Controlled sweat generation via ultrasound stimulation integrated in a wearable device

Abstract

Wearable sweat biosensors enable noninvasive molecular monitoring, but reliable and comfortable on-demand sweat induction remains challenging. We developed a soft, skin-conformal wearable device that integrates ultrasound-assisted sweat generation with electrochemical sensing. Low-frequency ultrasound enhanced transdermal delivery of carbachol from a hydrogel reservoir, inducing localized sweat secretion under resting conditions without electrical current or skin penetration. Under the matched 15-min experimental window used here, ultrasound-induced sweating showed lower observed skin irritation and more favorable user-rated comfort than the tested iontophoresis setting. Integrated sensors measured uric acid, pH and K+ in ultrasound-induced sweat, and sweat uric acid showed a positive correlation with serum levels in a preliminary human study. These results support wearable ultrasound as a low-irritation strategy for controlled sweat generation and downstream biomarker tracking under resting conditions.

Reliable sweat induction remains a challenge for wearable sweat biosensing. Here, the authors develop a skin-conformal wearable device that uses ultrasound to induce sweat under resting conditions for electrochemical biomarker tracking.

Introduction

Wearable biosensors have been recognized as technologies designed to tackle modern healthcare challenges designed to tackle modern healthcare challenges with the continuous advancement of medical diagnostics^1–9. The rapid advancement of wearable sweat sensing technologies has been paralleled by growing recognition of sweat as a valuable medium for healthcare applications^10–15, due to the various biochemical markers in sweat related to physiological states that can be analyzed noninvasively^16–22. Effective sweating is crucial for continuous and reliable analysis. Therefore, it is of great significance to evaluate existing sweating methods and develop improved sweat collection technologies. Sweat secretion can be classified into natural, physical, and chemical methods according to the underlying mechanisms. Natural perspiration, which commonly occurs in areas with dense sweat glands such as the fingertips, is primarily driven by internal physiological processes, including thermoregulation and eccrine gland activation²³. Natural sweat can be passively collected at rest without external stimulation²⁴. However, the timing of sweat secretion cannot be actively controlled by the user. Sweating through exercise is a commonly used method to collect large volumes of sweat²⁵. However, both sweat rate and composition of this method are highly dependent on environmental conditions, physical exertion, hydration status, and individual physiology²⁶. Thermal stimulation induces sweating through external heat input^27–29, but may cause discomfort or overheating. Chemical methods enable on-demand sweat secretion through external chemical stimulation. Iontophoresis is a widely used electrical stimulation technique^30–32, which employs an electrical current to drive a cholinergic agonist (such as pilocarpine) into sweat glands^31,33–37. This method allows localized sweating under resting conditions and has been integrated into wearable systems with microfluidic collectors³⁸. However, electrical stimulation may damage the skin, corrode electrodes, and often cause discomfort. For example, the FDA-approved GlucoWatch was withdrawn from the market due to reports of skin irritation and blister formation during continuous use (although it is used to extract interstitial fluid, ISF, not sweat)³⁹. Chemically induced sweating can also be achieved using drug-coated microneedle patches⁴⁰. Studies have demonstrated that dissolvable microneedles loaded with pilocarpine can generate sweat volumes comparable to those produced by iontophoresis^40,41. However, this strategy compromises the noninvasive nature originally intended for wearable sweat-sensing platforms. Following insertion into the skin, microneedles may provide a more effective platform for interstitial fluid (ISF) extraction and analysis. Accordingly, developing a sweating strategy that offers controllable minimal skin irritation and full compatibility with wearable systems remains an important goal.

Ultrasound technology has recently gained attention for its noninvasive nature and minimal skin irritation^42–46. Ultrasound is a longitudinal pressure wave with frequencies above the audible range (>20 kHz). In clinical settings, high-frequency ultrasound (>1 MHz) is typically employed in applications such as ultrasonography, tumor ablation, and lithotripsy^47–49. In contrast, low-frequency ultrasound possesses unique properties, including the ability to transiently increase skin permeability and enhance transdermal transport^50,51. This low-frequency modality has been extensively applied for ISF extraction⁵², drug delivery⁵³, promoting wound healing⁵⁴ and tissue regeneration⁵⁵. Transdermal ultrasound does not rely on electrical current or mechanical skin penetration, thereby reducing pain and irritation. Leveraging these advantages, we aim to achieve sweating via ultrasound-mediated drug delivery and to integrate it with wearable sweat-sensing systems. This study introduces wearable ultrasound as an actuator for integrated on-device biofluid management, enabling controlled sweat generation and guided delivery to the sensing interface for downstream biomarker analysis.

