Non-Implantable Prosthetic Devices to Stabilize Posture and Body Balance

Abstract

This is a narrative review that explores the development of non-implantable vestibular devices designed to address postural instability, particularly in aging populations and patients with vestibular hypofunction. It establishes that balance relies on complex sensory integration and that the functional decline of this system creates a significant medical need. Three principal technological strategies are examined: sensory substitution devices, galvanic vestibular stimulation (GVS), and immersive visual feedback systems. Sensory substitution devices, which convert balance data into auditory, tactile, or electrotactile cues, demonstrate significant promise. Examples like vibrotactile belts provide feedback that reduces postural sway, enhancing stability and patient confidence. Parallel to this, GVS—using electrical currents applied to the mastoids—emerges as a potent non-invasive method to modulate vestibular pathways, improving balance control and even inducing neuroplastic changes, especially with stochastic “noisy” signals. The most recently developed devices include augmented and virtual reality technologies that offer innovative visual feedback, creating enriched rehabilitation environments that accelerate recovery by promoting sensory reweighting and neural adaptation. This review concludes that while implantable prostheses are advancing, non-invasive devices offer versatile, affordable, and complementary solutions for balance restoration. The future success of non-invasive alternatives hinges on developing more sophisticated stimulation protocols that account for the complexity of natural movement and individual patient contexts, expanding therapeutic options for vestibular disorders.

1. Introduction

Bipedal posture control depends on the complex integration of vestibular, proprioceptive, visual, and auditory inputs [1,2]. The peripheral vestibular organs act as sensors of linear and angular accelerations of the head. The central nervous system (CNS) uses these signals to determine head orientation relative to gravity and its movement dynamics within the environment, integrating vestibular information with other sensory modalities to produce coordinated voluntary and reflex motor responses that maintain balance and stabilize the gaze [3,4]. The integration of vestibular, proprioceptive, visual, and auditory information contributes to spatial navigation, the creation of an inertial reference map, and cognitive processes involving body schema and attentional processes [5,6,7].

During locomotion—whether walking, driving a car, or piloting an aircraft—humans rely on stabilization processes shaped by learning and training. Yet sensory inputs remain essential for detecting perturbations and stabilizing body posture, particularly under extreme or unforeseen conditions. Disruption of these sensory inputs results in motor impairments that compromise normal function. Vestibular dysfunction, in particular, underscores the system’s central role in balance and posture. Patients with bilateral vestibular hypofunction (BVH), for example, often show profound difficulty stabilizing posture during walking. More critically, they may experience gaze instability, leading to significant visual degradation during movement [8]. The incidence of balance dysfunction—whether due to age-related decline in vestibular function (presbyvestibulopathy) or pathological processes such as Parkinson’s disease (PD) and poststroke conditions—rises markedly with advancing age; indeed, evidence of balance impairment has been documented in 85% of individuals over 80 years old, substantially increasing the risk of falls [9].

To address these challenges, both implantable and non-implantable vestibular devices have been developed (for recent reviews on implantable prostheses, see [10,11,12,13]). Several reviews have examined specific aspects of non-implantable balance devices: Sienko et al. [14] examined the dual role of sensory augmentation systems as real-time balance aids versus rehabilitation tools in vestibular patients. Minino et al. [15] conducted a systematic review of vibratory and acoustic stimulations for postural control. Mohammed et al. [16] provided a narrative overview of wearable technologies for vestibular hypofunction management. More broadly, Kahya et al. [17] and Mohammed [16] reviewed peripheral neuromodulation wearables for gait and mobility improvement in older adults and vestibular hypofunction. Other reviews target the specific effects of GVS on postural control [18,19,20], while others compare implantable approaches versus noisy GVS (nGVS) in terms of postural outcomes and vestibulo-ocular reflex (VOR) responses [12,21]. A systematic review considered sensory-based approaches, such as vibrotactile (Vbt) and auditory feedback, in comparison with virtual reality (VR)-based interventions and physical rehabilitation strategies [22]. In contrast with these focused surveys, the present review integrates all major non-implantable technological approaches—sensory substitution (auditory, tactile, and electrotactile), galvanic vestibular stimulation (GVS), and immersive visual feedback—into a unified description, explicitly addressing their applicability, limitations, and potential for use.

For this review, we included only data from peer-reviewed academic publications. Commercial devices and emerging gray literature may exist, but were not systematically reviewed. For studies addressing clinical use, we limited our discussion to controlled trials that included placebo or sham groups and were conducted under rigorous ethical norms for human investigation.

We collected articles from the past 16 years to provide a comprehensive, representative overview of the principal non-implantable technological approaches for postural stabilization, sensory substitution, GVS and immersive visual feedback across the range of clinical populations and applications addressed in this review.

Notably, none of these devices has a disease-specific clinical indication. The use of a device for a certain pathology mainly reflects the research group that developed it and commercial strategies, since targeting a specific condition facilitates approval and clinical trials. For example, the tongue electrotactile feedback device is currently used for multiple sclerosis, but it could, in principle, be applied to BVH or PD. Similarly, other devices lack a specific indication, having been tested and approved for a particular use rather than being restricted to it.

2. Sensory Substitution Devices

Many patients could benefit from medical devices that enhance postural stability. Individuals with BVH experience chronic imbalance and a markedly elevated fall risk [8,9]; balance and gait impairments also occur in bilateral chronic vestibulopathy [23], multiple sclerosis [24], PD [25,26], stroke—where lateropulsion and asymmetric weight bearing often persist beyond conventional rehabilitation [27,28]—traumatic brain injury [29], and the elderly population [9,17]. In older adults, balance deterioration is a major functional and clinical concern, making this a prevalent need.

