−30°C-operable bioadhesive hydrogel sensors for embryonic-like skin regeneration and real-time wound monitoring Authors: Linhao Hou, Wei Jing, Xueying Zhang, Pengfei Wei, Fanqi Hu, Dingfei Qian, Wenchao Wang, Bo Zhao, Yiqian Huang, Xuesong Zhang, Xinyu Liu, Yuguo Chen Published: 2026 ABSTRACT Hydrogel-based biosensors are promising for on-skin and in situ wound monitoring owing to their softness and biocompatibility, yet their practical deployment remains hindered by subzero failure caused by water crystallization, unstable adhesion on moist/exudative tissues, and limited ability to regulate the wound microenvironment. Here, we report an antifreezing, electroactive, and robust adhesive hydrogel bioadhesive sensor based on polyethylene glycol, a glycerol/water binary solvent, and carboxylated carbon nanotubes (PEG/Gly/CNT). The glycerol-water hydrogen-bonding network suppresses ice crystallization, enabling stable operation at -30°C, while an optimized 3 wt% CNT percolation network provides electroactivity and mechanical reinforcement. The resulting sensor exhibits strong wet-tissue adhesion (41.50 ± 1.62 kPa) and maintains high conductivity (2.07 ± 0.20 S/m) and high strain sensitivity (GF = 2.61) at -30°C, supporting wide-range linear strain sensing with rapid and stable signal readout under deformation. Beyond monitoring, the PEG/Gly/CNT facilitates microenvironmental regulation by endogenous bioelectrical signaling transmission at the wound interface. In a diabetic full-thickness skin wound model, the sensor enables continuous electrical readouts during healing and significantly accelerates wound closure (99.63 ± 0.41% recovery on day 12), accompanied by increased collagen type III deposition and an elevated MMP-9/α-SMA ratio, indicating embryonic-like remodeling. Moreover, conformal adhesion to dynamic tissues and real-time signal acquisition are demonstrated on Bama miniature pig cardiac wounds. This subzero-operable, bioadhesive, and electroactive hydrogel platform addresses key limitations of current hydrogel biosensors and offers reliable deployment and sensing in extreme-cold environments, while supporting tissue repair under physiological conditions and enabling cardiovascular monitoring. Introduction In recent years, hydrogel-based biosensors have attracted extensive attention in wound monitoring owing to their high water content, excellent flexibility, and superior biocompatibility [[1], [2], [3], [4], [5]]. However, their performance remains severely challenged under low-temperature conditions. When the temperature falls below 0°C, the water molecules within hydrogels readily freeze, resulting in volumetric expansion, microstructural damage, and a significant decline in electrical conductivity. These changes greatly restrict their stable application in outdoor or cold-climate wound care scenarios [[6], [7], [8], [9]]. In addition, conventional hydrogels exhibit inherently weak tissue adhesion, and their adhesive strength can drop by more than 60% at low temperatures, leading to detachment or adhesive failure. This instability compromises both sensor fixation and the accuracy of signal acquisition [10]. More importantly, most existing biosensors enable passive signal monitoring, such as temperature, strain, or pH, while lacking the capacity to regulate the wound microenvironment. Accumulating evidence indicates that endogenous electric fields play an essential role in directing cell migration, stimulating angiogenesis, and promoting tissue regeneration [11]. Nevertheless, current hydrogel-based biosensors are unable to induce or reconstruct these bioelectric signals. Therefore, developing a hydrogel-based biosensor with integrated antifreezing, strong adhesion, and electroactivity is of great significance. Such a system could not only enable real-time monitoring of the wound-healing process but also support local endogenous bioelectrical signaling to promote tissue regeneration. To mitigate the freezing limitations of hydrogel-based biosensors under low-temperature conditions, a variety of antifreezing strategies have been proposed to enhance stability and function in extreme environments. A common approach involves introducing inorganic salts (e.g., LiCl, NaCl) to depress the freezing point by leveraging the high osmotic pressure of ionic solutions to suppress ice crystallization [12,13]. However, such systems are prone to ion migration, leaching, and loss of mechanical integrity [14,15]. Another strategy is to construct polymer networks rich in multiple hydrogen bonds and/or hydrophobic interactions (e.g., poly(vinyl alcohol), polyacrylamide copolymers) to hinder ice growth by strengthening interchain associations. Yet this often increases network rigidity at the expense of flexibility and electrical conductivity [16]. Small-molecule cryoprotectants (e.g., ethylene glycol, dimethyl sulfoxide) have also been explored, but potential cytotoxicity limits their safe use in wound environments [17]. By contrast, glycerol/water (Gly/H2O) binary solvents have emerged as a standout solution, balancing