Introduction
Neuromuscular impairments resulting from neurological conditions such as stroke, spinal cord injury (SCI), or other neuromuscular disorders often lead to chronic pain and disability1 and present a significant global health challenge.2 Restoring function for these patient populations via neuromodulation technologies offers a promising path to improving motor function, easing pain, enhancing quality of life, and reducing disability. Neuromodulation has previously been defined as “the process of inhibition, stimulation, modification, or therapeutic alteration of activity in the central, peripheral, or autonomic nervous systems through electrical, chemical, or mechanical interventions.”3
Numerous advanced technologies are emerging, designed to modulate neuronal activity within the brain, spinal cord, and peripheral nervous system. This is in contrast to technologies that block nerve activity (for instance, as in the case of radiofrequency ablation, cryoablation, or neurectomy). Neuromodulatory approaches include direct brain stimulation, activation or inhibition of spinal and peripheral motor nerves, and advanced brain-computer interface technologies, among others.
These interventions are aimed at reducing pain and/or restoring or improving motor control, muscle strength, coordination, and overall functional mobility to help patients regain the ability to perform activities of daily living (ADLs).
The rapid evolution of neuromodulation necessitates developing evidence-based guidelines to inform clinical practice in functional restoration and pain management and to identify promising technologies warranting further investment and study. However, a sizeable portion of the neuromodulation literature and consensus guidelines have focused on the amelioration of pain as a primary end point.4–6 Given limitations in funding and technical challenges with study design, there are fewer studies that focus on outcomes that truly matter to patients: the real-world benefits of neurostimulation and advanced brain-computer interface technology for restoration of physical function. Standard-of-care pain rehabilitation techniques often yield suboptimal outcomes with many treatments focused primarily on symptom management rather than functional restoration.
There is a clear need for innovative interventions that can enhance neuroplasticity, reconfigure central pain processing, reduce pain, prevent deconditioning, and improve functional recovery. Deconditioning, characterized by a decline in physical fitness and muscle function from inactivity or immobility, exacerbates pain and further hinders recovery in patients with neuromuscular impairments. This guideline addresses the need for a standardized, evidence-based approach to applying neuromodulation technologies for neuromuscular restoration and pain relief in this patient population.
This document has been developed with the clinician and clinical stakeholders at the forefront, although a general discussion on mechanisms and scientific basis is also included for each technology.
To that end, a multidisciplinary panel of experts was assembled to review and assess quality of literature and provide practical clinical recommendations. The aim was to provide a comprehensive, evidence-based framework for healthcare professionals and policymakers, including physicians, advanced practice providers, physical therapists, psychologists, payers, and neuromodulation specialists, to assist in selecting, implementing, and optimizing neuromodulation techniques for neuromuscular restoration and pain relief.
Methods
The American Academy of Pain Medicine Foundation (AAPMF) convened an interdisciplinary panel (Supplementary Material S1), which met in Austin, Texas, on April 2, 2025. Criteria for panel participation included knowledge or specific expertise in restorative neuromuscular function and/or advanced neuromodulatory technologies related to neuromuscular function. Panel selection reflected a range of professional disciplines: anesthesiology, physical medicine and rehabilitation, addiction medicine, neurosurgery, orthopedic rehabilitation, physical therapy, pain medicine, material science, and biomedical engineering.
The screening process for panel members included a review of their conflicts of interest as well as their relevant expertise. Financial conflicts of interest were disclosed and mitigated before initiation of the consensus guideline development process. Panel members signed a non-disclosure agreement to ensure the integrity of the in-person meeting and the consensus development process.
Before the in-person meeting, panel members TE, AMB, and FMA developed an initial set of clinical questions and search terms, focused on defining the scope and organization of the research. They created an initial, broad list of search terms, organized by both the technology modality and functional restoration outcomes. The team compiled these questions into an email survey sent to all panelists, who voted on the inclusion of specific technologies and identified additional questions to address; panelists were also given the opportunity to suggest additional technology categories.
Examples of these clinical questions were designed to guide the subsequent literature review, evidence synthesis, and panel discussions. The final set of clinical questions is presented in Table 1.
General questions were also considered across all interventions, including:
A risk-benefit analysis, if available, including reported adverse effects for intervention-based techniques, was examined to determine if benefits outweigh potential harms.
Peer-reviewed journals were searched in PubMed (Medline); subsequently, to ensure a comprehensive capture, supplemental searches were performed in the Cochrane Library, Web of Science, and Scopus databases. The search strings, Boolean logic, field-limiting, tags, and post-search citation screening strategies for these databases are detailed in Supplementary Material S2. Inclusion criteria were English-language case reports, clinical studies, clinical trials, guidelines, meta-analyses, observational studies, practice guidelines, randomized controlled trials (RCTs), reviews, and systematic reviews in humans published between 2015 and 2025 for the following categories of interventions:
Selection of technology categories and associated topics was based on several factors: panelist expertise and familiarity, clinical relevance and novelty, and potential for future influence in clinical pain medicine. Despite a varying degree of evidence across categories, all were deemed relevant for the panel’s consideration. During manuscript review, edits to the subcategory titles were made to facilitate a coherent flow of the manuscript, and several of the above categories were combined.
Preparation for the in-person panel meeting involved two phases. First, panel members were tasked with reviewing and screening the literature search results, which were uploaded into Covidence (Veritas Health Innovation, Melbourne, Australia), a review software. Each panelist was assigned a subset of abstracts to independently screen for relevance to the guideline topics, according to predefined criteria.
The aim was to refine the literature set and familiarize panelists with pertinent research, particularly in areas outside their primary expertise. Following the initial literature review, panelists independently reviewed reference lists of retrieved articles and identified pertinent studies for inclusion.
Second, panelists were subdivided into four subgroups and assigned topics (eg, PNS, BCIs) for which they completed standardized grids to facilitate the extraction and synthesis of key information, including pivotal references, primary findings, and relevant outcome metrics. Given the breadth of interventions investigated, subgroups were aligned to include those panelists with particular expertise in specific technologies. The compiled information from these grids served as the foundation for subsequent discussions.
During the eight-hour in-person meeting, subgroups deliberated findings then presented a summary of conclusions and insights to the entire panel for broader discussion and consensus building.
Panelists evaluated literature using the United States Preventive Services Task Force (USPSTF) criteria7 as a general framework to evaluate the quality of evidence with modifications for neuromodulation studies (Table 2) and assigned “level of certainty regarding net benefit” as described in Table 3. This approach was previously used in practice recommendations published by the Neurostimulation Appropriateness Consensus Committee.8
The panel followed a Delphi process to review evidence, gain expert consensus, and create guideline recommendations in keeping with methodology described in Benzon et al.9 Panel-defined terms are shown in Table 4. The panel met virtually in August 2025 to review guideline progress and clarify the Delphi process.
Payer coverage patterns and reimbursement considerations for various neuromodulation technologies were obtained via a targeted, non-systematic internet search of publicly available information.
Regarding bias mitigation, the literature search, article selection, evidence review, and recommendation development processes were conducted independently of all industry funding. Industry sponsors had no role in literature selection, evidence grading, recommendation development, manuscript drafting, Delphi voting, or peer review, and did not have access to manuscript drafts prior to submission.
All panelists disclosed relevant financial and non-financial relationships prior to participation. Conflicts of interest were reviewed by American Academy of Pain Medicine (AAPM) leadership prior to panel selection. During the Delphi consensus process, recommendations were developed through structured independent voting by panel members.
Final recommendations required consensus across the multidisciplinary panel, and no single panelist could independently determine recommendation strength or evidence grading.
Recommendation grades were based on the totality of available evidence, including prospective studies, mechanistic rationale, functional restoration outcomes, durability data, safety considerations, and real-world clinical experience, using modified USPSTF criteria and Delphi consensus methodology. The panel acknowledges that for several emerging restorative technologies, large-scale non-industry-funded head-to-head replication trials remain limited, reflecting the evolving nature of the field.