To address this need, various approaches have been developed to improve balance. These include methods based on sensory substitution and sensory enhancement. Sensory substitution involves using sensory signals from an alternative sensory modality. This concept was originally introduced by Paul Bach-y-Rita to assist blind individuals with navigation [30]. In contrast, sensory enhancement amplifies the weakened sensory input of a specific modality. This distinction is useful when selecting the appropriate aid for maintaining balance, depending on the impaired sensory or neural input.

Users of sensory substitution devices must learn to interpret the signals provided. The adaptation period varies with the number of cues programmed and their mapping to the desired postural adjustment [31]. Most sensory substitution devices for balance and postural control utilize Vbt stimulation, auditory feedback, or electrotactile stimulation [15]. Vbt devices encode postural deviations as mechanical vibrations delivered to the skin surface, providing the user with a substitute spatial reference that guides voluntary postural correction. Non-invasive balance prostheses have shown particular promise in individuals with the poorest baseline performance, who often experience the greatest improvements. However, the roles of repetition, training duration, and intended use (rehabilitation vs. prosthetic substitution) require further study to establish effectiveness [14]. In general, the more time users spend with the device, the faster they decode its signals and regain stable postural control [32]. Although initially characterized by low specificity, advances in signal processing and artificial intelligence are enabling more individualized, adaptive, and physiologically responsive alternatives that elicit motor responses more precisely matched to the delivered stimulus [33,34]. A systematic review and meta-analysis of randomized controlled trials in neurological diseases found moderate-to-high effectiveness for wearable biofeedback rehabilitation in both PD and stroke populations, with auditory and visual biofeedback demonstrating the most consistent postural improvements [35]. A recent review evaluating smart wearable balance systems noted that only one incorporated machine learning and that most relied on motion tracking or VR, underscoring the early stage of intelligent integration in non-implantable balance prosthetics [36].

Overall, current trends in haptic feedback point to the development of compact feedback units targeting specific balance axes and integrated into everyday items, such as shoe soles or insoles that vibrate upon detecting improper weight distribution, or smart clothing equipped with integrated tactile interfaces. These innovations represent significant progress, offering more refined and effective forms of sensory substitution for maintaining postural balance [15,16].

2.1. Auditory Feedback

Hegeman et al. (2005) [37] first proposed auditory feedback for balance control, demonstrating that individuals with vestibular loss could integrate auditory cues into their postural strategies, thereby improving balance-task performance. A typical auditory feedback system encodes tilt angle into sound volume and pitch. For example, an accelerometer placed at the L5 vertebra estimates body sway near the center of mass and drives changes in sound pitch (anteroposterior sway) or volume (mediolateral sway). Gait performance and balance were evaluated in both healthy subjects and patients with BVH; both groups exhibited reduced sway during quiet standing, with more pronounced improvements in patients with BVH, particularly during trials with eyes closed [38]. A more complex system incorporating different acoustic cues related to movement—mapped through mathematical functions such as sigmoid, exponential, and step functions—was also tested. All auditory biofeedback signal modalities led to improved postural control while standing on a random moving platform [39].

Auditory stimulation not directly correlated to body position also appears to enhance balance. For instance, individuals with cochlear implants showed improved balance while listening to music [40]. Similarly, exposure to auditory white noise improved postural balance during walking in both young and older adults [41]. In patients with otolithic disorders secondary to mild traumatic brain injury, acoustic biofeedback during rehabilitation reduced sway angle and sway velocity relative to no feedback [42]. However, the long-term rehabilitative effects of auditory biofeedback remain to be explored [43].

The effect of auditory noise on standing balance was further investigated by measuring how long participants could stand on one leg with eyes closed. Exposure to threshold-level auditory white noise nearly doubled standing time compared with no-noise conditions. The benefit was greater in individuals with lower baseline balance ability, though no control group other than the absence of noise was included [44].

Auditory feedback has also been used to supplement visual and somatosensory inputs to improve balance across various clinical populations. In post-stroke patients, a two-week intervention using an auditory biofeedback prosthesis—that modulated output according to plantar foot pressure—resulted in a 13–29% reduction in frontal whole-body angular momentum range compared with a control group [27]. Another study examined dynamic balance in chronic stroke patients by measuring center of pressure (CoP) and center of mass (CoM) before and after 20 min of white noise exposure, finding increased anteroposterior CoP range and velocity; this suggests that white noise may enhance dynamic balance and increase gait speed [45].

Compared with other devices, such as Vbt, auditory biofeedback offers a lower-cost implementation and requires minimal body-worn hardware beyond a sensor and earphone; however, it is uniquely disadvantaged in social and noisy environments where acoustic feedback becomes impractical [46]. Moreover, continuous auditory feedback severely limits social interaction, as the user’s attention remains focused on balance control rather than engaging with their surroundings or communicating with others. Its clinical evidence base, while encouraging, is dominated by small single-site studies, limiting generalizability. Beyond these functional limitations, the practical challenge of developing a truly portable auditory device that operates effectively outside controlled laboratory or hospital environments remains unsolved; hence, the technology remains experimental.

2.2. Tactile Feedback

One of the earliest devices was the tongue electrotactile feedback (TEF) system (Figure 1), which exploits the tongue’s high spatial resolution and sensitivity to electrical current. It uses a 10 × 10 electrode array delivering 5–15 V [47]. The system can operate in either a vision substitution mode or a balance stabilization mode [48]. In the balance stabilization mode, a two-axis accelerometer commands the electrical stimulation.

