Next-generation wearable neuromonitoring technologies: signal acquisition, modalities, and clinical relevance

1. High-density surface electromyography

Traditional bipolar surface EMG utilizes two electrodes to measure the global electrical activity of a muscle, which typically lacks spatial resolution and is susceptible to crosstalk [19]. In contrast, HD-sEMG deploys two-dimensional grids comprising tens to hundreds of miniaturized electrodes closely spaced (e.g., 3-8 mm apart). This configuration provides a high-resolution spatiotemporal map of the electrical potential distribution across the skin surface [20].

The primary technical advantage of HD-sEMG lies in its advanced signal processing capabilities. Utilizing mathematical algorithms such as blind source separation, the complex interference pattern recorded at the surface can be decomposed into the discharge trains of individual motor units [21]. This non-invasive extraction of motor unit action potentials and their firing rates mimics some of the diagnostic information traditionally obtained via invasive needle EMG [22].

Despite its potential, wearable HD-sEMG faces technical hurdles. The computational load required for real-time decomposition is substantial, challenging the battery life of wireless sensors. Additionally, maintaining uniform impedance across a large grid during dynamic movements remains difficult, as localized sweating or electrode lift-off can degrade signal quality.

In terms of exploratory applications, flexible HD-sEMG patches have been investigated for tracking physiological markers of reinnervation. As a damaged nerve regenerates, surviving motor neurons undergo collateral sprouting. Conceptually, by wearing an HD-sEMG array, a patient’s progressive reinnervation could be quantified over weeks, potentially serving as an objective biomarker for nerve regeneration speed, although clinical validation of this approach is still in its early stages [23].

2. In-ear electroencephalography

Prolonged cortical monitoring is often hampered by the obtrusiveness of conventional scalp EEG systems, which require wet gels and are prone to movement artifacts. In-ear EEG, often integrated into customized earpieces or "hearables," records electrical activity using dry or semi-dry electrodes placed directly within the ear canal or around the auricle [24].

Anatomically, the ear canal is in close proximity to the temporal lobes, providing a stable vantage point for recording auditory evoked potentials, sleep spindles, and certain epileptiform discharges. Because the ear canal is relatively stationary compared to the scalp and naturally holds the sensor securely, in-ear EEG tends to suppress motion artifacts and offers enhanced wearability for long-term daily use [25].

The most significant limitation of in-ear EEG is its low spatial resolution. Due to the small inter-electrode distance and confined anatomical location, it is generally unable to capture focal activities occurring in the frontal or occipital cortices. Furthermore, the limited number of channels restricts complex source localization.

Despite spatial constraints, in-ear EEG has shown preliminary utility in automated sleep staging, which is relevant since sleep architecture is often disrupted post-surgery and in neurodegenerative diseases. Furthermore, clinical studies have highlighted the inherent limitations of intraoperative monitoring in predicting delayed postoperative neurological events [10,16]. To help bridge this gap, it has been proposed that wearable EEG tracking could serve as an adjunctive tool to monitor postoperative arousal and neural network integrity in ambulatory environments, though this application remains largely conceptual at present.

3. Flexible epidermal sensors for autonomic monitoring

Autonomic nervous system dysfunction is a critical component of various conditions, ranging from systemic neuropathies to disorders of gut-brain interaction, where wearable monitoring is increasingly being evaluated [26]. Recent advancements in materials science have led to ultra-thin, flexible epidermal sensors designed specifically for continuous biomarker and autonomic tracking. Utilizing elastomeric substrates embedded with nanomaterials, these sensors exhibit mechanical properties that conform closely to the human epidermis.

For instance, epidermal wearable biosensors have been specifically engineered to continuously monitor biochemical markers in sweat, offering non-invasive tracking capabilities for chronic diseases [27]. Their thin profile makes them minimally obtrusive to the wearer, supporting high patient compliance in both clinical and home settings.

However, translating these devices to daily use requires overcoming mechanical limitations. Maintaining long-term durability necessitates robust, wireless, skin-integrated system designs that can function reliably across various physical healthcare environments [28].

From a neurophysiological perspective, integrating the continuous measurement of electrodermal activity and heart rate variability provides a direct window into assessing autonomic function and sympathovagal balance [29]. Consequently, this modality extends the reach of neuromonitoring beyond motor and sensory pathways, providing a potential window into stress-induced physiological changes and autonomic neuropathies.