Table 5 shows outcome measures reported across the reviewed literature, providing essential clarity for clinical interpretation. Additional tables summarize the evidence (Table 6), clinical applications (Table 7), and the panel’s recommendations (Table 8). While Table 8 provides a formal summary of the certainty of evidence and strength of recommendations, these are further referenced qualitatively within the narrative text to maintain clinical context and readability.
The following section applies this organizational framework to the panel’s consensus findings across the reviewed technologies.
Results/discussion
Panel participation and consensus
Thirteen panelists completed the in-person meeting and longitudinal panel work. Twelve out of 13 panelists (92.3% response rate) completed two iterative rounds of voting. Consensus was predefined as a >70% threshold of agreement; all statements reached this threshold by the conclusion of the second round.
Between rounds, statements and their respective evidence grades were refined using panelist feedback to improve clarity and accuracy. Final consensus statements, including Quality of Evidence and Level of Certainty rankings, are summarized in Table 8. Payer coverage patterns and reimbursement considerations are summarized in Supplementary Material S3.10–37 The following subsections synthesize the evidence and clinical considerations, concluding with the panel’s recommendations for each technology.
Artificial intelligence-guided and robotic rehabilitation systems
Mechanisms of action
While AI-enabled technology and robotic systems in medicine have existed for decades, advanced, state-of-the-art AI-guided and robotic rehabilitation systems are emerging as a new frontier in functional restoration, applied chiefly to patients with neurological conditions like stroke, SCI, Parkinson’s disease, and cerebral palsy.38–40 These systems are designed to enhance, restore, or replace neuromuscular functions by (1) directly activating nerves and muscles to produce a specific movement or sensation or (2) leveraging neuroplasticity through high-intensity, repetitive, task-specific training to improve central circuit function.38 Adaptive algorithms can adjust assistance levels in real-time, adapting to a patient’s performance. Many systems also integrate VR to make therapy more engaging and motivating,41,42 while others use functional electrical stimulation (FES) to induce muscle contractions to help prevent muscle loss, preserve neurological circuits, and achieve movement.43 Machine learning, a subtype of AI, can help robotic systems learn from available data and make beneficial changes based on this information.39 AI can facilitate rapid processing of complex data for improved user experience as well. One example of a robotic rehabilitation system is the Hybrid Assistive Limb (HAL) system.44 This is an electromyogram (EMG)-controlled wearable exoskeleton system that utilizes surface EMG and combines voluntary movements with robotic support.
This device can lead to functional improvement in ambulation even outside of use of the HAL device.
Evidence review
Evidence on the effectiveness of these systems is mixed, complex, and evolving. Careful patient selection and continued research are necessary.
In upper-limb rehabilitation for stroke, robot-assisted arm training may improve ADLs, arm function, and strength, but the results should be interpreted with caution due to variations in study design.45 Some multicenter trials found that this type of training did not significantly improve upper-limb function compared to usual care,46–48 leading to the conclusion that it is not yet supported for routine clinical practice in some settings.47 Nevertheless, a specific EMG-triggered robotic system, the RETRAINER system, combined with FES showed promise in restoring arm function,43 and error-augmented bimanual therapy has been used for stroke survivors.49
In lower limb rehabilitation, robot-assisted gait training is a key application for both stroke and SCI.50,51 A systematic review found that while gait training can be effective, walking speed may not be the most appropriate outcome measure for patients with incomplete SCI.50 The use of exoskeletons, such as the HAL system, has demonstrated feasibility and safety in improving mobility for patients with both complete and incomplete SCI.51 Additionally, a robotic ankle-foot orthosis has shown promise for patients with foot drop.52
Emerging research focuses on robotic therapy in combination with other interventions to enhance outcomes. Systematic reviews have examined combining robotics with VR for lower-limb function in stroke survivors53 and with noninvasive brain stimulation for motor function in individuals with incomplete SCI.42 Additionally, a recent Cochrane systematic review evaluating VR for multiple sclerosis rehabilitation showed no clear benefit over no intervention for gait or posture.54
Further research is needed to provide more conclusive evidence on the effectiveness of these systems across different patient populations38–40 and the future role of AI in the use of biohybrid robots.55 Biohybrid devices combine living tissue, such as contractile cells, with synthetically-engineered materials and control systems.55
Clinical considerations
These systems promote functional recovery by driving neuroplasticity through enabling repetitive practice that helps the brain reorganize and regain control over muscles. A key concept with AI-guided and robotic rehabilitation systems is the prevention of muscle atrophy and preservation of the neuromuscular synapse. Another key benefit is their ability to deliver high-intensity, personalized therapy.
The high cost of robotic systems and inconsistent insurance coverage are barriers to widespread clinical adoption, limiting patient access to these technologies.38 Other limitations include fatigue related to the bulky and cumbersome nature of some rehabilitation technologies and to a low frame per second rate that can lead to poor performance and patient safety issues.56Recommendation: AI/Robotics Statement 1: AI-Guided and robotic rehabilitation systems for upper and lower limb rehabilitation may provide meaningful functional restoration (Grade C; certainty: low). Recommendation: AI/Robotics Statement 2: AI-Guided and robotic rehabilitation systems for upper and lower limb rehabilitation may provide clinically significant pain relief (Grade I; certainty: low).
Virtual reality/augmented reality interfaces
Mechanisms of action
VR/AR interfaces utilize alternative environments to enhance rehabilitation and functional recovery by leveraging neuroplasticity through engaging, repetitive, and feedback-rich experiences.57 These technologies create controlled, immersive surroundings thought to enhance neural reorganization and formation of new neural pathways.58 By integrating realistic visual, auditory, and proprioceptive feedback, VR/AR technology facilitates multisensory integration, which is essential for motor learning and coordination. The gamified platforms boost patient motivation and engagement, leading to higher adherence to therapy protocols.59,60 Furthermore, these interfaces enhance motor planning and execution by enabling patients to visualize and perform goal-oriented tasks, strengthening the link between cognitive intent and physical movement.61 The real-time feedback provided allows for immediate guidance and correction, which improves movement precision and coordination.
Evidence review
VR/AR therapies have demonstrated efficacy in improving functional status and managing pain. For chronic low back pain, VR-based training has been shown to reduce pain and fall risk and improve quality of life.62,63 In stroke rehabilitation, VR-supported exercise therapy is an effective tool for upper extremity motor rehabilitation.64 A Cochrane review found that VR-based rehabilitation, when combined with conventional therapy, can lead to earlier functional improvements.65 VR therapy has been found to be more effective than minimal interventions in enhancing static and dynamic balance and gait performance in stroke patients and older persons.66,67 Individual RCTs also provide support. The RHOMBUS II trial60 and other studies57,68,69 have investigated the use of virtual gaming and immersive VR for upper limb and gait rehabilitation in acute and subacute stroke patients.
In the context of musculoskeletal pain, studies have investigated VR for chronic neck pain,70 and an interactive AR-based telerehabilitation trial is evaluating its effectiveness for adhesive capsulitis.71 The VR-SPIRIT trial is a notable RCT for pediatric patients with congenital cerebellar diseases, examining the role of VR in improving social prediction and intensive rehabilitation training.72 Another trial is exploring the combined effect of VR and repetitive rTMS for children with spastic cerebral palsy.73 Beyond motor function, the immersive nature of VR can act as a powerful distraction, effectively reducing pain perception and managing fatigue, thereby enabling patients to engage more fully in therapeutic exercises.74
Clinical considerations
While VR/AR technologies hold great promise, several clinical considerations must be addressed. The primary adverse events are related to cybersickness, which can include nausea, dizziness, and disorientation. These effects are typically mild and can be mitigated by adjusting the virtual environment or limiting session duration.