When tested in patients with unilateral vestibular deficits, the head-mounted TEF reduced center-of-pressure displacement relative to controls, indicating improved upright balance [49]. Using plantar pressure sensors instead of a head accelerometer, the TEF enabled healthy subjects to counteract the destabilizing effects of an extended head posture [50]. Furthermore, the device has also been applied to patients with gait or balance impairments caused by multiple sclerosis [51].

A torso-mounted TEF configuration with a two-axis accelerometer attached to a belt successfully countered GVS-induced balance perturbations that mimic elderly instability [52,53]. In chronically dizzy patients with vestibular pathology, four days of one-hour TEF training significantly improved posturographic scores, especially when visual or somatosensory input was restricted; older patients and those with greater vestibular loss benefited more, and gains surpassed those from standard balance physiotherapy alone [54].

TEF technology, commercially available as the BrainPort^® (Wicab, Inc., Fitchburg, WI, USA), has proven utility in blind individuals and BVH patients [55]. The recommended clinical regimen for BVH is the use of the TEF device twice daily for 20 min, which provides lasting improvements in postural stability.

A related device, the portable neuromodulation stimulator (PoNS), delivers non-specific tongue stimulation as a neuromodulation adjunct to physical therapy, yielding neuroplasticity and balance gains in multiple sclerosis [51]; thus, tongue-based interfaces extend into broader neurorehabilitation.

Despite its demonstrated efficacy, TEF has several limitations: (1) the intraoral array interferes with speech and swallowing, restricting use to static or low-demand settings; (2) users require multiple daily training sessions over days to decode spatiotemporal patterns [54]; (3) the limited spatial resolution of the electrode grid may reduce directional-feedback precision, especially in severe vestibular loss; and (4) hygiene and maintenance requirements may lower long-term adherence.

A different approach for development of balance devices uses Vbt feedback that is applied to the torso, head, and fingertips to convey body orientation. An early device delivered Vbt on the anterior and posterior torso surfaces, encoding mediolateral and anteroposterior tilt as well as sway speed [56]. This device also encoded sway speed and was tested in both healthy subjects and patients with vestibular loss, showing that body sway was significantly reduced in both groups [56].

Subsequently, an ambulatory Vbt biofeedback system was tested in patients with bilateral vestibular loss. In its initial configuration, the device uses a tri-axial accelerometer worn on the trunk and twelve equally distributed eccentric vibro-motors around the waist. A small cohort study in subjects with balance impairment demonstrated improved stability in both patients with bilateral vestibular areflexia and vestibular hypofunction subtypes [57]. Users reported increased confidence and better quiet stance performance [32].

This system evolved into the commercial balance belt (Elitac Wearables, Maastricht, The Netherlands), which activates waist tactors when trunk tilt exceeds 2.5° and silences them below 1.5°, thresholds determined by pilot testing for efficacy and comfort [58] (Figure 2). It uses accelerometers and gyroscopes to detect changes in body position and generate corresponding vibrations around the waist. In studies in a cohort of 39 BVH patients with baseline mobility/balance scores of ≤5/10, ~80% reported positive effects in laboratory tests and 68% still benefited after one month of daily use [58]; subsequent studies in ~120 BVH patients confirmed positive outcomes [59]. The balance belt remains one of the few commercially available sensory substitution prostheses.

A headband-based variant, Swaystar Balance Freedom™, uses an L5 gyroscope and eight vibrators around the head for directional sway feedback. Use of the device in persistent postural–perceptual dizziness (PPPD) patients improved stance balance and yielded larger gait improvements [60].

Moving beyond belts, a thimble-like device delivers subtle Vbt to the fingertip upon sway detection, mimicking the stabilizing effect of light touch on a stationary surface [61]. Interestingly, this subtle haptic feedback mimics the well-known phenomenon, where gently touching a stationary surface significantly reduces postural sway. Users achieved sway reductions equivalent to direct physical contact with a stationary surface (Figure 3A) [61]. This concept evolved into a virtual tactile point providing light-touch vibrations [61,62]. A related system creates a “virtual viscosity field” that resists movement proportionally to body velocity. This system was tested in 32 younger and 19 mature subjects; it produced expansions of stability limits and reductions in postural sway [63].

A haptic feedback device based on soft pneumatic actuators applied to the lower limbs delivers localized pressure via controlled expansion, stimulating cutaneous receptors [64]; preliminary results show improved balance and proprioception, suggesting potential for applications in neurological rehabilitation and human–robot interactions (Figure 3B).

The effects of simultaneous Vbt and auditory feedback were assessed in subjects during quiet stance. Actuators were placed on the forehead to avoid perturbing trunk or waist proprioception. Electromyography (EMG) was recorded from muscles of the leg, trunk, and upper arm, while pelvic and upper trunk angular movements were sensed in the roll and pitch planes. BVH subjects were studied without feedback and then were trained with combined Vbt and auditory feedback. The auditory feedback threshold was higher than that of Vbt feedback (80% and 40% of the 90% pelvis sway range, respectively). Sound volume increased linearly from 50 to 70 dB. Eight head vibrators indicated the movement direction and amplitude through varying vibratory intensity. The results showed that muscle synergy amplitudes changed in conjunction with underlying sway reductions. The combination of EMG recording and pelvis movement analysis offers a method for evaluating both non-implantable and implantable prosthesis [65].

