Struggling to get restorative sleep without reaching for pills? Researchers at the University of Texas at Austin have built a soft, skin-attached patch that uses gentle ultrasound to help people reach REM sleep about 43 minutes faster and stay in it roughly 16 minutes longer, with no drugs or invasive procedures. In a small 28-person study published in Nature Communications, the device also tracked brain activity in real time, offering a new window into how deep-brain stimulation could shape sleep and stress resilience at home.
What NEUSLeeP Is and Why It Matters
NEUSLeeP stands for Non-invasive Electrophysiological Recording and Ultrasound Neuromodulation Sleep Patch. It was developed by a team led by biomedical engineering Ph.D. graduate Kai Wing Kevin Tang and principal investigator Huiliang Wang in the Cockrell School of Engineering at UT Austin.
The patch addresses a long-standing problem in sleep medicine. Current noninvasive tools such as transcranial electrical stimulation, magnetic stimulation, and auditory stimulation cannot precisely target the deep subcortical circuits that govern sleep stages. Deep brain stimulation can target those circuits, but it requires surgery. Transcranial focused ultrasound offers precise deep-brain targeting without incisions, yet existing systems are bulky and impractical for overnight use in a bedroom.
NEUSLeeP aims to bridge that gap. It is a lightweight, flexible platform designed to be worn comfortably throughout the night while both monitoring sleep and delivering targeted stimulation.
- Purpose: Enhance rapid eye movement (REM) sleep, the stage tied to emotional regulation, stress adaptation, and dreaming.
- Published: 4 June 2026 in Nature Communications (Tang et al., 2026)
- ClinicalTrials.gov identifier: NCT07190287
- Size: About 103 grams and 3.5 mm thick, roughly 80% thinner than conventional ultrasound stimulation devices.
How the Wearable Ultrasound Sleep Patch Works
The core innovation is integration: sensing and stimulation in one skin-attached system that remains stable as you move during natural sleep.
The three integrated layers
The device combines three custom-engineered components inside a soft bioadhesive substrate:
- Concentric Ring Ultrasound Transducer Array (CRUTA): An eight-channel piezoelectric array centered at 0.65 MHz (megahertz, a measure of ultrasound frequency). By adjusting timing delays between rings, the array tunes the focal depth to reach the subthalamic nucleus (STN), a small deep-brain hub that helps regulate movement and sleep, through the thin temporal window above the cheekbone. The skull itself helps tighten the beam, improving spatial focus.
- AMPS-based Sleep Gel (ASG): A clear, soft hydrogel that provides ultra-low impedance electrodes for sleep recording. It tracks brain waves (EEG, electroencephalography), eye movements (EOG, electrooculography), and muscle tone (EMG, electromyography). It retained about 76% of its mass after 72 hours in testing and showed about 50% lower electrical resistance in the key EEG frequency range compared with commercial gels, enabling stable overnight lab-grade sleep recording with only six electrodes.
- Eco-PEIE-Gel bioadhesive substrate: An Ecoflex-Gel modified with polyethyleneimine ethoxylate. Skin adhesion reached 0.749 N/cm, a 2024% increase over unmodified Ecoflex-Gel, while staying soft and flexible (about 18 kPa, kilopascals, a measure of stiffness where lower means softer). Redness scores remained very low after 8 hours, indicating minimal skin irritation.
Where and how it stimulates
During the study, the patch targeted the left subthalamic nucleus (STN), a small structure deep in the basal ganglia network that helps coordinate movement and sleep-wake transitions. Investigators used MRI brain scans to verify the line-of-sight and measured the distance to the STN (average 72.2 millimeters, about 2.8 inches, from the temple window) to set the focal depth precisely.
Stimulation used a gentle, low-duty-cycle protocol. That means ultrasound was on only 5% of the time: brief pulses at 100 times per second (100 Hz), with a pressure of 0.90 MPa (megapascals, a measure of acoustic pressure), delivered as 30 seconds on and 30 seconds off for 5 minutes per block. Blocks were repeated every 90 minutes to align with the natural ultradian sleep cycle. Earlier pilot work showed this 100 Hz setting produced an inhibitory, calming effect on STN activity, matching approaches used in clinical deep brain stimulation associated with improved REM.
At the same time, hydrogel electrodes recorded brain waves, eye movements, and muscle tone to identify sleep stages in real time. When the team compared the patch’s readings to a standard 32-sensor lab headcap in three participants, the two systems agreed on sleep staging almost perfectly. Researchers reported a statistical agreement score of 0.88 on a 0-to-1 scale where anything above 0.80 is considered near-perfect agreement.
Study Results: 43 Minutes Faster to REM, 16 Minutes More REM
The headline findings come from two consecutive overnight sessions in the UT Austin Sleep Lab. The first night was sham (no ultrasound), the second night delivered active stimulation. This fixed order was chosen because the durability of ultrasound effects was unknown and carryover could have contaminated a counterbalanced design, a limitation the authors acknowledge.