In this study, we present a soft and skin-conforming wearable device that integrates ultrasound-based sweat generation with electrochemical sweat sensing. The system features a printed circuit board (PCB) housing both the ultrasound excitation and electrochemical sensing circuits, along with microfluidic channels and electrochemical sensors for tracking potassium, pH, and uric acid (UA) in induced sweat. By applying low-frequency, low-intensity ultrasound to gently stimulate sweating, the device avoids the discomfort associated with physical exertion or electrically induced methods. Ultrasound stimulation caused minimal skin irritation and improved overall user comfort under the tested conditions. Preliminary validation showed a positive correlation between sweat and serum UA concentrations (r = 0.90), supporting the potential clinical relevance of sweat UA as a noninvasive biomarker. This work incorporates ultrasound-induced sweating into a wearable sensing platform, highlighting the potential of wearable ultrasound for on-demand sweat generation and biomarker tracking under resting conditions.

Results

Design and Construction of the Wearable Ultrasound Sweat Collection and Sensing Platform. Figure 1a presents a system-level overview of the ultrasound-electrochemical hybrid wearable platform. The system comprised an ultrasound patch connected to a control circuit. As shown in Fig. 1a and S1, the patch consisted of several vertically stacked layers: (i) a sweat inlet with a drug delivery port, (ii) laser-induced graphene electrode arrays for biomarker detection, (iii) a microfluidic channel for sweat transport, (iv) a lead zirconate titanate (PZT) ultrasound transducer, and (v) a PZT placement zone with a sweat outlet. These components were assembled using adhesive to form a tightly sealed integrated structure. The control circuitry included an ultrasound drive module and an electrochemical detection unit. Ultrasonic activation was achieved as follows: a microcontroller unit (MCU) initiated the process by sequentially transmitting signals to both a pulse generator and a multiplexer. Upon receiving the trigger, the pulse generator produced excitation pulses that actuated the ultrasound transducer array. Simultaneously, the MCU activated the electrochemical sensing module and collected analog signals, which were digitized using the MCU’s built-in analog-to-digital converter (ADC). The digital data were transmitted in real time via a Bluetooth module to an external device (smartphone or computer), where a custom application processed and visualized the physiological information.

Fig. 1. Design of the fully integrated wearable system for ultrasound induced sweat and electrochemical detection.

a Schematic diagram of the layered structure of the device, which integrates an ultrasound module (upper left inset) consisting of a piezoelectric transducer array, microfluidic channels, and sensing electrodes, connected to a flexible printed circuit board (FPCB) containing an ultrasound drive module and an electrochemical detection module. b Schematic cross-sectional view of the mechanism of ultrasound-induced sweating. c Finite Element Simulation of Acoustic Field Distribution in the Integrated System. d Optical image showing the ultrasound transducer adhered to the skin surface. e Optical photograph of the flexible control circuit. f Schematic diagram of the system showing a sweat-sensing patch based on a piezoelectric transducer and microfluidics, integrated with flexible circuitry. g Literature-informed qualitative comparison of representative sweat-induction strategies for wearable applications, rather than a standardized quantitative comparison.

Figure 1b illustrates the operational mechanism of the ultrasound patch. Acoustic waves emitted from the transducer propagated through a hydrogel matrix and penetrated the skin, promoting the transdermal absorption of carbachol and inducing sweat secretion. Figure 1c shows finite-element simulations of the acoustic field distribution within the integrated patch, providing a spatial map of acoustic pressure in the hydrogel and near the skin. Following sweat generation, microfluidic channels guided the sweat into predefined pathways, reducing the risk of contamination and evaporation (Fig. 1d). Excess sweat was directed into a collection chamber via microchannels, ensuring consistent electrode coverage through glandular pressure. Electrochemical sensors integrated on the platform selectively detected key sweat biomarkers, including potassium ions, pH, and UA. As shown in Fig. 1e, these multimodal sensing capabilities were enabled by seamless integration with a flexible printed circuit board (FPCB) for simultaneous signal acquisition and downstream biomarker analysis (Fig. 1e, f). Figure 1g provides a literature-informed qualitative comparison of representative sweat-induction strategies for wearable applications, in which ultrasound is positioned as a low-irritation, wearable-compatible approach. (Table S1)

Ultrasound Induced Enhancement of Skin Permeability. We conducted both theoretical modeling and experimental validation to investigate molecular diffusion and increased skin permeability caused by ultrasound. Red dye was used as a tracer to monitor diffusion beneath the ultrasound transducer and assess molecular transport (Fig. 2a). Optical imaging captured the lateral spread and penetration depth of the dye, which were used to quantify diffusion intensity (Fig. 2b). Application of ultrasound significantly accelerated dye transport, as evidenced by earlier onset of dispersion and deeper skin penetration. The ultrasound patch maintained a stable frequency and amplitude during activation, which were tuned to match the resonant frequency of the piezoelectric material (0.6 MHz, Fig. 2c). The driving frequency of 0.6 MHz was selected to closely match the resonant frequency of the piezoelectric element, considering both the layered structure and skin coupling, thereby ensuring efficient acoustic excitation and stable ultrasound output. In addition, we tested higher frequencies (1–2 MHz) and lower frequencies (200 kHz). Higher and lower frequencies were generally less effective in inducing sweating, indicating that 0.6 MHz provides a favorable balance for effective and stable ultrasound-induced sweating.