Vbt feedback currently represents the most clinically mature non-implantable approach for balance intervention. It is supported by the highest level of real-world evidence, including two-year adherence data [59] and the availability of CE-marked commercial products. Its primary competitive advantage lies in delivering subconscious postural cues without occupying sensory channels required for environmental awareness.

Beyond general balance, Vbt has shown benefit for stance and gait in multiple sclerosis and PD [25,66], serves as a countermeasure for post-spaceflight body tilt [67], and demonstrates moderate-to-high effectiveness in meta-analyses of PD and stroke rehabilitation [35]. Unlike GVS, Vbt avoids transcranial current injection, making it less invasive and thereby reducing the barrier to clinical adoption.

However, Vbt devices face notable challenges: trunk-mounted sensors miss head and neck movements, omitting cephalic contributions that are critical for multi-segmental balance [68,69]; belts for sensors and power can restrict movement and lower compliance [68,69]; and perhaps most concerning is the inadequate characterization of the physiological range over which these devices operate effectively, a knowledge gap that complicates both device optimization and clinical prescription.

2.3. Visual Feedback: Augmented and Virtual Reality

Augmented reality (AR) overlays virtual visual cues onto the real world, while VR immerses the user in a simulated environment; both provide sensory feedback to improve balance (Figure 4). AR glasses can project a stable horizon line or stepped targets to guide posture [70], and VR can present challenging scenarios such as swaying rooms or moving crowds [71,72]. These visual environments promote sensory reweighting, enabling the central nervous system to prioritize reliable visual or residual vestibular inputs over erroneous signals [72].

Recent evidence indicates that VR-based rehabilitation improves balance across diverse populations, benefits demonstrated through improved Berg Balance Scale (BBS) scores in PD and stroke patients [73,74] and reduced dizziness with better posturographic measures in peripheral vestibular disorders [75]; however, long-term retention of these gains remains unclear [76].

However, immersive VR often triggers sensory conflicts due to oculovestibular decoupling, leading to a condition known as “cybersickness”, a distress syndrome caused by exposure to head-mounted displays (HMDs) [77], arising from the mismatch among visual, vestibular, and proprioceptive signals [78]. The Simulator Sickness Questionnaire (SSQ) is the standard severity measure of cybersickness [78]. Evidence indicates that 60–95% of HMD users experience some degree of cybersickness, with 6–12.9% discontinuing exposure prematurely [78]. In vestibular patients, partial vestibular loss makes afferent activity susceptible to incongruent visual motion, potentially worsening symptoms. Research on VR combined with GVS shows that incongruent vestibular stimulation increases postural instability and motion sickness [79], while synchronizing GVS with visual ego-motion—oculovestibular recoupling—may mitigate cybersickness [80].

Multimodal feedback combining auditory, Vbt, and visual cues also improves postural stability and comfort, particularly in balance-impaired populations [81]. Adding VR/AR exercises to conventional rehabilitation produces marked short-term (0–3 months) improvements in Dizziness Handicap Inventory scores, with a significant standardized mean difference favoring VR/AR [82].

Immersive visual feedback likely accelerates balance recovery by inducing neuroplasticity and sensory reweighting, beyond mere repetition. A study using Computerized Vestibular Retraining Therapy improved balance and produced lasting increases in visual and vestibular sensory ratios [83], indicating that exposure to controlled virtual environments that challenge balance leads to more efficient utilization of residual vestibular information and reduces over-reliance on visual cues, common in patients with vestibulopathy. Notably, these neuroplastic effects were more pronounced in patients with moderate-to-severe baseline disability [83].

Taking the specificity of therapy a step further, a recent meta-analysis of 12 randomized controlled trials with 600 participants found that immersive VR is significantly more effective than non-immersive VR for reducing vertigo symptoms [84]. It is worth noting that cybersickness, common initially, generally subsides after four weeks of training [82]. Indeed, combining VR with GVS demonstrated that when vestibular stimulation was applied in the opposite direction to the visual stimulus (incongruent), subjects experienced significantly greater postural instability and cybersickness relative to the no-stimulation control [79].

Beyond vestibular disorders, VR balance games improve postural control in stroke and PD, reflected in a higher Berg Balance Scale and Timed Up-and-Go scores compared with conventional exercises [73,74,85]. VR can expose patients to challenges that would be unsafe in real life, such as walking on a narrow beam or standing on a moving platform. AR, meanwhile, allows patients to practice maintaining balance in their real-life environment with additional cues, such as visual markers indicating the direction of weight shift, which may aid in the transfer of skills to daily activities [79,81,82].

One innovative approach combined an immersive VR headset with a Vbt belt, providing tactile cues on the trunk that signal tilt direction, reinforcing training and reducing visual dependence through co-activation of somatosensory pathways [86].

The field has grown rapidly; publications on “VR and postural control” nearly tripled from 2013 to 2023 [72]. Overall, AR and VR provide rich, interactive environments that offer patients enhanced balance feedback and accelerate postural control improvements.

Visual feedback systems employing augmented and VR technologies introduce unique challenges related to sensory conflict. Immersive VR frequently triggers oculovestibular decoupling, causing cybersickness, dizziness, and autonomic instability that may transiently worsen the patient’s condition [79,82]. Although these effects tend to diminish with continued use, the required adaptation period extends to weeks, which may discourage patients from persisting with treatment. Furthermore, the long-term retention of balance improvements achieved through VR training remains inadequately characterized, raising uncertainty about whether these interventions produce lasting benefits [76]. Despite these limitations, VR/AR provides the richest sensorimotor training environment and strongest neuroplasticity evidence [83,84]; its greatest value lies in its use as a complementary rehabilitation tool, not as a continuous prosthetic aid.