Of 28 enrolled adults (16 healthy, 12 with mild insomnia symptoms, ages 19 to 38), 26 provided usable overnight recordings. Sleep was scored in 30-second epochs by a certified rater blinded to condition.
- REM share of total sleep rose from 16.3% to 20.9% A 4.6 percentage-point increase, equal to about 15.9 minutes of extra REM per night.
- Time to first REM fell from 177 minutes to about 135 minutes: A 24% reduction, or 43 minutes faster on average.
- Both groups improved REM proportion: Healthy sleepers rose from 15.1% to 19.4%, insomnia group from 17.9% to 23.2%. The latency benefit was statistically strong in healthy sleepers and directional in the insomnia group.
- No disruption to other sleep: Wake-ups during the night and overall sleep efficiency (the share of time in bed actually spent asleep) were unchanged (healthy: 80.0% sham vs 81.4% with stimulation; insomnia: 81.4% vs 80.6%). The three non-REM stages were also unchanged: light drowsiness (N1), light stable sleep (N2), and deep slow-wave sleep (N3).
- Comfort and safety: Participants described the patch as comfortable. Vital signs including blood oxygen (SpO2, the percentage of oxygen in the blood), heart rate, and blood pressure showed no meaningful changes, and adverse effects were minimal. Skin irritation was negligible.
For context, prior subthalamic deep brain stimulation studies reported REM increases of 1.3% to 6.2% and latency reductions of about 15 minutes.
NEUSLeeP’s 4.6% increase and 43-minute acceleration sit at the strong end of that range, achieved without surgery.
Beyond Sleep: Heart Rate Variability and Emotional Brain Changes
This is where the study becomes more than a sleep gadget story. The researchers also examined whether enhanced REM translated into daytime resilience.
Improved heart rate variability in healthy sleepers
Heart rate variability (HRV) is a measure of how much the time between heartbeats fluctuates. Higher variability generally means the body is more adaptable to stress, as the calming branch of the nervous system (the parasympathetic system) is more active. In this study, HRV was calculated from 5-minute heart recordings (ECG, electrocardiogram) using a standard metric called RMSSD, which captures beat-to-beat differences. Higher scores generally track with better stress regulation.
- In healthy participants, overnight change in HRV improved markedly: -5.89% on the sham night versus +41.5% on the stimulation night (p = 0.014).
- In the insomnia group, HRV did not show a clear stimulation effect (26.9% sham vs 4.79% with stimulation), consistent with prior work showing blunted autonomic responsiveness in insomnia tied to hyperarousal.
A statistical model that accounts for where participants started showed that baseline HRV strongly predicted morning-after HRV. Even after that adjustment, healthy participants had higher morning HRV than those with mild insomnia, and the benefit of stimulation was larger in the healthy group.
Brain circuit modulation seen on fMRI
Morning-after functional MRI during an emotional face-matching task revealed selective changes rather than global shifts:
- Healthy group: Reduced activity on the same side as stimulation in basal ganglia and midbrain hubs, including two parts of the substantia nigra (SNc and SNr, involved in dopamine signaling and activity control), in response to emotional faces (anger, fear, happy), alongside reduced activity in temporal and midbrain circuits and increased activity elsewhere in the cortex. In resting brain scans taken without a task, activity fluctuations rose at the stimulated STN, and blood flow increased in the left nucleus accumbens, a region tied to reward and motivation.
- Insomnia group: A more heterogeneous pattern with elevated amygdala reactivity across conditions, a profile often reported in insomnia linked to deficits in emotion and arousal regulation.
As UT co-investigator Gregory Fonzo summarized: “REM sleep is not just about dreaming, it is about emotional reset and stress adaptation. By enhancing REM, we may help people better cope with stress and improve their overall well-being.”
How NEUSLeeP Compares to Current Sleep Solutions
Poor REM quality is associated with depression, anxiety, and PTSD, yet most existing treatments do not directly target REM circuits and often bring side effects such as daytime drowsiness, memory effects, or limited durability. NEUSLeeP is still a research prototype tested under a registered protocol, not a consumer product, but its profile is distinctive.
| Approach | How It Works | Strengths | Limitations |
|---|---|---|---|
| NEUSLeeP (focused ultrasound) | Skin-attached patch delivering focused ultrasound to the STN plus simultaneous brain-wave monitoring (EEG) | Noninvasive, millimeter-precise deep targeting; real-time monitoring; REM-specific gains without altering NREM; HRV and circuit effects suggest stress benefits | Small sample (n=26 usable), single-blind fixed order, not yet randomized or counterbalanced; prototype not for sale |
| Medication (e.g., zolpidem, melatonin, clonazepam) | Pharmacologically alters sleep onset or maintenance | Broadly available; can shorten sleep latency | Side effects include residual drowsiness, fatigue, memory and mood effects; does not precisely target REM circuits |
| Cognitive behavioral therapy for insomnia (CBT-I, a structured talk-based program) | Behavioral and psychological restructuring of sleep habits | Durable, guideline-recommended; no pharmacologic side effects | Requires trained providers and adherence; limited direct REM neuromodulation |
| Deep brain stimulation (STN) | Implanted electrodes inhibit STN at 60–120 Hz | Demonstrated REM improvements in Parkinson’s cohorts | Invasive surgery with long-term complication risk |
| Wearable trackers only | Accelerometry and surface EEG estimate stages | Low burden; useful for monitoring | No active neuromodulation; accuracy varies |
The comparison underscores the originality of NEUSLeeP: it is the first system reported to combine noninvasive deep-brain monitoring via ultrasound with stable electrophysiology for continuous natural sleep while targeting a specific subcortical node tied to REM regulation.