Fig. 2. Characterization of ultrasound performance.

a Optical images showing the effect of ultrasound on the accelerated diffusion of a colored dye in water. b Quantitative analysis of the effect of ultrasound on dye diffusion in water. c Representative waveform of the ultrasonic waves emitted by the ultrasound patch during operation at 0.6 MHz. The measurement was repeated independently three times with similar results. d Finite element analysis (FEA) of the acoustic pressure distribution of the piezoelectric transducer on a human body. e Schematic of skin impedance measurement. f, g show the changes in skin impedance under ultrasound stimulation at a frequency of 0.6 MHz. Data in (f, g) are presented as mean ± SD from n = 3 independent experiments.

Mechanical vibrations were generated through the piezoelectric effect when the PZT sheet was driven by alternating current, emitting ultrasonic waves (Fig. S2). These results confirmed the system’s ability to produce a stable ultrasound field at the skin interface, which supports effective stimulation of sweat gland activity. To optimize acoustic performance, a finite element analysis (FEA) was used to simulate acoustic pressure distribution across the skin surface. The simulated field extended up to 4 mm beneath the transducer. Acoustic attenuation was visualized in both the X-Y and Y-Z planes (Fig. 2d). Simulations revealed a uniform pressure distribution with a central peak, indicating consistent exposure across the treated area. In the right-hand sub-panels of Fig. 2d, the upper image shows the acoustic pressure distribution at the position of the first dashed line in the left panel, while the lower image illustrates the distribution along the centerline from the first dashed line to the bottom. Both demonstrate the attenuation of acoustic pressure in the X–Y and Y–Z planes. Since lower skin impedance is closely associated with enhanced permeability and improved drug delivery, we developed an electrical model to evaluate impedance changes (Fig. 2e)⁵⁶. A direct voltage was applied across two skin-contacting electrodes, and total resistance was measured across the stratum corneum, epidermis, and dermis. In the absence of ultrasound, skin impedance measured approximately 270 kΩ (Fig. S3). Upon ultrasound activation, impedance rapidly decreased from ~260 to ~60 kΩ (Fig. 2f). This significant reduction, together with the observed dye diffusion (Fig. 2g), confirmed that ultrasound enhanced skin permeability, thereby improving the efficiency of transdermal delivery and enabling effective sweating. Together, the pronounced reduction in skin impedance under ultrasound and the accelerated dye transport observed beneath the ultrasound transducer in the model system demonstrate that ultrasound facilitates mass transfer across the interface under our operating conditions. The impedance decrease suggests a transient weakening of the skin barrier, while the faster dye spreading under identical concentration gradients indicates an increased effective transport rate. These findings consistently point to an ultrasound-induced enhancement of transdermal delivery capacity, even though the precise microscopic mechanism remains to be fully elucidated.

Comfort and safety assessment of ultrasound patch. A model system with agarose-based hydrogel drug carriers was established to evaluate the functional performance of the ultrasound patch in transdermal delivery. Fluorescent dyes, including fluorescein isothiocyanate (FITC), calcein, and rhodamine B, served as model drugs (Fig. 3a and S4). As shown in the fluorescence images, these compounds were transported from the hydrogel matrix into porcine skin along the ultrasound propagation direction. The fluorescence intensity beneath the skin in the ultrasound-treated groups was significantly higher than that in the untreated control (US 5 V, P = 0.0417; US 15 V, P = 8.81 × 10⁻⁵; US 25 V, P = 7.93 × 10⁻⁶; two-tailed Student’s t-test; Fig. 3b). Under identical dye concentrations and exposure times, the ultrasound-treated skin exhibited higher fluorescence intensity and deeper penetration of FITC compared with the untreated control, indicating an increased transdermal transport rate across the skin barrier. This accelerated intradermal diffusion is consistent with an effective enhancement in skin permeability under ultrasound stimulation. HE staining of ex vivo porcine skin revealed distinct structural differences between the control and ultrasound-treated groups. The control skin exhibited compact and well-organized tissue architecture, whereas the ultrasound-treated skin showed mildly loosened collagen fibers and enlarged inter-fiber spaces (Fig. 3c). These structural features reflect a transient increase in tissue permeability that facilitates transdermal drug delivery. Consistent with the impedance measurements and the fluorescence-diffusion results in porcine skin, this mildly loosened yet continuous dermal architecture further supports a transient increase in skin permeability without overt structural damage under ultrasound stimulation. A FEA was performed to simulate the temporal evolution of the liquid volume fraction within the microfluidic channel (Fig. S5). The simulation indicated complete wetting of the sensing region within approximately 210 s at a sweat inflow rate of 1.5 µL·min⁻¹. During actual testing, the microfluidic patch efficiently sampled sweat with high temporal resolution (Fig. 3d, e). With 1.5 µL·min-1 as the inlet flow rate, the refreshing time taken to reach 90% of the new solute concentration was around 240 s, for a change of solute concentration from 0 to 0.12 mol·m⁻³ (Fig. 3f and g).