3. Sensory Enhancement Strategies

Galvanic Vestibular Stimulation

Non-implantable devices that deliver GVS have shown considerable promise parallel to advances in vestibular implants [87,88,89,90,91]. GVS is a non-invasive form of sensory enhancement that amplifies diminished vestibular input by applying direct or alternating electrical current over the mastoid processes (for recent reviews, see [20,92]). The technique is safe: no health-related side effects have been reported, even in older adults [93,94]. Because it requires no surgery, GVS can be used in individuals with mild vestibular hypofunction (e.g., age-related decline) [92,95,96], as an adjuvant in vestibular rehabilitation [97], and in extreme conditions such as microgravity, where it may prevent neurosensory conflict [98,99]. A recognized limitation is the relatively low spatial specificity of current spread, which could be addressed through complex stimulation patterns and multi-electrode arrays for focused transcranial delivery [100].

Regarding its action mechanism, GVS modulates vestibular input primarily by altering the firing of primary vestibular afferents at the spike initiation site [101,102], with additional contributions from synaptic activation in hair cells [103,104]. Recordings in awake, behaving macaque monkeys have demonstrated that transmastoid GVS robustly activates both canal and otolith afferents in parallel [105]. Irregularly discharging afferents show higher sensitivity to applied current, in agreement with earlier work on direct-current stimulation of the vestibular nerve [101,102]; however, regular and irregular afferents across all semicircular canal and otolith classes exhibit comparable thresholds [105]. Consequently, the behavioral responses—perceptual, postural, and oculomotor—reflect convergence and integration of signals from both sets of end-organs.

Diverse stimulation parameters and configurations are employed for GVS (Figure 5A). Sinusoidal GVS (sGVS) preferentially targets irregular afferents; direct-current GVS (dcGVS) provides sustained polarization; nGVS applies zero-mean, low-amplitude random noise (<30 Hz) that enhances afferent sensitivity, most likely via stochastic resonance; and pulsed GVS (pGVS) delivers brief, high-intensity current pulses tailored to the physiological response of interest. For pGVS, pulse durations of 100–400 ms engage medium-latency vestibulospinal reflexes, whereas durations below 100 ms preferentially activate short-latency (~9 ms) vestibulo-ocular reflex (VOR) pathways [106]. Pulse trains at ~60 Hz are typically used to study eye movements [106], while lower frequencies (2–10 Hz) probe postural reactions. Duty cycles are kept low (<10%) to avoid charge accumulation and respect safe charge density limits (<40 µC/cm²) [107]. These parameters have not been standardized, contributing to between-study variability [19]. Current amplitudes range from 0.5 mA to peaks of 5 mA for dcGVS and acGVS, and up to 10 mA for short pulses (<100 ms) [19,107].

Electrode placement and polarity critically determine the effects of a given configuration [107]. Common montages include bilateral bipolar (transmastoid), bilateral monopolar, and unilateral monopolar [108,109] (Figure 5B). GVS primarily influences posture [110,111,112], eye movements [113,114,115], the autonomic nervous system [116,117], and cognitive processes [118,119].

The classic postural reaction to GVS is a slight tilt of the body’s vertical axis toward the anode, accompanied by a sensation of motion toward the cathode [120,121,122,123]. This percept, which does not resemble any natural movement, reflects the combined activation of otolithic and semicircular-canal afferents [122]. Postural adjustments depend on the individual’s stance, the supporting surface, and available sensory information [124]; the locomotor response has a latency of around one second from stimulus onset, even when stimulation is self-administered [125]. When GVS is applied during specific gait phases (e.g., heel-off), two distinct strategies emerge: repositioning the foot toward the perceived side of fall, and modulating ankle torque to shift the center of mass [126,127]. A four-electrode configuration delivering a 3 mA DC step can induce directional virtual head motion around all three orthogonal axes, including yaw rotation [91]. Electromyographic recordings during 2 mA, 400 ms transmastoid GVS reveal short-latency (~60 ms) and medium-latency (~100 ms) components in triceps brachii, tibialis anterior, and soleus muscles, indicating engagement of reticulospinal, vestibulospinal, and rubrospinal pathways [128]. A linear relationship between GVS duration and H-reflex/M-wave amplitude further suggests supraspinal modulation of spinal motoneuron excitability [129].

GVS produces a significant oculomotor effect. GVS biases eye position toward the anode and away from the cathode [130]. Three-dimensional video-oculography during 5 mA transmastoid GVS reveals substantial inter-individual variability in ocular displacement, largely attributable to differences in electrode placement, impedance, and cerebrospinal fluid shunting [131,132]; nonetheless, within-subject responses are repeatable and predictable from current magnitude and montage. Using pGVS (60 Hz, 0.9–10 mA, and 100 ms pulses), the VOR latency is approximately 9 ms at currents ≥2.5 mA, increasing to ~32 ms near the perceptual threshold (0.9 mA) [106]. The evoked eye movements comprise torsional deviation away from the cathode, vertical displacement, and horizontal rotation, all scaling with current intensity. Crucially, these pulse-evoked ocular reflexes are not suppressed by voluntary eye movements, highlighting the potential to restore low-frequency, low-velocity VOR function [106].