Limitations, Safety Notes, and What Comes Next
Any balanced coverage of a small, early-stage study must separate promising signals from premature conclusions.
- Study design limits causal certainty: Sham always preceded stimulation, increasing order and expectancy effects. The authors note the study was single-blind, not double-blind, without parallel control groups or an active stimulation control at a non-sleep site. Night-to-night variability alone can shift sleep metrics.
- Small, young sample: 26 usable datasets from adults 19 to 38. Generalizability to older adults or clinical populations remains untested.
- Acoustic targeting precision: At 72.2 millimeters depth, the beam narrows substantially when passing through the temporal bone, yet it still covers nearby structures such as the substantia nigra (SNr and SNc). The absence of blood flow or activity changes on the opposite side of the brain supports spatial selectivity, but finer resolution is a stated engineering goal.
- Research device status: The paper and the University of Texas at Austin news release describe NEUSLeeP as a prototype under patent application, being advanced with UT’s Discovery to Impact commercialization unit. It is not available for purchase and has not undergone the larger trials needed for clinical utility.
The team’s stated next steps include larger, randomized, double-blind trials with counterbalanced sham, active control conditions, concealed assignment, and pre-registered outcomes. Planned clinical directions include PTSD, depression, and chronic insomnia, building on the established links between disrupted REM and those conditions, plus expanded use for at-home monitoring, neuroscience research, and personalized sleep therapies. The study is registered on ClinicalTrials.gov as NCT07190287.
Frequently Asked Questions
How does a wearable ultrasound patch boost REM sleep?
The patch delivers low-intensity focused ultrasound through the skull to the left subthalamic nucleus, a small deep-brain hub connected to brainstem circuits that help orchestrate REM. The 100-pulses-per-second setting is designed to calm STN activity, which downstream may free up limbic and brainstem pathways such as the pedunculopontine nucleus (PPN) that regulate REM transitions. Because the array can steer its focus with timing adjustments and is coupled through a sticky hydrogel, it can maintain that targeting overnight while brain-wave sensors track sleep stages.
How much extra REM sleep did participants actually get?
Across the 26 participants with usable recordings, REM as a share of total sleep rose from 16.3% to 20.9%, adding about 15.9 minutes. Time to first REM dropped from 177 minutes to about 135 minutes, a 43-minute acceleration. Other stages and overall sleep efficiency did not change significantly.
Is the ultrasound sleep patch safe and comfortable?
In this study, yes within its limits. Participants rated the patch comfortable, skin redness scores were low after 8 hours, and vital signs did not shift meaningfully during morning-after sessions in a separate 16-person brain-imaging (fMRI) group. Safety was assessed with continuous blood oxygen and heart rate monitoring and follow-up checks one week later. Larger safety datasets are still needed before any home-use claims could be made.
Can this device treat insomnia, depression, or PTSD?
Not yet. The study included a small insomnia subgroup (mild insomnia, scoring 6 to 10 on the Pittsburgh Sleep Quality Index, a standard 0-to-21 questionnaire where 0 to 5 is considered good sleep) whose REM proportion improved similarly to healthy sleepers, but latency and autonomic benefits were less pronounced. The authors propose future trials in those conditions but emphasize this was a feasibility study, not a treatment trial. No conclusion about therapeutic efficacy can be drawn from 26 participants over two nights.
When will NEUSLeeP be available to buy?
There is no commercial release timeline. The technology is patent-pending and being explored with the university’s commercialization office. Any future product would require larger randomized trials, regulatory review, and manufacturing scale-up. For now, improving sleep hygiene, light exposure, consistent schedules, and evidence-based options such as CBT-I (cognitive behavioral therapy for insomnia) remain the most accessible approaches.
Conclusion
The NEUSLeeP patch offers a compelling proof of concept: a soft, adhesive, 103-gram device that quietly monitors sleep and nudges a deep brain circuit to deliver more and faster REM without drugs or surgery. Its early data are striking: 4.6 percentage points more REM, 43 minutes quicker to first REM, preserved sleep architecture, and in healthy adults, a notable lift in heart rate variability and selective modulation of emotional circuits.
Those signals come with honest caveats: a small sample, a fixed sham-first order, single blinding, and prototype status. The next chapter, as outlined by the UT Austin team and funders including DARPA’s REM-REST program, the NIH, and the NSF, is larger, rigorously controlled trials that test whether this platform can move from fascinating neuroscience toward real help for people whose nights, and days, depend on better REM.