Fig. 3. Ultrasound for sweat induction.

a Diffusion depth of fluorescein isothiocyanate (FITC) dye in porcine skin observed by confocal microscopy under different ultrasound intensities. b Statistical analysis of green fluorescence intensity. (mean ± SD; n = 6 independent samples per group; two-tailed Student’s t-test for pairwise comparisons (vs. control); significance as indicated). c HE staining results of ex vivo porcine skin after ultrasound exposure. d Photographic image showing microfluidic sweat sampling during simulated perspiration. e The percentage of liquid present in the microfluidic channel during the collection process. f Simulations of the solute concentration distributions over the bottom surface of the microfluidic channel at different time points, with an inlet flow rate of 1.5 µL·min⁻¹. g Average solute concentration in a microfluidic channel during finite element simulations. h Surface temperature distribution of the ultrasonic patch during operation. (Data are presented as the mean ± SD of n = 3 independent experiments). i Evaluation of skin irritation intensity after application of ultrasound patch and commercially available iontophoresis device. (Data are presented as the mean ± SD of n = 15 independent experiments). j Cytotoxicity and biocompatibility test results. (Data are presented as the mean ± SD of n = 3 independent experiments).

The self-heating effect of the transducer was also examined. As shown in Fig. 3h and S6, the temperature of the transducer increased modestly from 25.3 ± 2.68 to 26.85 ± 1.2 °C during a single operation cycle. This minor rise suggested negligible thermal risk to the skin and is far below the threshold required to stimulate sweating, confirming that the observed sweat secretion was not thermally induced but instead originated from ultrasound-mediated transdermal delivery. To further exclude the possibility that ultrasound exposure alone triggers sweating, we performed an agonist-free negative control by replacing the carbachol-loaded hydrogel with a K⁺-containing hydrogel of matched ionic strength; no visible skin wetting/sweat droplets were observed under otherwise identical ultrasound conditions (Fig. S7). To determine the appropriate ultrasound induction duration, we compared sweat secretion under 5, 10, 15, and 20 minutes of ultrasound application (Fig. S8). Sweat was not detectable at 5 minutes (0 ± 0 µL), increased modestly to 5.40 ± 0.27 µL at 10 minutes, and reached a stable plateau at 15 minutes with a sweat volume of 10.45 ± 0.52 µL, with no further improvement observed at 20 minutes (10.35 ± 0.52 µL). Based on these results, 15 minutes was selected as the standard induction duration for all subsequent experiments. This matched window was used to compare the practical performance of the two on-body sweat-induction modes within the same wearable-session duration, rather than to represent the individually optimized operating conditions of each modality. For comparison, both ultrasound- and iontophoresis-based sweat induction were performed under matched experimental settings, while noting that the two methods rely on distinct transport mechanisms. The ultrasound module operated at approximately 5 mW, while the commercial iontophoresis device applied a current of 1.5 mA at 5–10 V ( ≈ 7.5–15 mW). The skin conditions following both ultrasound- and iontophoresis-induced sweating was subsequently compared (Fig. S9, blue circles). Ultrasound treatment produced minimal erythema and swelling, with no visible rash. In contrast, under our test condition (1.5 mA, 15 min), iontophoresis produced more pronounced transient erythema after a single stimulation, and follow-up imaging showed that the redness could persist for several days (Fig. S10). We note that skin responses to iontophoresis depend on stimulation duration, current, electrode, gel formulation, and individual sensitivity; therefore, this comparison is intended to benchmark comfort under a matched 15-min, wearable-relevant induction window rather than to generalize across all clinical iontophoresis protocols. For the one-week repeated-use evaluation, the ultrasound patch was applied once per day to the same forearm location. Each session consisted of 15 minutes of continuous ultrasound stimulation, using the same operating parameters as in the single-session induction experiments. Under this regimen, no erythema, rash, or barrier disruption was observed after seven consecutive days, and the skin impedance remained stable throughout the entire period (Fig. S11), indicating the absence of cumulative irritation. In addition, transepidermal water loss (TEWL) measurements in six volunteers showed a transient increase immediately after ultrasound stimulation, followed by a gradual recovery toward baseline within 30 min (Fig. S12), supporting the reversible nature of the ultrasound-induced barrier perturbation. Because iontophoresis-induced erythema lasted for several days, repeated stimulation at the same site was not ethically feasible; therefore, long-term repeated-use evaluation was performed only with ultrasound while maintaining comparable single-session stimulation duration for both methods.