Table 1 summarizes the key characteristics of the major wearable neuromonitoring modalities, including their signal acquisition principles, strengths, limitations, and representative clinical relevance.

Table 1.

Characteristics of emerging wearable neuromonitoring modalities

1. Postoperative care in spine and peripheral nerve surgery

Surgeries for cervical radiculopathy and myelopathy carry a risk of postoperative complications, such as C5 palsy, which typically manifests 2 to 7 days post-surgery. IONM has established the utility of tracking motor-evoked potentials to predict upper extremity motor paresis during such procedures [30,31]. Drawing upon these established IONM principles, it is conceivable that a wearable HD-sEMG array on the deltoid muscle could allow clinicians to extend motor unit recruitment monitoring into the postoperative ward. In this exploratory framework, observing a delayed drop in recruitment efficiency might alert the surgical team to impending neuropraxia. Similarly, building on the concepts of intraoperative recurrent laryngeal nerve assessment [32], wearable sensors placed on the neck could theoretically be used to track the postoperative recovery trajectory of vocal fold neuropraxia after thyroidectomy. It should be noted, however, that these postoperative applications have not yet been validated in controlled clinical trials and remain at a conceptual stage.

2. Neurorehabilitation and corticomuscular coherence

In stroke rehabilitation, assessing whether motor improvements stem from true neurological recovery or compensatory movements is challenging. Corticomuscular coherence (CMC) provides a mathematical measure of the functional coupling between the primary motor cortex (via EEG) and the contracting muscle (via EMG) [33]. By deploying a multimodal wearable system, clinicians could estimate CMC continuously during rehabilitative exercises. A progressive increase in beta-band CMC (15-30 Hz) over weeks of therapy has been hypothesized to reflect functional rewiring of the corticospinal tract, and could, if validated, serve as a digital biomarker to help time adjunct therapies such as robot-assisted training [34].

3. Tracking progression in neurodegenerative diseases

Optimizing dopaminergic medication in Parkinson’s disease requires an accurate understanding of the patient’s "On" and "Off" times, which typically relies on subjective patient diaries [8,11]. Wearable kinematics combined with surface EMG have been explored as an objective method to capture resting tremor characteristics, bradykinesia, and muscle rigidity under naturalistic conditions [35]. Over time, repeated sampling of these parameters can generate a personalized record of motor symptom fluctuations. While this continuous peripheral data is also being investigated for its potential to inform advanced closed-loop stimulation parameters [36], its primary near-term value lies in providing clinicians with an objective, continuous metric to guide routine medication adjustments.

Table 2 summarizes representative clinical scenarios in which wearable neuromonitoring technologies may provide meaningful longitudinal information beyond conventional intermittent assessment.

Table 2.

Representative clinical applications of wearable neuromonitoring technologies

1. Signal pre-processing and noise reduction

Real-world ambulatory data is inherently noisy and complex. AI algorithms, such as hybrid neural autoencoders, have been utilized to process and encode complex sensory and neural data efficiently [37]. In the context of wearable devices, the application of similar neural network architectures has the potential to assist in isolating physiological signals from background noise and baseline drift, which is a necessary step before any reliable clinical interpretation can occur. However, prospective validation of such pipelines on ambulatory wearable data remains limited.

2. Pattern recognition and prognostic assessment

Once the data is pre-processed, ML models can be utilized to handle the temporal dependencies of physiological time-series data. By analyzing extracted features, AI models may identify latent patterns related to patient mobility and neurological status. For instance, ML algorithms analyzing continuous gait parameters and wearable sensor data have demonstrated the ability to assess fall risk accurately among vulnerable populations [38,39]. Furthermore, AI methodologies have demonstrated robust capabilities in predicting severe neurological outcomes in critical care scenarios, such as post-cardiac resuscitation [40]. While originating from acute care, this illustrates the broader prognostic potential of AI in neurological monitoring; by applying similar pattern recognition frameworks to ambulatory wearable data, ML may in the future assist clinicians in stratifying patients based on objective digital biomarkers. At present, however, this translational step has not been broadly demonstrated, and further clinical studies are needed to establish the reliability and generalizability of such approaches.