Most advanced GVS systems are now focused on the use of noisy GVS. This particular waveform of GVS has attracted particular interest because low-amplitude stochastic input can enhance the signal-to-noise ratio through stochastic resonance, a phenomenon demonstrated in the semicircular canal afferent discharge of chickens [133]. Low-level nGVS significantly modulates the amplitude of ocular vestibular-evoked myogenic potentials (oVEMPs) without affecting their latencies, likely by amplifying utricular responses via this mechanism [88]. Behaviorally, nGVS (<30 Hz; 0.1–0.5 mA) reduces CoP oscillations during quiet stance [134].

Pioneering clinical work applied transmastoid white-noise nGVS (0–1 mA; 30 s) to healthy individuals and patients with bilateral vestibular hypofunction (BVH) and observed improved stabilometric parameters, velocity, sway area, and RMS of CoP in 91% of patients, with an optimal amplitude of 456 ± 82 µA [87]. In a subsequent study of 30 older adults, two nGVS sessions (30 min twice and a single 3 h session) yielded a mean optimal intensity of ~200 µA; postural stability remained enhanced for >2 h, and a second session produced further benefit [135]. Similar improvements have been reported for passive body motion perception [136] and gaze stabilization in older adults, where 1.6 mA nGVS partially restored ocular counter-rolling gain, a response mediated predominantly by otoliths at low motion frequencies [137]. Task-specific effects were noted: nGVS robustly improved direction recognition during inter-aural translation, whereas only 50% of subjects showed mild improvement in yaw rotation, consistent with a predominant otolithic site of action [138]. In healthy upright subjects, 0.2 mA nGVS reduced VOR gain, while 0.6 mA increased center-of-pressure path length, with no significant correlation between the two measures [115].

Contemporary nGVS systems fall into two categories: (i) automatic devices in which stimulation amplitude is modulated in real time by head-mounted gyroscopes and accelerometers [89,90,139], and (ii) pre-programmed devices that deliver constant-parameter stimulation throughout use [140]. These systems have been tested across diverse groups: pilots [99,141,142], athletes [143], older adults [144], and patients with BVH, PD, or post-stroke instability [145,146].

In summary, nGVS has emerged as an effective, non-invasive option for managing BVH, with growing evidence of its utility in PD and post-stroke postural instability [18,145] (

Table 1. Prospective medical applications of GVS.

Author	GVS	Subjects	Result
Wuehr et al., 2016 [147]	Transmastoid nGVS (sham trial) and nonzero-amplitude nGVS set to 80% of the individual cutaneous threshold for GVS.	BVH patients (n = 13) were tested while walking.	Walking improvement in BVH patients was more notorious during slow stride than at a fast pace. nGVS did not provoke nystagmus, vertigo or pain to any participant. Results showed that GVS improved gait performance in pathological conditions.
Iwasaki et al., 2018 [148]	Transmastoid nGVS.	19 healthy controls and 12 patients with BVH.	nGVS had significant effects on gait velocity, stride length and stride time in healthy subjects as well as in patients with BVH.
Chen et al., 2021 [149]	nGVS intensities (0–1000 μA). Amplitude determined by standing stability.	Ten BVH patients and 16 healthy participants. nGVS applied in straight walking and 2 Hz head yaw walking in light and dark conditions.	In the light, the CoM deviation decreased in straight walking for the BVH. In the dark, both healthy and BVH showed decreased lateral deviation during nGVS. The chest–pelvic ratio angle significantly decreased in BVH for 2 Hz head yaw walking. nGVS reduced walking deviations in BVH patients.
Wuehr et al., 2023 [150]	nGVS (mean intensity: 0.36 ± 0.16 mA). Optimized for each subject to stabilize in posturographic assessment.	Eleven patients with BVH (mean age: 54.0 ± 8.3 years; 7 females).	nGVS improves vestibular perceptual performance determined as direction recognition thresholds for head-centered roll tilt motion on a 6DOF motion platform.
Nguyen et al., 2023 [151]	GVS (sinusoidal, direct current, and noisy), amplitude (0.4, 0.8, and 1.2 mA), and duration (5 and 30 min).	Patients with either unilateral or BVH (n = 18) or cerebellar ataxia (n = 13) were enrolled in the study.	Patients with unilateral vestibulopathy experienced the most favorable change in dizziness perception and imbalance with nGVS or sGVS at 0.4 mA for 30 min, followed by DC GVS at 0.8 mA for 5 min. nGVS, use of 0.8–0.4 mA, for 30 min was most effective in BVH and cerebellar ataxia patients.
Wuehr et al., 2024 [152]	Transmastoid nGVS of various levels from 0 to 0.7 mA.	BVH patients (n = 19) and paired controls.	Body sway versus nGVS amplitude showed a bell-shaped function in 63% of patients, thus indicating a stochastic resonance with optimal improvements of 31% at an average intensity of 0.3 mA. Patients with a stronger stochastic resonance-like response showed the most meaningful improvement in static balance.
Curry et al., 2024 [153]	Vestibular nerve stimulation (VeNS) using Modius Sleep device 30 min daily for 4 weeks.	Randomized, sham-controlled trial in 147 participants with moderate-to-severe insomnia (Insomnia Severity Index (ISI) ≥ 15).	After 4 weeks, mean ISI score reduction was greater in the VeNS than sham group. Mean ISI score decreased by 5.8 (95% CI: [−6.8, −4.81], approaching the clinically meaningful threshold of a 6-point reduction.
Mitsutake et al., 2024 [154]	Square-wave transmastoid GVS of 3000 μA and 200 μA, and nGVS of 200 µA.	Twenty-six healthy volunteers in two groups: balance training combined with nGVS and sham GVS. Nine consecutives 60 s GVS periods.	nGVS group showed significantly increased post-intervention H-reflex amplitude.
King et al., 2025 [155]	nGVS amplitude of ±0.35 mA, wideband of 0.001–300 Hz, bipolar mastoids and C4 reference, for 20 min three times weekly for six weeks.	40 older adults randomly assigned to a stimulation group of noisy electrical vestibular stimulation (nEVS intervention) or sham group.	Following a regimen of multiple GVS (nEVS in this work), improvements in balance persisted for up to six months. This suggests the potential for long-term training effects, possibly due to neuroplastic changes in the vestibular system.
Fujimoto et al., 2025 [156]	Bipolar transmastoid. nGVS 100–2000 µA for 30s.	Randomized, double-blind, placebo-controlled trial in 39 patients with unilateral or bilateral peripheral vestibulopathy.	30% of patients demonstrated significantly greater reductions in CoP velocity at 100 μA and 1700 μA.
Menon et al., 2025 [157]	GVS 0.1 mA, with steps of 0.1 mA, until skin sensitivity. Two-pole (transmastoid) and three-pole (two additional electrodes on the temples) stimulation.	12 participants with PD.	GVS improved visuomotor target tracking in individuals with PD. Both two-pole and three-pole stimulation were effective. The most effective stimulus across all subjects was a waveform with an envelope frequency of 30 Hz and a carrier frequency of 110 Hz, which improved motor performance by 25% relative to the sham stimulus.
Oh et al., 2025 [158]	Bipolar transmastoid. Direct current (DC) of 0.8 to 1.0 mA, 30 min daily for 10 days.	Single-blind, randomized, sham-controlled trial in 83 acute unilateral peripheral vestibulopathy (AUPV) patients using GVS (cathode on lesion side).	Improved visuospatial memory performance. Findings support GVS as a neuromodulatory intervention to enhance spatial memory and facilitate cognitive recovery in AUPV.
Cheung et al., 2025 [159]	Use of VeNS amplitude of 0–1 mA until subjects felt a swaying sensation. Frequency: 100 Hz. Modius Sleep device (Nurovalens®).	Participants, including 43 adults exhibiting insomnia symptoms and 40 paired sham controls, underwent 20 VeNS sessions, one hour prior to bedtime during 30 min.	Findings suggest that VeNS was effective in reducing insomnia severity and improving participants’ physical well-being immediately after the 4-week intervention and at the 3-month follow-up compared with the sham VeNS group.

