted version.

This research was supported by NIH 1 DP5 OD 029571-01.

CB is employed by Handspring. AC was a voluntary intern at Handspring. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.