To provide an objective assessment of skin irritation, we calculated the erythema index (EI) using $EI=100\left(R-G\right)/\left(R+G\right).$The relative EI values were 0.964 ± 0.087 for ultrasound and 1.242 ± 0.118 for iontophoresis, indicating milder erythema after ultrasound stimulation under the matched comparison condition used in this study (Fig. 3I). Cytotoxicity was assessed using the CCK-8 assay to evaluate long-term biocompatibility (Fig. 3J). The CCK-8 assay results demonstrated cell viabilities of 88.23 ± 5.162% at 24 hours, 92.17 ± 0.566 % at 48 hours, and 91.86 ± 9.122 % at 72 hours, indicating good cytocompatibility associated with the material and device. Consistently, live/dead cell staining (Fig. S13) further confirmed the predominance of viable cells with negligible apoptosis, supporting the favorable cytocompatibility of the system. To address the actual bio-interface, cytotoxicity tests were performed using extracts of the agarose hydrogel matrix that directly contacts the skin, which exhibited high cell viability (Fig. S14). Furthermore, to simulate the operational state, the cytotoxic profile of the PZT material was assessed under continuous ultrasound activation, with results confirming that the device operation does not introduce additional cytotoxic effects (Fig. S15). In addition, the spatial-peak temporal-average (SPTA) intensity of the ultrasound patch was calculated to be approximately 130.7 mW·cm⁻² under our driving conditions (0.6 MHz). This value is well below the FDA safety limit for diagnostic ultrasound exposure (720 mW·cm⁻²), confirming the safety of our stimulation parameters (see Supplementary Information for detailed calculation).

Structural optimization and electrochemical performance evaluation of a wearable ultrasonic sweat sensing platform. Figure 4a presents the core components of the ultrasound patch, comprising a PZT transducer, electrochemical sensing electrodes, and a PDMS based microfluidic chamber for structural support (Figs. S16 and S17). FEA assessed mechanical resilience under a 150 N tensile load applied to the left side of the patch (Fig. 4b). The copper serpentine interconnects and PDMS substrate effectively dispersed the stress concentration. A compression test further confirmed the patch’s mechanical robustness and flexibility (Fig. S18). The graphene-based sensor array was fabricated using a CO2 laser engraving method^12,57. The surface morphology revealed a porous and conductive structure, which can provide a high specific surface area for electrochemical sensing applications (Fig. S19). The array was configured to detect K⁺, pH, and UA in sweat. To ensure accurate sensing, it was essential to prevent interference between the electrochemical sensors and ultrasound transducers. To mitigate potential crosstalk, the sensors were spatially separated from the ultrasound components (Fig. 4c and Fig. S20). The microfluidic inlet is positioned adjacent to (rather than directly above) the hydrogel reservoir to minimize the possibility of hydrogel contamination. After ultrasound-assisted delivery into the skin, the agonist can diffuse within the local tissue microenvironment; therefore, induced sweating is not necessarily restricted to only the area directly beneath the transducer and can be collected from the neighboring inlet region. As the transducers operated with high-frequency pulses, signal drift in adjacent sensors was a potential concern. During the test, the ultrasound patch was placed directly on top of standard analyte solutions, and electrochemical signals were continuously recorded under stable conditions. The transducer was intermittently switched on and off in repeated cycles to directly assess whether ultrasound activation influenced sensor output. Electrochemical signals were recorded during 20-second ultrasound activation and deactivation cycles to monitor interference. These results demonstrate that ultrasound excitation did not affect electrochemical sensing accuracy (Fig. 4d–f).

Fig. 4. Electrochemical performance of wearable ultrasound patch.

a Schematic diagram of the ultrasound patch. b Finite element simulation of the ultrasound patch under tensile deformation. c Ultrasound patch applied to human skin. Crosstalk of the electrochemical signals of pH d & K⁺ e (OCP) and UA f (DPV) to ultrasound. The open circuit potential responses of the pH g and K⁺ h sensors, and the differential pulse voltametric response of the UA i sensor, along with their respective linear calibration curves.