VeNS: vestibular nerve stimulation; AUPV: acute unilateral peripheral vestibulopathy; ISI: Insomnia Severity Index.

). Persistent heterogeneity in amplitude, waveform, and electrode placement, however, has so far precluded a formal meta-analysis and the derivation of pooled effect estimates [145]. Future work that systematically compares configurations under homogeneous conditions will be essential to optimize stimulation protocols and translate nGVS into routine clinical practice.

Although GVS is one of the most studied non-invasive approaches, it suffers from fundamental limitations in specificity and standardization. It activates both canal and otolith afferents non-specifically, failing to replicate natural vestibular encoding, and may produce unwanted eye movements during postural adjustments. Large inter-subject variability—arising from electrode placement, electrode resistance, and cerebrospinal fluid distribution—further complicates clinical application [157].

The absence of standardized dosing units is even more problematic: studies use movement sensation thresholds, maximum tolerated current, or other metrics, preventing systematic comparison [160]. The lack of studies comparing different stimulation configurations under homogeneous conditions prevents decisions about optimal parameters [107]. Current intensities vary widely across studies, ranging from 0.5 to 5 mA, with no consensus on appropriate values for specific conditions or patient populations [19]. GVS is a specialized form of transcranial current stimulation, which is now widely used in clinical settings and may serve as a useful guide for establishing GVS protocols [161]. Paradoxically, low-amplitude noisy stimulation can degrade balance in some healthy individuals, indicating that these interventions may sometimes disrupt rather than enhance postural control [162].

4. General Limitations and Challenges of Non-Implantable Devices

Despite considerable recent advances, significant limitations persist across all non-implantable approaches to balance stabilization. These span modality-specific drawbacks, technical and physiological constraints, user population considerations, and gaps in research and clinical translation. Recognizing these challenges is essential for guiding innovation and improving outcomes for patients with balance disorders.

Table 2. Comparison of non-implantable devices for balance.

Technology	Mechanism	Clinical Evidence	Target Population	Advantages	Limitations
Auditory biofeedback	Sensory substitution (acoustic encoding of sway)	Moderate (RCTs in BVH, stroke, and mTBI)	BVH, stroke, and mTBI	No ear canal occlusion; portable; land ow cost	Acoustic signal competes with and may mask environmental sounds, social limitation, and no ambulatory standardization
Vbt/haptic biofeedback [58,59]	Sensory substitution (tactile encoding of trunk tilt)	Moderate–high (RCTs and long-term follow-up studies)	BVH, PD, MS, and elderly	Ambulatory, subconscious, wearable, and commercially available	Incomplete postural info (no cephalic axis), cognitive load in elderly, and trunk-only sensing
TEF	Sensory substitution (electrotactile tongue array)	Moderate (RCTs, BVH, and MS)	BVH, MS, and dizzy patients	High spatial resolution and neuroplasticity effects; limited accessibility	Intraoral, interferes with speech/eating, significant learning period, and limited portability
GVS	Sensory enhancement (direct vestibular afferent modulation)	High (many RCTs and systematic reviews)	BVH, PD, elderly, and astronauts	Modulates vestibular pathway directly; neuroplastic potential; intermediate accessibility	Non-specific activation, high inter-subject variability, and no standardized dosing
VR/AR [36,78,82,84]	Visual feedback substitution or enhancement	High (meta-analyses and systematic reviews)	BVH, PD, stroke, and peripheral vestibular disorders	Rich sensorimotor environment, neuroplasticity, and highly accessible	Cybersickness, no long-term retention data, and requires technical infrastructure