Risks such as falls are managed through supervised use and careful environmental setup. From a functional perspective, these technologies promote functional recovery by enhancing the brain’s capacity for neuroplasticity and reorganization. However, as is also true of robotics, a significant barrier remains in insurance and payer coverage, as dedicated VR/AR platforms often lack specific reimbursement codes.
Future research should focus on determining optimal protocols, developing personalized and adaptive systems, and conducting large-scale, long-term RCTs to establish definitive evidence of efficacy and cost-effectiveness.Recommendation: VR/AR Statement 1: VR and AR therapy is efficacious for motor and gait rehabilitation in individuals after a stroke (Grade B; certainty: moderate). Recommendation: VR/AR Statement 2: VR and AR therapy may be considered for patients with chronic pain, after weighing considerations such as diagnosis, cost, access, and potential side effects (Grade B; certainty: high).
Brain-computer interfaces
Mechanisms of action
BCIs enable direct communication pathways between the brain and an external device. The fundamental principle involves detecting and interpreting brain signals, typically from the sensorimotor cortex, to control external devices or provide neurofeedback. BCI systems can facilitate both neural rehabilitation/functional improvements and provide assistive benefits to patients with loss of function.75 By providing real-time neurofeedback and rewards, BCI systems can improve user performance in both BCI accuracy and limb function.76,77
Different BCI systems use various methods for signal acquisition and processing:
BCIs can enhance neuroplasticity by influencing motor learning, promoting increased attention to a motor task, reinforcing desired motor behaviors, or requiring the activation or deactivation of specific brain signals.76
Evidence review
BCI technology has shown promise in various neurorehabilitation contexts, particularly for restoring muscle function in people with severe paralysis. A meta-analysis of 16 articles with 382 participants showed that BCI training based on brain activity induced motor recovery after stroke.75 Furthermore, a recent double-blinded RCT demonstrated that motor imagery-contingent BCI training in patients with chronic stroke was significantly more effective for upper limb functional recovery and neural plasticity compared to BCI feedback that was not contingent on EEG-detected motor activity.86 Systematic reviews and other literature support the potential of BCI systems to promote functional recovery by providing real-time neurofeedback, making EEG-based BCI interventions a promising approach for stroke rehabilitation.76,87,88 Applications of BCIs in stroke rehabilitation have been summarized by Yang et al,87 as follows:
For rehabilitation in advanced-stage ALS, EEG-based BCI methods, P300-based, motor imagery-based, and hybrid BCI, have shown promise in providing communication and control.78 BCI-controlled exoskeletons are also being explored. However, challenges remain regarding the accuracy and reliability of non-invasive BCI systems and the inconsistent effect sizes of BCI training on motor recovery across studies.90
The effectiveness of BCI interventions is assessed through:
There is information available on the safety profile of BCIs. A systematic review of BCIs for stroke patients (especially upper limb motor function) has shown a favorable safety profile.96 The Stentrode With Thought-Controlled Digital Switch (SWITCH) Study, a single-center, prospective, first-in-human study, evaluated the safety of a fully-implanted endovascular BCI for severe paralysis.80 The study included four patients with ALS and one with primary lateral sclerosis and reported no serious adverse events, vessel occlusion, or device migration over a 12-month follow-up. The BrainGate Neural Interface system has a favorable safety profile with no unanticipated adverse device events or device explantations due to safety events.
Subdural grids carry risks, including neurological infection, intracranial hemorrhage, and increased intracranial pressure requiring revision surgery.97 Endovascular BCIs, which access the sensorimotor cortex via blood vessels, provide a potentially safer minimally-invasive alternative to open-brain surgery for electrode placement. Moving beyond a specific patient’s safety, a systematic review has also examined the broader risks of BCI use, such as legal, societal, and other physical health concerns.98
Clinical considerations
BCIs hold significant potential for restoring communication and motor function in individuals with severe paralysis resulting from conditions such as stroke or ALS. Non-invasive EEG-based BCIs offer potential avenue for rehabilitation, while implanted BCIs, are advancing as viable options, especially for those with severe paralysis. The ability of BCIs to leverage and enhance neuroplasticity suggests a crucial role in future neurorehabilitation strategies.
However, the field continues to address challenges related to accuracy, reliability, and consistency of outcomes, which can be highly variable among individuals.90 Non-invasive motor BCI systems, such as brain/neural exoskeletons, have accuracy and reliability rates ranging from 65% to 80%; these rates may be insufficient, for assistive use—especially compared to implantable BCIs—and reflect inconsistent effect sizes of BCI training on motor recovery across studies, with standardized mean differences ranging widely from 0.16 to 1.20.90 Further research is needed to refine BCI technology, improve its consistency, and identify patient populations most likely to benefit from these interventions to ensure their widespread clinical adoption. A main limitation of this technology is the lack of human clinical trials given the associated costs and complexities of implementation.Recommendation: BCI Statement 1: BCI may be considered for motor rehabilitation after stroke or paralysis and must be balanced with patient-specific factors, anticipated benefits, and risks of the procedure (Grade C; certainty: low). Recommendation: BCI Statement 2: BCI may be considered for pain relief after stroke or paralysis (Grade I; certainty: low).
Electrical nerve stimulation
Broad categorizations for devices within this heading include PNS, muscle stimulation (eg, FES), TES, and VNS.
Peripheral nerve stimulation/transcutaneous electrical stimulation/functional electrical stimulation
Mechanisms of action
PNS devices are used for functional restoration through several key mechanisms.99 These devices stimulate neural pathways to promote neuroplasticity,99,100 which is crucial for motor recovery after central nervous system injuries like stroke.101,102 PNS can also directly activate motor neurons, causing muscle contractions and facilitating movement. This mechanism helps compensate for impaired neural pathways to restore functional mobility.103,104 PNS also modulates pain and nociceptive inhibition,105 a process where it activates inhibitory pathways to reduce chronic pain perception.106–109 Furthermore, PNS can be used to provide proprioceptive feedback, which enhances balance and coordination,110 and can also promote nerve and tissue regeneration to prevent muscle atrophy and maintain structural integrity.111,112
TES, including TENS, and transcutaneous spinal cord stimulation (tSCS), involves the delivery of electrical current through the skin to modulate neural activity. For pain modulation, TENS is generally understood to work by stimulating nerve fibers, including A-alpha, A-beta, and C fibers, which can activate descending inhibitory pathways and block pain signals. For structures like the spinal cord or deeper muscles, advanced TES devices deliver programmable electrical current waveforms designed to permit high-amplitude stimulation without discomfort.