Figure S21 schematically illustrates the sensing mechanisms for each analyte. The K⁺ and pH sensors operate based on potentiometric detection, where ion-selective and pH-sensitive membranes induce voltage shifts relative to a reference electrode in response to target ion concentrations. The UA sensor employs differential pulse voltammetry on an Au-decorated graphene working electrode, where UA is electro-oxidized at the modified electrode surface to generate a current proportional to UA concentration. All functional modules, including the electrochemical detection and ultrasound control circuits (Fig. S22), were integrated onto the FPCB, enhancing both wearability and functionality (Fig. S23). The performance of each sensor was tested using standard solutions containing the corresponding analytes (Fig. 4g–i). Figures 4g and 4h show the representative open-circuit potential responses of the pH and K⁺ sensors, measured in standard solutions ranging from pH 4 to 8 and K⁺ concentrations from 1 to 32 mM, respectively. A clear linear relationship was observed between voltage and analyte concentrations, with sensitivities of 65.46 mV·pH⁻¹ for the pH sensor and 54.05 mV·dec⁻¹ for the K⁺ sensor. For UA detection, the electrode was tested in UA standard solutions ranging from 2 to 128 μM, exhibiting a prominent oxidation peak at approximately 0.26 V, corresponding to a sensitivity of 0.03199 μA·μM⁻¹ (Fig. 4i). In addition, the three sensors showed good anti-interference capability against common sweat constituents and electroactive species (Fig. S24). Continuous measurements over 80 min demonstrated stable outputs with minimal signal drift (Fig. S25), and the reproducibility of independently fabricated sensors showed highly consistent normalized sensitivities among three devices for each analyte (Fig. S26). The limits of detection were estimated to be 5.3 μM for K⁺, 0.19 μM for UA, and 0.10 pH for pH sensing. The UA LOD (0.19 μM) is comparable to representative wearable sweat UA sensors reported previously and is sufficiently low for detecting low UA levels relevant to resting sweat (Table S2).

Sweat Biomarker Tracking and User Experience. The wearable integrated ultrasound-electrochemical patch was applied to the forearm of a healthy volunteer to demonstrate its capability to induce and monitor sweat secretion under resting conditions (Fig. 5a). The ultrasonic drive module activated the PZT element on demand to induce sweating. Subsequently, the electrochemical sensing module processed the biomarker signals in sweat, and the data were wirelessly transmitted to the user interface via a Bluetooth module (Fig. 5b). First, we evaluated the comfort of ultrasound-induced sweating. Volunteers rated their experiences with exercise-induced sweating, ultrasound-induced sweating, and iontophoresis using VAS (Figs. S27 and Fig. S28)^58–61, along with assessments of erythema and thermal sensation. Fifteen participants scored each method from 0 (most satisfied) to 10 (least satisfied) (Fig. 5c). As shown in Fig. 5d, ultrasound-induced sweating caused only mild erythema compared with exercise, yet it offered superior comfort and thermal sensation relative to both exercise and iontophoresis. During ultrasound stimulation, the skin surface temperature slightly decreased due to sweat evaporation compared with baseline conditions (Figs. S29 and S30), and no noticeable effects were observed after device removal aside from residual sweat (Fig. S31), further supporting the comfort and compatibility of the wearable patch during prolonged use. A comparative experiment was then conducted on four participants to assess whether biomarker concentrations differed between ultrasound- and iontophoresis-induced sweat (Fig. 5e and S32). Additional comparative results between ultrasound- and iontophoresis-induced sweating, including affected area, relative drug depletion, area-normalized sweat volume, and sweat duration, were obtained from a controlled experiment (Fig. S33). To further clarify the kinetic differences between the two stimulation modes, we also analyzed stimulation-duration-dependent area-normalized sweat output and sweat onset time under identical carbachol hydrogel concentration (Figs. S34 and S35). We further analyzed the time-resolved area-normalized sweat rate for both iontophoresis and ultrasound under the same conditions (Fig. S36), providing a kinetically resolved functional comparison between the two stimulation modes. These results suggest that, despite a slower onset than iontophoresis, ultrasound can achieve comparable area-normalized sweat output under optimized stimulation conditions. In addition, the relative decrease in hydrogel carbachol concentration after 15 min of ultrasound stimulation was quantified separately as an indirect indicator of drug release from the hydrogel (Fig. S37). The extracted data revealed no significant differences in biomarker concentrations between the two stimulation methods, confirming the reliability of ultrasound for sweat induction. A controlled dietary study involving four participants was further performed to verify the clinical applicability of ultrasound-induced sweating for noninvasive metabolic monitoring. Dietary intake of purine-rich food and alcohol is known to elevate serum UA levels (Fig. 5f), and this metabolic change can also be reflected in sweat composition⁶². Elevated UA levels were detected in both serum and sweat after overnight fasting followed by a purine-rich meal, and in a separate human trial five hours after a standard lunch, repeated paired measurements collected over a 300-min window showed that sweat UA tracked blood UA over time within each participant (Fig. 5g and S38). The positive correlation coefficient (r = 0.90) between serum and sweat UA concentrations (Fig. 5h and S38) supports the feasibility of using ultrasound-induced sweat UA as a biomarker for noninvasive metabolic analysis. Overall, within the matched wearable-session window used here, participants rated ultrasound-induced sweating as more comfortable, supporting the practical potential of controlled acoustic stimulation as a low-irritation approach for on-demand sweat induction.