Abbreviations: MS, multiple sclerosis; mTBI, mild traumatic brain injury; RCTs, randomized controlled trials. Evidence levels: high—supported by meta-analyses and multiple independent RCTs; moderate–high—supported by RCTs with long-term follow-up data; and moderate—supported by small-to-medium RCTs with limited generalizability.

provides a comparative synthesis of the main non-implantable approaches reviewed, evaluating each one along the dimensions of mechanism, clinical evidence level, practical considerations, and limitations.

User population characteristics introduce additional layers of complexity. Elderly individuals, who represent the largest potential beneficiary group for these technologies, face a particular cognitive burden. It seems that the CNS prioritizes postural performance over cognitive tasks [163]. People with BVH could use Vbt devices for balance control, and no difference from age-matched controls was found during dual-task conditions. In both BVH and age-matched controls, the reaction time performance was significantly degraded [164]. Sensory substitution inherently demands attention, and age-related declines in processing speed and dual-task performance may limit device effectiveness in older users [16,165].

The heterogeneity of pathologies further complicates device development and prescription. Identical feedback systems produce different responses in patients with bilateral vestibular loss versus those with proprioceptive deficits, yet current approaches rarely account for these differences [150]. Although patients with the poorest baseline performance show the largest improvements, this variability makes it difficult to establish standardized prescriptions and predictions of individual responses.

From a technical perspective, current devices provide incomplete kinematic characterization. Most deliver linear acceleration information for balance control but fail to contribute to gaze stabilization, a fundamental vestibular function [166], a significant gap relative to implantable devices. More fundamentally, they inadequately distinguish self-generated voluntary movements from external perturbations, the “exafference problem” [167]. Lacking internal models or efference copy, devices react rather than predict, potentially assisting counterproductively during voluntary motion. Advances in postural signal processing, including rapid mathematical analysis, could improve discrimination of stability during normal and abnormal conditions [168].

When compared with implantable alternatives, non-invasive devices face inherent limitations in functional scope. They primarily address postural control without fully replicating the vestibular system’s contributions to the vestibulo-ocular reflex, spatial navigation, and body schema representation [11,13,169]. Hybrid approaches combining multiple stimulation modalities are promising but introduce synchronization challenges and may increase the cognitive burden.

Significant gaps also hinder research progress. The absence of standardized protocols—varying stimulation parameters, electrode configurations, and outcome measures—prevents meaningful meta-analyses and slows knowledge accumulation [170]. Most evaluations are restricted to controlled laboratories, with little real-world effectiveness data. Long-term adherence and lasting rehabilitative effects remain insufficiently explored. Patient stratification by specific pathology, severity, and comorbidity is rare, limiting clinical applicability. A recent systematic evaluation of commercial smart wearable balance systems confirms that evidence-based implementation guidelines are largely absent, and direct comparative studies are scarce [36].

The next generation of balance prostheses must discriminate self-generated movements from external perturbations [171,172,173]. Current reactive devices lack efference copy and ignore voluntary motion, sometimes providing counterproductive assistance. Future systems should incorporate intent recognition algorithms powered by machine learning (ML); by fusing multimodal inputs (EMG, IMU, and pressure sensors), ML models can predict movement initiation and distinguish intent from instability [174,175].

The inherent limitations of single-modality devices, such as the lack of specificity in GVS or the cognitive burden of haptic decoding, can be mitigated by hybrid systems. Integrating VR with GVS has enabled oculovestibular recoupling—synchronizing galvanic pulses with visual ego motion—thereby reducing cybersickness and autonomic instability [176,177]. Although still in pilot stages, combining Vbt feedback (for center-of-mass stability) with GVS (for head/gaze stabilization) could offer a more holistic restoration by addressing both trunk and cephalic reference frames.

Specific technical solutions developed include adaptive threshold algorithms or bone conduction transducers for auditory feedback (to leave the external ear canal unoccupied); incorporation of head-mounted IMU units and soft-robotic actuators for vibrotactile/haptic systems (to capture cephalic contributions and improve comfort); miniaturization and wireless integration for TEF (to reduce intraoral burden); GVS integration with VR and graduated exposure protocols for visual feedback systems; and multi-electrode arrays with individualized amplitude calibration and ML models to personalize stimulation in GVS. These complement the existing discussion on hybrid VR–GVS systems for oculo-vestibular re-coupling [156,157] and on ML-based intent recognition [151,152,153,154,155].

In conclusion, while non-implantable postural devices offer accessible and promising alternatives for balance restoration, their development and clinical application face substantial challenges spanning technical limitations, physiological constraints, user population heterogeneity, and research gaps. Overcoming these through innovative engineering solutions, rigorous clinical investigation, and personalized approaches is essential to fulfilling the therapeutic potential of these technologies for the growing population with postural disorders, mainly of vestibular origin.