These waveforms can consist of modulated frequencies (eg, base frequency and overlapping frequency) on multiple independent channels. This allows for stimulation over the skin to reach the spinal cord dorsal roots, thereby activating spinal networks.113 The precise mechanisms of tSCS are complex, potentially involving the activation of motor pools via dorsal roots or interneurons, which can lower their activation threshold by residual supraspinal input.114 Different patterns of tSCS can be employed, including single pulses, trains of pulses, and waveforms, with or without carrier frequencies. Many studies aimed at improving functional recovery after SCI have utilized alternating current tSCS.114 Stimulation currents often involve a carrier frequency (eg, 2.5-10 kHz) with a burst frequency (eg, 30 Hz) or isolated pulse protocols with frequencies between 5 and 90 Hz and pulse widths between 0.5 and 2.0 ms.115
Evidence review
PNS devices have been used for decades to improve function across various pathologies. The field of PNS for functional restoration is well-established, with a long history of application in diverse forms.116 Pivotal trials have demonstrated its safety and efficacy in enhancing motor recovery by inducing cortical reorganization and increasing muscle mass.101,102,117
PNS reduces pain and improves functional capacity in conditions like chronic low back pain and chronic shoulder pain. The mechanism involves modulating pain pathways and increasing the release of endogenous opioids reducing pain perception and interrupting chronic pain signaling.105,106,109,118 A recent study on shoulder pain’s real-world outcomes using 60-day temporary percutaneous stimulation (targeting the suprascapular nerve, axillary nerve, or brachial plexus) found that 83% of patients avoided permanent implant surgery, shoulder surgery, or radiofrequency ablation over a mean 21-month follow-up (up to 5 years).119
PNS is also used to restore neuromuscular dysfunction.110 A small pilot RCT showed that 2-week temporary PNS of the axillary nerve can provide post-stroke patients with hemiplegic shoulder pain meaningful improvements in shoulder biomechanics similar to results from physical therapy.120 FES has been used to prevent muscle degradation and improve gait in patients with diabetic peripheral neuropathy and to recover lower limb function in Conus-Cauda Syndrome.112 While electrical stimulation can be used to promote axonal regeneration and functional recovery after peripheral nerve injury, this technique typically involves local field stimulation applied to the injured tissue.121
Electrical stimulation modalities such as FES and tSCS are used to reduce spasticity and enhance motor control.103,104,111 A systematic review found that tSCS improves lower limb function and reduces spasticity, aiding in mobility restoration.111 One notable study demonstrated that targeted neurotechnology can restore walking in some individuals with SCI.122
The applications explored for TES techniques include pain management and restoration of motor and autonomic functions. Following SCI, tSCS has shown positive results in restoring hand and arm function.113 A systematic review of tSCS studies reported evidence of induced muscle activation in both lower and upper limbs.115 Stimulation was typically applied at the T11-T12 and C4-C7 interspinous spaces, respectively, leading to increased motor responses measured by surface EMG, voluntary movement, muscle strength, or function. These studies reported increases in the amplitude of spinal reflexes, EMG activity during active movement of various limb muscles (eg, quadriceps, hamstring, triceps surae, tibialis anterior, extensor digitorum, flexor digitorum), and coactivation of trunk muscles (erector spinae and external oblique).
To evoke adaptive neuroplasticity, tSCS interventions are most effective when combined with activity-based rehabilitation, such as movement, strength, coordination, assisted gait, transfer training, or manual dexterity and grip strength exercises. Patient positioning during stimulation can also influence the threshold and magnitude of the motor response due to improved access to proprioceptive afferents.
A study protocol for a randomized crossover trial is aimed at investigating the effects of TENS on pain, function, and descending inhibition in individuals with non-specific chronic low-back pain.123 A multicenter, randomized, double-blind, controlled trial of TENS for pancreatic cancer-related pain found that TENS not only helped with pain but also improved appetite and constipation.124 TENS has also been shown to reduce movement-evoked fatigue.
To improve gastrointestinal (GI) motility, TES is being investigated in disorders like dysphagia, gastroesophageal reflux disease, gastroparesis, functional dyspepsia, chronic intestinal pseudo-obstruction, constipation, and irritable bowel syndrome.125 TES for GI motility is typically delivered for less than 60 minutes per session, once every other day to twice daily.
Preliminary evidence suggests that peripheral TES may have positive effects on specific cognitive functions in healthy individuals and those with mild cognitive impairment.126 For treatment of spasticity due to stroke or cerebral palsy, a self-administered TES concept using a garment with integrated electrodes (Mollii) has been evaluated in an RCT.127
The effectiveness of TES is evaluated using a variety of measures as suits the targeted condition:
Clinical considerations
In patients with neural paralyses or damaged/injured end organs, electrical stimulation devices can replace sensory and motor function. Electrical stimulation is also used in conjunction with orthoses, prostheses, and exoskeletal devices to restore and augment function. Electrical stimulation can be applied to the surface of skin using surface electrodes to target superficial nerves and muscles, while implanted electrodes are beneficial for more specific activation and access to deeper structures.
PNS therapy is a multi-faceted approach to functional improvement.108,128,129 It enhances neuroplasticity and motor control, modulates pain pathways, and improves proprioceptive feedback and motor coordination. It can also reduce spasticity and enhance cognitive engagement when paired with interactive tasks. For patients in whom at least partial nerve and muscle function remain, electrical stimulation promotes the maintenance and gain of muscle mass and functional recovery through high-intensity, repetitive practice, which helps the brain reorganize and regain control over muscles.
TES offers a non-invasive and versatile approach to neuromodulation with growing clinical applications. While TENS remains a widely-used modality for pain management, newer frontiers in tSCS demonstrate potential for functional restoration, particularly in people with SCI. Emerging roles for TES include managing GI motility disorders, improving cognitive function, and addressing spasticity.
The ability to customize stimulation patterns and combine TES with activity-based rehabilitation suggests potential for personalized and integrated treatment strategies.
The effectiveness of tSCS for neuroplasticity and motor recovery is enhanced when combined with activity-based rehabilitation and with proper patient positioning. While promising, the field is characterized by diverse stimulation patterns and parameters, necessitating further research to standardize optimal protocols for specific conditions. The risks of TES are generally low, with the main considerations being proper device placement and patient selection to maximize therapeutic benefit.
Despite the potential for moderate to substantial net benefit for patients with motor or autonomic impairment and published systematic reviews showing benefit, this field is still emerging with a low-to-moderate certainty regarding net benefit.Recommendation: PNS Statement 1: PNS therapy for motor rehabilitation may provide benefit through improving motor control after stroke and primary joint instability (Grade C; certainty: moderate). Recommendation: PNS Statement 2: PNS therapy is clinically proven to help patient’s pain with specific chronic pain diagnoses, such as shoulder pain (Grade A; certainty: moderate). Recommendation: TES Statement 1: Transcutaneous spinal cord stimulation (tSCS), a form of TES, is an emerging technology that may provide meaningful functional recovery to patients after spinal cord injury (Grade B; certainty: moderate).
Recommendation: TES Statement 2: Although TENS (Transcutaneous Electrical Nerve Stimulation) therapy remains a common therapy for chronic pain, there is emerging evidence that tSCS may also provide meaningful relief for patients with chronic pain (Grade C; certainty: low).
Vagus nerve stimulation
Mechanisms of action
The vagus nerve is a critical component of the autonomic nervous system, transmitting afferent information that influences homeostasis and modulates central and peripheral neural circuits. VNS leverages this pathway to exert therapeutic effects across various neurological conditions. The proposed neurobiological mechanisms are multifaceted:
VNS can be delivered invasively through surgically-implanted pulse generators or non-invasively via transcutaneous stimulation. The most common invasive approach involves stimulating the left cervical vagus nerve to minimize cardiac side effects. Non-invasive methods, such as transcutaneous auricular VNS (taVNS) and transcutaneous cervical VNS (tcVNS), have demonstrated comparable effects to invasive methods but are more non-specific.140,141
Evidence review
VNS has demonstrated therapeutic potential for a range of conditions, with established efficacy in epilepsy, depression, and headache disorders.133,142,143 Emerging evidence supports its use for motor restoration following neurological injury.