robust antifreezing capability with excellent biosafety [18]. Glycerol forms extensive hydrogen bonds with water molecule, effectively suppressing ice nucleation and growth, thereby allowing hydrogels to remain soft and conductive at temperatures as low as -20°C and below. For example, Wang et al. prepared PVA/PAM/SA/Gly hydrogels, owing to glycerol's ice-inhibiting effect, they observed an endotherm shift to lower temperatures with the disappearance of the crystallization peak, enabling the hydrogel to remain stretchable at -20°C [19]. Moreover, glycerol's intrinsic humectancy and biocompatibility help retain moisture and maintain a favorable wound milieu [20]. Therefore, employing a Gly/H2O solvent system to build antifreezing hydrogel biosensors not only markedly improves low-temperature operability, but also ensures safety and stability, providing an ideal platform for real-time wound monitoring and regenerative therapies in cold environments. Beyond antifreezing performance, hydrogel-based biosensors also suffer from insufficient adhesion to wound tissues, particularly in moist or highly exudative environments, leading to detachment and unstable signals [21,22]. To enhance interfacial adhesion, several strategies have been explored. One approach introduces biomimetic catechol-containing moieties (e.g., dopamine, catechol) that form covalent or hydrogen bonds with thiol, amine, or carboxyl groups on tissue surfaces, thereby strengthening attachment [23,24]. Another incorporates reversible dynamic bonds to achieve robust wet adhesion and self-healing in hydrogel networks [25,26]. However, these methods can be limited by controllability, biosafety, and long-term stability. Recently, polyethylene glycol N-hydroxysuccinimide (PEG-NHS) based hydrogels have gained attention owing to their excellent biocompatibility, tunable crosslinking, and high reactivity toward tissue-surface amines [27]. Ece et al. reported that PEG-NHS hydrogels achieved shear adhesion strengths of up tp ∼90 kPa on skin and tracheal tissues [28]. PEG-NHS thus enables rapid and cytocompatible tissue adhesion while providing a versatile matrix for the construction of hydrogel bioadhesive sensors. In parallel, embedding carbon nanotubes (CNTs) into hydrogel networks improves mechanical toughness and electrical conductivity, yielding electrically active interfaces [29]. Such composites not only support real-time monitoring of physiological parameters (e.g., strain, bioelectrical signals), but also faciliatates endogenous bioelectrial signal transmission at wound sites to guide cell migration, thereby accelerating tissue regeneration [30]. Compared with conventional non-drug-delivering biomaterials that function primarily as passive scaffolds, electroactive composites enable an integrated sensing-feedback-regulation modality, offering distinctive advantages for intelligent regenerative medicine. Existing studies have mainly focused on isolated properties such as antifreezing behavior, tissue adhesion, or strain sensing, whereas the integration of these functions with bioelectric microenvironment regulation for regenerative wound care remains limited. Here, we report a hydrogel-based bioadhesive sensor that combines antifreezing performance, robust tissue adhesion, and high sensitivity. Built on an adhesive PEG matrix and a Gly/H2O binary solvent system, the bioadhesive sensor incorporates carboxylated CNTs to form a stable conductive network. Comprehensive characterization shows that the PEG/Gly/CNT bioadhesive sensor maintains excellent mechanical compliance under extreme cold (-30°C) with stable electrical response, and exhibits a highly linear signal in the small-strain regime. Notably, the CNT network provides a conductive interface that facilitates endogenous bioelectrical signal transmission at the wound site, promoting embryonic-like healing and thereby accelerating tissue repair (Scheme 1). The bioadhesive sensor achieves concurrent dynamic monitoring and bioelectric modulation in a full-thickness skin wound model, facilitating embryonic-like regenerative remodeling and rapid wound closure, and enabled precise acquisition of cardiac motion-associated resistance signals in Bama miniature pigs. Collectively, this antifreezing, strongly adhesive, and electroactive hydrogel biosensor demonstrates significant potential for health monitoring in harsh environments and for intelligent wound management, offering an integrated platform that unites real-time sensing with regenerative therapy.Scheme 1Conventional hydrogel-based biosensors readily freeze under low-temperature conditions, leading to reduced tissue adhesion, detachment from tissues, and signal failure, thereby compromising stable sensing. In this work, a polyethylene glycol (PEG)-based hydrogel bioadhesive sensor is endowed with antifreezing capability through a glycerol/water (Gly/H2O) binary solvent and reinforced with carboxylated CNTs to enhance mechanical property and electrical conductivity, yielding a sensor with strong adhesion, robust low-temperature