Fig. 5. Wearable ultrasound electrochemical patch in vivo performance: biomarker tracking and user feedback assessment.

a Photograph of a subject wearing the device at rest. b System level block diagram showing the flow of power and signals in the wearable ultrasound electrochemical patch, from the power source through the transducer and biosensors to the user interface. c Ratings for each scale range from 0 (most satisfied) to 10 (least satisfied). d Scale ratings from 15 volunteers comparing the user experience of the ultrasound patch, a commercially available iontophoresis device, and exercise induced sweating. In (d), box plots show the median as the centre line, the 25th and 75th percentiles as the box bounds, and whiskers extend to the most extreme data points within 1.5× the interquartile range; open squares indicate mean values, and individual points represent scores from n = 15 independent participants. e Three participants were monitored during 80 minutes of daily activities (talking, working, and resting). For both ultrasound-induced and iontophoresis-induced sweating, a 15-minute stimulation period was used for drug delivery, followed by naturally sustained sweating during the 80-minute monitoring phase. Each participant underwent both stimulation modes in a controlled comparative experiment, and sweat samples were collected under each condition for analysis. f Schematic diagram of UA metabolism. g Variation in sweat and serum UA levels in four subjects within 5 h following the intake of a purine-rich meal. During this experiment, ultrasound stimulation was reapplied every 60 min to repeatedly induce sweat secretion during the monitoring period. h Correlation between ultrasound-induced sweat and serum UA concentrations in human subjects.

Discussion

In this study, we have developed a soft, skin-conformal wearable platform that integrates ultrasound-assisted sweat generation with multiplexed electrochemical sensing, providing a noninvasive and user-friendly strategy for on-demand sweat induction and downstream biomarker tracking under resting conditions. This platform highlights wearable ultrasound as an effective actuator for integrated on device biofluid management, enabling active sweat generation and guided transport to the sensing interface for downstream analysis. Finite element simulations and experimental validation confirmed that the system generated stable ultrasonic output. The significant reduction in skin impedance, which reflects enhanced permeability of the stratum corneum, together with accelerated dye diffusion, validated the platform’s efficacy in facilitating transdermal drug delivery. By leveraging low-frequency ultrasound to mediate the transdermal delivery of carbachol, the device achieved a stable sweat secretion under resting conditions. Under the matched 15-min comparison condition used in this study, ultrasound-based sweat induction showed lower observed skin irritation and more favorable subjective comfort ratings than the tested iontophoresis setting, as evidenced by lower skin redness scores and VAS ratings across 15 participants. This difference is consistent with the distinct stimulation modes of the two approaches: iontophoresis relies on electrical current to drive transdermal delivery, whereas ultrasound delivers energy through a mechanical acoustic field without passing current through the tissue. Rather than establishing superiority under each modality’s individually optimized operating conditions, these results support ultrasound as a low irritation, wearable-compatible strategy for on demand sweat generation at rest. Integrated electrochemical sensors enabled selective detection of UA, pH, and K+ in ultrasound-induced sweat. Biomarker concentrations in ultrasound-induced sweat closely matched those obtained from iontophoresis-induced sweat, and UA levels exhibited a positive correlation with serum values (r = 0.90), supporting the clinical relevance of sweat-based diagnostics. Beyond sensing performance, the platform’s lightweight construction, mechanical flexibility, and low-power operation support its suitability for extended on-body use. Given the widespread clinical application of ultrasound in physical therapy, wound healing, and drug delivery, this work lays the foundation for future development of closed-loop systems integrating both biosensing and therapy. Such advancements could facilitate wearable biomarker tracking and timely intervention, further supporting the development of next-generation personalized and digital healthcare.