In the context of stroke, VNS has been investigated to aid recovery of motor function, particularly when paired with rehabilitation. VNS paired with rehabilitation has shown significant improvement in upper limb function.100–102 A pivotal Phase III trial (VNS-REHAB) demonstrated that invasive VNS combined with rehabilitation led to significantly greater improvements in upper limb function in patients with long-term, moderate-to-severe arm impairment following ischemic stroke compared to rehabilitation alone.101 Additionally, an RCT in patients with acute stroke showed that taVNS combined with conventional rehabilitation resulted in significant improvements in upper and lower limb motor function, sensory disturbance, and quality of life.144 Another study in subacute stroke patients also reported improved upper limb motor function and functional independence with taVNS.145 The use of VNS is also being explored in other neurological conditions such as Parkinson’s Disease, traumatic brain injury, MS, and Alzheimer’s Disease, with preliminary evidence suggesting a neuroprotective and restorative role.133
Clinical considerations
The evidence suggests that VNS holds significant promise as a therapeutic modality for motor restoration by promoting neuroplasticity and enhancing recovery. The emergence of non-invasive transcutaneous approaches (eg, taVNS and tcVNS) offers a safe, well-tolerated, and potentially scalable alternative to invasive VNS, which carries surgical risks and financial burdens.141 Ongoing research is focused on optimizing stimulation protocols, distinguishing between responders and non-responders, and further elucidating the long-term efficacy and mechanistic consequences of prolonged stimulation.132,146,147 Translational clinical trials are critical for evaluating the therapeutic benefits across different stages of neurological injury and validating VNS as a mainstream tool for neuromuscular restoration.Recommendation: VNS Statement 1: VNS may promote neuroplasticity and help patients with motor rehabilitation after stroke (Grade C; certainty: moderate). Recommendation: VNS Statement 2: VNS for pain management is an emerging area of research and may provide some benefit to patients with chronic pain (Grade C; certainty: low).
Multifidus neurostimulation
Mechanisms of action
Multifidus muscle restoration, specifically through peripheral motor nerve stimulation of the lumbar medial branch nerves using neurostimulation (eg, ReActiv8 by Mainstay Medical, currently the only FDA-approved device for this purpose), addresses a primary pathophysiologic mechanism underlying chronic mechanical axial low back pain: Spinal instability driven by multifidus dysfunction.148–152 The multifidus muscle is crucial for segmental spinal stabilization. Its persistent inhibition results in atrophy or fatty infiltration, which contributes to impaired neuromuscular control, leading to increased biomechanical stress, chronic and recurrent low-back pain, and disability.153,154 Neurostimulation re-engages the multifidus muscle through low-frequency electrical stimulation of the dorsal ramus of bilateral L2 medial branches.154,155
This targeted activation is hypothesized to:
The therapeutic effect is considered disease-modifying and is aimed at restoring proper neuromuscular control and spinal stability over time.158,159
Evidence review
Evidence for multifidus neurostimulation is supported by high-quality RCTs and long-term cohort studies. The RESTORE RCT confirmed the superiority of neurostimulation over optimized medical management in reducing disability and pain in patients with multifidus-driven axial low back pain.148,160 Clinical studies have demonstrated that restoring multifidus function leads to meaningful patient-reported improvements in pain, physical function, quality of life, and decreased reliance on healthcare resources.151,153,158,161 Long-term studies show these benefits are durable, with sustained improvements reported from one-to-five-year follow-up intervals,158,159,161–166 including in older people traditionally considered high-risk for surgery.167 A recent systematic review and meta-analysis by Copley et al168 reinforces the efficacy of medial branch stimulation.
Anatomic and mechanistic studies suggest that targeted activation of the multifidus can reverse deleterious structural changes.156,157,169 This is clinically relevant for patients with a history of lumbar spinal surgery, where improved neuromuscular function, ADLs, and quality of life have been reported.170,171 A consecutive case series in patients with chronic low back pain after prior lumbar spinal surgery reported early outcome profiles similar to those without previous surgery, showing serial reductions in pain and disability scores over 12 months.170 Furthermore, a 5-year longitudinal study demonstrated that neurostimulation provides robust and sustained improvements in pain, function, and quality of life for patients with chronic mechanical low back pain associated with multifidus dysfunction, supporting its long-term real-world efficacy.172 A case report also highlights the use of neurostimulation in a patient with a large disc herniation.173 Furthermore, a clinical care pathway has been described to ensure optimal outcomes.171,174
Consensus statements151,175 and a scoping review by Tieppo Francio et al155 underscore the importance of multifidus muscle dysfunction as a driver of chronic mechanical low back pain. By synthesizing clinical, imaging, and electrophysiologic evidence, they further validate the use of therapies that target this neuromuscular deficit.155
Clinical considerations
For diagnosing multifidus dysfunction and identifying candidates for multifidus neurostimulation, a comprehensive multimodal evaluation should integrate patient-reported outcomes with objective physiological and imaging biomarkers.176,177 Patient-reported outcome measures (PROMs) are considered the gold standard for quantifying functional impact and monitoring treatment response. These include:
Diagnosis of multifidus dysfunction relies on a clinical profile, physical exam, and imaging. Patients typically present with axial, mechanically-driven low back pain lasting >6 months, without radicular or neuropathic features. Pain is often aggravated by transitional movements (eg, sit-to-stand) and/or trivial loading activities (eg, brushing teeth) and improves with external spinal support.
Physical exam employs validated tests to provide objective evidence of impaired neuromuscular control and may include:
Objective physiological and imaging biomarkers are under investigation. While current studies on neurostimulation have yet to directly quantify multifidus muscle changes post-treatment, several modalities are being explored. These include:
According to high-level evidence from the RESTORE RCT and long-term studies, ideal candidates for neurostimulation148,159,167:
Exhibit evidence of multifidus dysfunction on clinical exam and/or imaging
Fatty infiltration of the multifidus muscle is a structural biomarker of chronic spinal deconditioning and neuromuscular dysfunction, consistently associated with longer symptom duration, segmental instability, and degenerative spinal changes.179,180 However, the relationship between structural muscle changes and clinical symptoms is not always linear. Multifidus dysfunction is primarily a functional impairment of neuromuscular control, not solely a structural abnormality.155 Fatty infiltration and atrophy may indicate disuse or chronic overload but do not reliably predict pain presence or severity, functional disability, or treatment outcomes.181 While MRI is a crucial imaging tool for confirming atrophy and fatty infiltration and for ruling out other surgical pathologies, it is not diagnostic in isolation and its findings should always be integrated with comprehensive clinical and functional assessments. Functional improvements, particularly with neurostimulation, are more reliably predicted by the clinical phenotype (axial mechanical pain without radicular features) and functional testing (eg, PIT, MLT) than by MRI alone.
Therefore, treatment decisions for neuromodulatory interventions should prioritize functional assessment of motor control over structural imaging alone. Because multifidus dysfunction is a dynamic and potentially reversible impairment, guidelines recognize neurostimulation as a viable and effective option, especially when other conservative treatments have failed.151,153,182 Adverse events in the RESTORE trial included pocket site pain (8.1%), device overstimulation of tissue (5.1%), lead fracture (3.0%), implant site seroma (2.0%), and implant site dermatitis (2.0%).148 In this study, 23 of the 31 adverse events had resolved by one year.Recommendation: Multifidus Neurostimulation Statement 1: Multifidus muscle stimulation is viable, safe, has a well-established mechanism of action, and improves functional outcomes for adults with refractory mechanical low back pain and evidence of multifidus muscle dysfunction with moderate-to-severe pain and disability (Grade A; certainty: high). Recommendation: Multifidus Neurostimulation Statement 2: Multifidus muscle stimulation may significantly reduce low back pain in adults with refractory mechanical low back pain and evidence of multifidus muscle dysfunction with moderate-to-severe pain and disability (Grade A; certainty: high).
Surgery, including regenerative peripheral nerve interface
Mechanisms of action
RPNIs are a surgical technique designed to improve advanced prosthetic control and prevent the development of painful neuromas. An RPNI is created by surgically implanting a transected peripheral nerve, or its individual fascicles, into an autologous skeletal muscle graft. As the axons regenerate, they reinnervate the muscle, which acts as a “bioamplifier.”183,184 This process converts microampere nerve action potentials into robust compound muscle action potentials (CMAPs) that can be reliably recorded using surface or epimysial electrodes.
Preclinical studies have shown that RPNIs produce large-amplitude signals with high signal-to-noise ratios, maintaining stability for over a year.183 This makes RPNIs well-suited for providing intuitive, multichannel control for advanced myoelectric prosthetic devices.
Evidence review
RPNIs have demonstrated efficacy in two key areas:
Clinical considerations
RPNIs offer an advance in functional restoration by providing amputees a biological bridge to sophisticated prosthetics183,184 and effectively treating neuroma pain.186–188 Surgically, RPNIs are easy to construct and require no implantable hardware. Skin flaps can create cutaneous mechanoneural interfaces to provide tactile sensations, while agonist-antagonist myoneural interfaces provide proprioception; both can be surgically constructed alongside RPNIs.190,191 Current research is focused on optimizing surgical techniques,189 integrating RPNIs with next-generation prosthetics, and exploring the potential for sensory feedback.183,184Recommendation: RPNI Statement 1: RPNI therapy may provide clinical benefit for select patient conditions and diagnoses, such as patients with neuromas or advanced prosthetic control (Grade C; certainty: moderate). Recommendation: RPNI Statement 2: RPNI therapy may provide clinically significant pain relief for specific patient conditions and diagnoses, such as prevention and/or treatment of neuroma pain (Grade B; certainty: moderate).
Scrambler therapy
Mechanisms of action
Scrambler therapy, a form of electrical stimulation therapy, is a non-invasive neuromodulation method that reorganizes maladaptive pain signaling pathways.192 Applied through cutaneous electrodes, it aims to alleviate chronic neuropathic pain by stimulating C-fibers.193 Unlike TENS which primarily stimulates A-beta fibers to activate the gate control mechanism, scrambler therapy replaces endogenous pain signals with synthetic non-nociceptive signals to mask the sensation of pain.194
During treatment, a clinician modulates the stimulation intensity, which is set above the patient’s perceptive threshold but below the patient’s pain tolerance as guided by direct verbal feedback from the patient.195 This modulation can be adjusted for each individual channel or electrode pair. The device typically delivers a maximum current of 5.5 mA with a stimulation intensity of 12.5 V. ST pulse width usually ranges between 6.8 and 10.9 milliseconds, with a frequency of 43 to 52 Hz.195 Scrambler therapy utilizes a much larger pulse width compared to TENS, which typically uses pulse widths of <250 microseconds.194
Evidence review
Scrambler therapy received 510(K) clearance from the FDA in 2009 for treating acute, postoperative, and chronic intractable pain.195 Clinical trials and meta-analyses have investigated its efficacy. A pilot RCT by Marineo et al196 reported a significant pain reduction in the scrambler therapy group compared to the control group, with mean VAS scores dropping from 8 to 0.7 points (91% reduction), compared to a reduction from 8.1 to 5.8 points (28% reduction) at one month (P < .0001). These differences were maintained at two and three months, with mean pain scores in the scrambler group at 1.4 and 2 points, respectively, versus 5.7 and 5.9 points in the control group (P < .0001).
A meta-analysis by Jin et al197 concluded that scrambler therapy appears to be effective for chronic pain but emphasized the need for further large RCTs. There is also a lack of long-term (≥12-month) RCT data that assesses efficacy and durability of results.
Case and series reports support its use in specific chronic pain conditions, such as diabetic peripheral neuropathy pain,198 pain associated with multiple system atrophy-Parkinsonian subtype,199 and complex regional pain syndrome.193 Most studies describe treatments as once-a-day sessions, typically lasting 30-45 min, administered over 10-to-14 consecutive days. A full 10-to-14-day treatment protocol is recommended when meaningful pain reduction is achieved.195 The primary outcome measure used in the reviewed studies for assessing the effectiveness of scrambler therapy is the VAS self-reported measure of pain intensity.196
Scrambler therapy demonstrates a favorable safety profile with few reported adverse effects. The original 2297 cases reported for European Euromark certification for safety indicated no substantial side effects.194 A systematic review by Wang et al,200 encompassing 1152 total patients, reported only two cases of contact dermatitis and one case of minor ecchymosis, with no serious adverse events. Another systematic review of 21 studies, mostly on the use of scrambler therapy for the treatment of neuropathic pain, reported only one adverse effect: contact dermatitis.195
Clinical considerations
Scrambler therapy presents a possible non-pharmacologic option for the management of chronic pain, particularly neuropathic pain. Its non-invasive nature and favorable safety profile make it an attractive alternative or adjunct to conventional pain management strategies. The evidence from pilot trials and meta-analyses suggests significant pain reduction, which can lead to improved functional status.
While initial evidence is encouraging, the field would benefit from larger, well-designed RCTs to definitively confirm its efficacy and establish optimal treatment protocols. Further research could also explore its long-term effectiveness and comparative effectiveness against other advanced pain interventions.Recommendation: Scrambler Statement 1: Scrambler therapy may provide meaningful clinical pain relief for specific diagnoses, such as complex regional pain syndrome or diabetic peripheral neuropathy (Grade C; certainty: moderate). Recommendation: Scrambler Statement 2: Scrambler therapy for neuromuscular functional restoration is an emerging area of clinical research and may provide meaningful improvement for patients (Grade I; certainty: low).
Spinal cord stimulation
Mechanisms of action
SCS for motor restoration utilizes implanted epidural electrodes to deliver electrical neurostimulation, modulating excitability within spinal and supraspinal networks and facilitating the transformation of dysfunctional neural networks into higher functional states.201 This neuromodulation can lead to motor restoration by reactivating dormant motor pathways and promoting the formation of new circuits.201 Primary anatomical targets of epidural stimulation include large diameter dorsal root afferents, dorsal column interneurons, ventrolateral white matter tracts, and intermediate zone and anterior horn cells over the cervical and lumbar enlargements.
The underlying mechanisms include network reorganization, where stimulation restores communication between surviving neurons,202 and neuroplasticity.203,204 By amplifying weak corticospinal signals, SCS can also enable descending motor control to make voluntary movement more effective in patients with SCI.122 By promoting sprouting, enhancing signal integration, and driving long-term plasticity, these mechanisms can help patients achieve a shift from movements reliant on stimulation to partially independent voluntary control, persisting even after stimulation is ceased. Lastly, SCS can facilitate the reactivation of central pattern generators (CPGs) in the lumbar spinal cord, which can generate rhythmic locomotor activity, enabling stepping movements even without direct supraspinal input.122,205 When integrated with proprioceptive and supraspinal inputs, this results in high-fidelity, smoother, and more natural movements.
Evidence review
SCS has been investigated for its potential to improve motor function in a variety of conditions. In stroke rehabilitation, SCS has been shown to enhance upper and lower limb motor function, particularly when combined with physical therapy.206–208 For SCI, SCS has demonstrated the potential to improve motor function and facilitate overground walking in individuals with both incomplete and motor-complete chronic SCI.122,205,209–211 SCS is also being explored for motor function impairment in MS and to address spasticity in cerebral palsy, although evidence is limited but growing.212 For Parkinson’s Disease, SCS may improve gait in patients with neuropathic pain, but more rigorous studies are needed to confirm its efficacy in pain-free individuals.213 Multiple studies have reported secondary findings of improved autonomic function in the long term, including volitional micturition and bowel continence, sexual function, blood pressure regulation, and improved body muscle composition.
The field is being advanced by AI and closed-loop systems. AI has been used to optimize treatment parameters, and algorithms can analyze complex neural data to predict optimal stimulation settings and adapt therapies to individual needs.214–217 Closed-loop SCS represents an advancement by using evoked compound action potentials to automatically adjust stimulation strength in real time.218–220
The efficacy of SCS is evaluated using objective measures, including216:
Clinical considerations
SCS represents an advancement in therapeutic neuromodulation for motor restoration. Most studies have utilized paddle-type electrodes implanted via a laminotomy, with lead wires tunneled to a subcutaneous implantable pulse generator. The integration of AI and closed-loop technologies holds promise for creating personalized, adaptive systems that can continuously adjust stimulation parameters using real-time feedback.221 These systems will be capable of maintaining motor output fidelity despite factors like muscle fatigue and variations in lead placement.222 While the development of highly accurate virtual frameworks to understand stimulation mechanisms remains a challenge, these technological advancements signal a shift toward more personalized, efficient, and effective therapies.Recommendation: SCS Statement 1: SCS for functional motor restoration can be used as a therapeutic option through activation of dormant circuits, neural network reorganization, and neuroplasticity (Grade C; certainty: low).