Methods

Materials

All reagents used in this study were of analytical grade and were used without further purification. Ultrapure water (18.2 MΩ·cm) was prepared using a Milli-Q system. Polydimethylsiloxane (PDMS, Sylgard 184) was obtained from Dow Corning. Aniline, potassium chloride (KCl), and carbachol were purchased from Macklin Biochemical Co., Ltd. (Shanghai, China), while hydrochloric acid (HCl) was supplied by Sinopharm Chemical Reagent Co., Ltd. Polyvinyl chloride (PVC), sodium tetraphenylborate (NaTPB), valinomycin, polyvinyl butyral (PVB), cyclohexanone, and UA were sourced from Tianjin Heowns Biochemical Technology Co., Ltd. FITC was acquired from Aladdin Biochemical Technology Co., Ltd. Multi-walled carbon nanotubes were provided by Jiangsu XFNANO Materials Tech co., Ltd. Normal human dermal fibroblasts (NHDF) were obtained from Shenzhen HuiNuo Biotechnology Co., Ltd. The Cell Counting Kit-8 (CCK8) reagent kit was procured from GLPBIO Inc. (United States). PET membranes were supplied by East China Composite Insulation Filter Fabric Factory (Shanghai, China), and double-sided adhesive tapes were provided by Deyi Tape (Shenzhen, China). An ultrasonic coupling agent was purchased from Jinyang Medical Materials Factory (Ningjin, China). The piezoelectric (PZT) ultrasonic transducer was obtained from Dihui Electronics Co., Ltd. (Shenzhen, China), and high-temperature resistant resin (Yellow-20) was supplied by BMF Ltd. The propidium iodide (PI) was purchased from Shanghai Beibo Biotechnology Co., Ltd. The hematoxylin–eosin (H&E) staining kit was obtained from Beijing Solarbio Science & Technology Co., Ltd. TEWL was measured using a VAPO SCAN device (ASCH JAPAN Co., Ltd., Tokyo, Japan). Fresh ex vivo porcine skin was obtained from a local commercial supplier.

Fabrication of the microfluidic sweat collection module

The microfluidic module was fabricated through a multi-step process. The outlet and channel layer were prepared by casting polydimethylsiloxane (PDMS) into a mold (26 mm × 26 mm × 2 mm) fabricated via projection micro-stereolithography (PμSL) 3D printing. The inlet layer, designed for microparticle delivery, was patterned directly onto a polyethylene terephthalate (PET) film (26 mm × 26 mm × 0.5 mm) using standard laser engraving techniques. After curing, both the PDMS structure and PET film were ultrasonically cleaned for 5 min, followed by air plasma treatment at 100 W for 2 min. The final assembly was completed by laminating the PET film, PDMS layer, and double-sided adhesive tape to form the integrated microfluidic module.

Construction of the ultrasound-electrochemical hybrid wearable platform

The ultrasound-electrochemical hybrid wearable platform was constructed through a series of integrated steps. Initially, five piezoelectric transducers (5mm-diameter PZT) were connected in parallel to ultrasonic driver board using positive and negative copper wires, which were laser-engraved into serpentine patterns. Then, these piezoelectric transducers were fixed within the fabricated PDMS. The driver board itself integrated commercially available electronic components, including a power supply, a power management system, and a microcontroller. The wearable system was powered by a rechargeable lithium battery (5 V, 2000 mAh) connected to the driver board. It converts low-frequency input signals into high-frequency output signals using an integrated signal generator and voltage amplifier. These amplified high-frequency signals drive the piezoelectric transducers to emit ultrasonic waves, which in turn facilitate the transdermal delivery of carbachol from the hydrogel on the skin surface into the subcutaneous layer, thereby inducing sweat secretion.

Statistical analysis

Data are presented as mean ± SD unless otherwise stated. Statistical comparisons were performed using a two-tailed Student’s t-test only where indicated (see figure captions). A p value < 0.05 was considered statistically significant.

Ethics

Human experiments were carried out following a protocol approved by the Animal and Human Ethics Committee of Shenzhen University (protocol number PN-202500095). Each participant gave written informed consent before enrollment. Consent to publish information that identifies individual participants was not obtained because no identifiable participant information is presented in the manuscript or Supplementary Information. Photographs were selected or cropped to avoid identifiable facial or personal features, and human-subject data are reported in aggregate or as group-level statistics. A total of 15 healthy volunteers, aged 23–32 years, participated in the ultrasound-induced sweating-related human experiments described in the manuscript and Supplementary Information. Sex information was not collected and was not used as a variable in the study design or analysis. Participants were not compensated. No live animal experiments were performed; ex vivo porcine skin was used only for transdermal diffusion and histological evaluation, and animal ethics approval was not required.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

L.C., X.H., B.S. and J.W. conceived the project. G.Z., Y.J. and Z.Z. developed the methodology. L.C., X.H. and B.S. carried out the experiments and investigations. T.X. designed the project, supervised the work, and secured funding. Litong Chen wrote the original draft. All authors contributed to reviewing and editing the manuscript and approved the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks Wang Minqiang, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the Shenzhen Science and Technology Program (JCYJ20240813142503006), the National Natural Science Foundation of China (22427806), the Synthetic Biology Research Center of Shenzhen University, the Key Laboratory of Marine Microbiome Engineering of Guangdong Higher Education Institutes (2024KSYS011), and the Shenzhen Overseas Talent Program.

Data availability

All data supporting the findings of this study are available within the article and its supplementary files. Any additional requests for information can be directed to and will be fulfilled by the corresponding authors. Source data are provided with this paper.

Code availability

No custom code was used in this study.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73789-4.