Recommendation: SCS Statement 2: Clinical utility of SCS for pain and functional improvement in patients with motor impairment may be considered in specific clinical situations as the physician balances the potential benefits and risks to the patient (Grade C; certainty: low).
Transcranial magnetic stimulation
Mechanisms of action
TMS is a non-invasive intervention that modulates brain activity by placing a coil on the scalp to generate a rapidly changing magnetic field. This field induces electrical currents that depolarize superficial axons, activating cortical networks.223 While TMS primarily targets axonal terminals located in the crown top and lip regions of cortical gyri and can also co-stimulate the peripheral nervous system,224 TMS does not directly activate subcortical structures like the thalamus or basal ganglia.223 The effectiveness of stimulation is influenced by the direction of the induced current in the motor cortex, which depends on the coil’s orientation and position relative to brain gyri and sulci.223
When applied over the motor cortex (M1), TMS induces descending volleys in the pyramidal tract, leading to muscle twitches recorded as motor-evoked potentials (MEPs) via EMG.223 MEPs are used to assess corticospinal tract excitability. The specific neurons activated depend on their geometry, myelination, spatial relationship to the electric field, and their physiological state, which is influenced by synaptic inputs and intrinsic membrane properties.224
Repetitive TMS (rTMS) can induce lasting changes in brain activity, a phenomenon thought to be related to long-term potentiation and long-term depression, which are mechanisms of synaptic plasticity.225 For instance, rTMS protocols can reduce GABAergic synaptic strength,225 shifting the excitation/inhibition balance within cortical networks. This facilitates or restores the capacity of stimulated areas to express associative synaptic plasticity that is specific to particular inputs or tasks.225 These activity-dependent changes, including calcium-dependent plasticity and spike-timing-dependent plasticity, are crucial for the brain’s ability to adapt and reorganize.226
While TMS pulses can be monophasic or biphasic, rTMS typically uses biphasic pulses due to their lower energy requirements.223 rTMS protocols involve delivering identical stimuli with consistent interstimulus intervals (ISIs) and are categorized by stimulation frequency223:
The effects of TMS on neuroplasticity and network connectivity are complex and influenced by various factors. Different rTMS protocols, such as TBS,227 are believed to regulate motor cortical excitability, induce neural network reorganization, and influence cerebral circulation metabolism.228 In conditions like stroke, rTMS is thought to improve motor function through mechanisms like the interhemispheric inhibition model, the vicariation model, and the bimodal balance-recovery model.228,229 Furthermore, “priming” protocols, which describe a session of stimulation before the main therapeutic session, can significantly affect the changes in corticospinal excitability and neuroplasticity induced by noninvasive brain stimulation techniques like TMS,230 highlighting the importance of precise stimulation parameters and individualized approaches.
Evidence review
TMS is a widely used tool for diagnostics, mapping, monitoring the motor system, research in cognitive neuroscience, and therapeutic interventions for various neurological and psychiatric disorders.231 First described in 1985 by Barker et al,232 TMS was positioned as a pain-free, non-invasive alternative to transcranial electrical stimulation for studying of the corticospinal tract.233 Initially used for diagnostic purposes, TMS evolved to probe motor cortex excitability,234 cognitive processes,235 and induce long-term changes in cortical excitability through rTMS236 and TBS.237 Key advancements include using focal TMS to map training-induced reorganization in stroke patients238 and pharmaco-TMS to characterize motor cortical excitability.239 Pascual-Leone and colleagues demonstrated that high-frequency rTMS can increase motor cortex excitability,240 that focal TMS can map practice-dependent plasticity,241 and that rTMS holds therapeutic potential for various brain disorders.242
TMS has emerged as a promising tool for restoring muscle function in various neurological conditions. In stroke rehabilitation, a large body of evidence from various study designs—including randomized controlled trials (RCTs), other trial designs, systematic and non-systematic reviews, and meta-analyses—mostly supports its effectiveness, especially when combined with conventional therapies.228,243–271 High and low frequency rTMS and TBS have demonstrated positive effects on limb motor recovery and ADLs.260–262 TMS has been investigated in multiple randomized RCTs,243–256,265,266,269,271 other trial designs,257,264,268 systematic228,258 and non-systematic259 reviews, and meta-analyses260–263,267,270 the majority of which support the intervention’s efficacy, particularly when combined with conventional therapies.
Fu et al228 found improved motor function in stroke patients, but called for more large-scale, high-quality, multicenter studies to clarify the mechanism of TBS and explore the optimal TBS treatment. Kubis259 reviewed studies showing that both high-frequency and low-frequency rTMS have various effects on functional disability and neurological scales. Cha and Kim243 found that rTMS combined with strengthening exercises improved motor function in subacute stroke patients more than exercise alone.
Du et al244 found that both high-frequency and low-frequency rTMS improved motor function in early stroke patients by modulating motor cortical activation.
Wang et al245 demonstrated that high-frequency rTMS over the contralesional cortex was superior to low-frequency rTMS or sham stimulation for promoting motor recovery in severe hemiplegic stroke. Chen et al246 showed that combined 1 Hz and 10 Hz stimulation had a synergistic effect on motor recovery. Valovičová et al247 found that 10 Hz ipsilesional rTMS improved motor function in the acute stage of ischemic stroke.
Lüdemann-Podubecká et al248 reported that 1 Hz rTMS over the contralesional premotor cortex improved hand motor function in subacute stroke. Kwon et al249 found that combined rTMS and tDCS resulted in better motor performance in stroke patients. Cho et al250 showed that both rTMS alone and rTMS + tDCS improved upper limb function in subacute stroke.
A few studies have reported mixed or negative results in stroke recovery: Xi et al263 found no significant difference in objective measures of upper limb function (Box and Block Test, Fugl-Meyer Assessment, modified Barthel Index); however the meta-analysis observed a significant improvement in MEPs, leading investigators to suggest that the neuroplasticizing effect of rTMS might translate into functional improvement through enhanced neuroelectrical signaling. Tarri et al reported that a five-day course of PAS did not improve motor function in stroke patients.272 Blesneag et al273 found no significant difference between rTMS and sham groups at three months post-stroke.
Overall, the reviewed studies suggest that TMS, particularly high-frequency rTMS, can be effective in promoting motor recovery after stroke. However, the optimal protocols and patient selection criteria continue to be refined.
In Parkinson’s Disease, a smaller body of literature suggests rTMS shows promise in improving motor symptoms, particularly freezing of gait and overall motor function.274–277 Studies highlight the benefits of both high-frequency and low-frequency rTMS. For example, Chang et al274 demonstrated that rTMS, with or without concurrent transcranial direct current stimulation, improved freezing of gait, motor function, and ambulatory function in Parkinson’s patients. Khedr et al275 found that while both 1 Hz and 20 Hz rTMS improved motor performance, 20 Hz rTMS yielded more sustained effects.
Chung et al276 further supported these findings, showing that both 1 Hz and 25 Hz rTMS improved gait speed in Parkinson’s patients, with sustained effects at three months. A systematic review by Nardone et al277 suggests that combining rTMS with treadmill training enhances physical therapy outcomes and points to potential future targets like the supplementary motor area and the use of H-coils for deeper brain stimulation. However, this review also noted the ineffectiveness of TBS for gait disturbances and called for larger, well-designed studies to optimize protocols and assess long-term effects.
In SCI, the evidence for rTMS is still emerging. Tazoe and Perez278 reviewed studies on rTMS for sensory and motor function impairments, spasticity, and neuropathic pain in SCI, finding mixed results across studies. They emphasized that rTMS parameters and post-injury reorganization in neural circuits are crucial for understanding its effects.
In a systematic review, Nardone et al279 highlighted TMS as a valuable diagnostic and research tool for assessing functional levels and monitoring changes after SCI. While initial studies suggest that rTMS may induce acute, short-duration beneficial effects on spasticity and neuropathic pain, the authors stress that this evidence is preliminary, and well-designed clinical trials are warranted. Case reports and reviews on PAS reported some encouraging preliminary data.
Shulga et al280 presented a case where PAS enabled and promoted walking rehabilitation in an individual with incomplete paraplegia. A subsequent review by Shulga et al281 on high-PAS, a novel variant combining high-intensity TMS and high-frequency PNS, observed functional improvements in a cohort of patients with incomplete SCI across various injury types and severities, indicating its potential for improving motor and functional independence. However, larger, robust studies are needed to confirm these preliminary findings.
The efficacy of TMS protocols is assessed primarily through MEPs and outcome measures that include improvement in motor function, ADLs, and neurological scales. However, confounding factors can influence results and interpretation in TMS studies. For example, the silent period, a period of EMG silence following an MEP during voluntary contraction, is thought to reflect intracortical inhibition, but both cortical and spinal mechanisms contribute to its generation.282
A study by Magnuson et al283 found that neither 1 Hz nor 10 Hz rTMS significantly modulated corticospinal excitability, and there was poor reproducibility across sessions for most rTMS protocols. This raises questions about the effectiveness and reliability of widely used rTMS techniques.
Clinical considerations
TMS, particularly high-frequency rTMS, can be effective in promoting motor recovery after stroke, although optimal protocols and patient selection criteria require further refinement. TMS has diagnostic and therapeutic applications across all stages of stroke rehabilitation (acute, subacute, and chronic) and across a range of severities. Diagnostically, TMS can evaluate effects of stroke on corticomotor excitability, intracortical function, and interhemispheric interactions by assessing MEPs.284
Research is ongoing to determine the most effective patient protocols. A key concept in stroke recovery involves rebalancing asymmetric interhemispheric inhibition corticomotor excitability. Standard TMS approaches aim to facilitate ipsilesional M1 excitability or inhibit contralesional M1 excitability.
However, this approach may not be effective for all patients, particularly those with severe stroke and extensive ipsilesional damage. In such cases, patients may rely more heavily on the contralesional dorsal premotor cortex,285 suggesting the need for individualized treatment strategies. A network meta-analysis by Li et al286 high-frequency-rTMS as one of the most effective non-invasive brain stimulation protocols for improving upper-extremity motor function and ADLs after stroke.Recommendation: TMS Statement 1: TMS therapy may provide meaningful motor recovery after stroke (Grade B; certainty: high).
Recommendation: TMS Statement 2: The evidence for TMS therapy for patients with neurodegenerative disease or spinal cord injury is still emerging and does not have any indication for motor recovery or spasticity (Grade C; certainty: low).
Limitations
This guideline has several limitations. First, it is an expert panel’s synthesis of the available literature rather than a formal systematic review or meta-analysis. While a rigorous literature search and quality assessment was conducted, the panel members’ collective experience and expertise may introduce some level of subjective bias.
The literature search was limited to English-language studies published between 2015 and 2025, which may exclude papers that are older, exclusive to other databases, or published in other languages. While the formal search window was limited to 2015-2025, the panel utilized an iterative process to include seminal and foundational research outside this timeframe to ensure a complete technical and historical context for each technology. Included studies varied in design, with a notable scarcity of high-quality, large-scale RCTs specifically focused on functional restoration as a primary outcome.
Many studies had small sample sizes, heterogeneous patient populations, and inconsistent outcome measures. This variability makes direct comparisons between different neuromodulation technologies challenging and limits the generalizability of findings. The technology categories selected for this review are not an exhaustive list of all nascent, clinically-relevant topics.
There is a wide range of evidence across topics, and inclusion of a technology here does not necessarily equate its evidence, relevance, or value compared to others. The authors also recognize that multiple valid methods exist for categorizing and classifying these technologies.
The data presented in Supplementary Material S3 are subject to substantial limitations. Given the evolving nature of payer policies and reimbursement structures, this supplement should be interpreted as a general overview of contemporary coverage considerations rather than a formal economic analysis or cost-effectiveness assessment. Despite these limitations, the panel deemed this summary essential to provide the reader with practical, real-world context on access and financial considerations for these emerging neuromodulation technologies.
The results indicate that inclusion in standard practice is a challenge. While established therapies like SCS and TENS are often covered by major payers, including Medicare, novel treatments such as BCI, scrambler therapy, and multifidus neurostimulation often lack dedicated coverage and are frequently classified as investigational. All neuromodulation technologies, from initial setup (eg, robotic systems exceeding $2 million) to procedure and course costs (eg, rTMS costing up to $15,000), are associated with high costs.
This necessitates that clinical adoption decisions include a rigorous cost-benefit analysis to ensure the financial investment justifies the patient’s anticipated clinical outcome. Treatment access depends on individualized insurer review and a patient’s capacity for substantial out-of-pocket expenses, representing a major limitation to equitable clinical adoption.
Despite the high upfront costs of many of these technologies, these therapies may deliver value-based care that decreases long-term health care costs, decreases the need for further interventions and/or surgeries, and may reduce health care utilization. These technologies may represent economic impact drivers that reduce long-term care burdens, similar to literature for spinal cord stimulation for chronic pain.287
High development and testing costs have limited the clinical applicability of several of these devices. Consequently, it is important to recognize that the guideline’s lack of strong support often reflects a technology’s current stage of development rather than its overall efficacy.
Patient selection remains critical. Clinical trials for BCIs, VR/AR, and robotic rehabilitation systems may not capture a technology’s true potential and efficacy if features are not personalized to patient needs. Future versions will likely offer customization, further affecting clinical efficacy and functional outcomes.
The guideline’s recommendations are based on current evidence, and the field of neuromodulation is rapidly evolving. As such, new technologies and research findings may emerge that could alter these recommendations, which, in any case, are not mandatory and should not replace clinical judgment. Additionally, the guideline’s focus is on technologies for neuromuscular restoration and pain relief in patients with neurological conditions, and its applicability to other patient populations or pain types may be limited.
The lack of consistent insurance coverage for many innovative technologies presents a barrier to their widespread adoption, regardless of their clinical efficacy. While this was discussed, a detailed economic analysis was beyond the scope of this paper. The guideline also does not provide specific guidance on how to train or credential practitioners to safely and effectively use these technologies, which is a critical step for clinical implementation.
Future research should address gaps by conducting more large-scale, pragmatic RCTs with standardized outcome measures and by exploring the economic and implementation aspects of these technologies.
Conclusions
To standardize neuromodulation practices, the AAPM Foundation convened a multidisciplinary panel of experts. The panel reviewed current literature on neuromodulation technologies and neurocomputer interfaces, focusing on objective measures of neuromuscular restoration and pain relief to create a consensus guideline. The aim is to present a comprehensive, evidence-based framework for healthcare professionals, including physicians, physical therapists, payers, and neuromodulation specialists to assist in the selection, implementation, and optimization of neuromodulation techniques.
While each of these technologies has unique mechanisms and applications, a common theme is leveraging neuroplasticity to promote functional recovery. The field is rapidly evolving, trending toward personalized, adaptive, and multi-modal interventions. As technology advances, these tools are poised to play a crucial role in the future of pain medicine and neurorehabilitation.
Further research is needed to optimize protocols, identify ideal patient populations, and address the clinical and financial barriers to widespread adoption. It is also crucial to create patient-specific, customizable protocols.
Supplementary Material