tolerance, and electroactivity. The bioadhesive sensor maintains long-term, stable sensing under cold conditions and facilitates endogenous bioelectrical signal transmission, thereby enabling real-time monitoring of wound healing while promoting tissue regeneration. Conventional hydrogel-based biosensors readily freeze under low-temperature conditions, leading to reduced tissue adhesion, detachment from tissues, and signal failure, thereby compromising stable sensing. In this work, a polyethylene glycol (PEG)-based hydrogel bioadhesive sensor is endowed with antifreezing capability through a glycerol/water (Gly/H2O) binary solvent and reinforced with carboxylated CNTs to enhance mechanical property and electrical conductivity, yielding a sensor with strong adhesion, robust low-temperature tolerance, and electroactivity. The bioadhesive sensor maintains long-term, stable sensing under cold conditions and facilitates endogenous bioelectrical signal transmission, thereby enabling real-time monitoring of wound healing while promoting tissue regeneration. Materials and methods Materials PEG-SG and PEG-NH2 were purchased from Xiamen Nanopeg Co., Ltd. (China). The molecular weights of both polymers were 10 kDa. Glycerol was purchased from Sigma-Aldrich (purity ≥ 99.0%, USA). The raw multi-walled CNTs were obtained from Chengdu organic chemistry Co., Ltd. (China) with a diameter of 5-15 nm. Unless otherwise specified, all other reagents were of analytical grade and purchased from Beijing Chemical Reagent Co., Ltd. (China). Preparation of CNTs 3 g CNTs were immersed in anhydrous ethanol for 30 min, filtered under vacuum, and air-dried. The pre-treated CNTs were then dispersed in 240 mL of a mixed acid solution of concentrated H2SO4 and HNO3 (3:1, v/v). The mixture was sonicated in a thermostatic water bath at 50°C for 2 h, with uniform dispersion checked every 30 min. After completion, the reaction was cooled to room temperature and diluted with deionized water to three times its original volume. The suspension was allowed to stand for 24 h, and the supernatant was discarded. The remaining mixture was filtered, and the obtained solid was repeatedly washed with deionized water until the filtrate reached pH = 7.0. The product was vacuum-dried to obtain carboxyl-functionalized CNTs (CNT-COOH). Similarly, 3 g CNTs were washed 2-3 times with anhydrous ethanol, filtered, and air-dried. Then, 10 g of KOH and 5 g H2O2 were dissolved in 200 mL of deionized water to form the reaction solution. The pre-treated CNTs were added, and the mixture was sonicated at 60°C for 3 h. After cooling to room temperature, the mixture was filtered, and washed repeatedly with deionized water until the filtrate was neutral. The product was vacuum-dried to obtain hydroxyl-functionalized CNTs (CNT-OH). 3 g CNTs were washed 2-3 times with anhydrous ethanol, filtered, and air-dried. Subsequently, 8 g aqueous ammonia (NH3·H2O) and 5 g 3-aminopropyltriethoxysilane were added to 150 mL of anhydrous ethanol and stirred to form the amination solution. The pre-treated CNTs were then added, and the reaction mixture was refluxed at 70 °C for 8 h under continuous stirring. The mixture was filtered under vacuum and the filter cake was alternately washed with anhydrous ethanol and deionized water until the filtrate reached pH = 7.0. The final product was vacuum-dried to yield aminated CNTs (CNT-NH2). 3 g untreated CNTs were washed 2-3 times with anhydrous ethanol, filtered, and air-dried. Then, 50 mL of fuming H2SO4 was slowly added to 200 mL of concentrated H2SO4 under stirring to obtain the sulfonation solution. The pre-treated CNTs were added, and the mixture was stirred at 80 °C and 300 rpm for 6 h. After the reaction, the solution was poured into an ice bath to quench and cool to room temperature. The mixture was filtered, and washed with deionized water until the filtrate reached pH = 7.0. The product was vacuum-dried to obtain sulfonated CNTs (CNT-SO3H). Preparation of PEG/Gly/CNT bioadhesive sensor Glycerol (Gly) and deionized water (H2O) were mixed at different volume ratios of 0:10, 1:9, 2:8, 3:7, 4:6, and 5:5, and the resulting mixed solutions were labeled as solutions 0-5, respectively. Subsequently, 1 g PEG-SG and 1 g PEG-NH2 were each dissolved in 5 mL of the corresponding mixed solution, thoroughly mixed, and injected through a dual-syringe system to form PEG/Gly hydrogel biosensors, denoted as PEG/Gly0. PEG/Gly1 to PEG/Gly5 hydrogels were prepared using the same procedure with the corresponding mixed solutions. For clarity, PEG/Gly0 and PEG/Gly4 were referred to as PEG and PEG/Gly, respectively. To prepare CNT-doped hydrogels, a Gly:H2O mixed solvent (4:6, v/v) was used as the solvent system. CNTs with different surface functional groups (-COOH, -OH, -NH2, and -SO3H) were dispersed in the mixed solvent by ultrasonication for 20 min, followed by mixing through a dual-syringe system, named PEG/Gly/CNT-COOH, PEG/Gly/CNT-OH, PEG/Gly/CNT-NH2, and PEG/Gly/CNT-SO3H, respectively. Additionally, PEG/Gly/CNT-COOH hydrogels were prepared with different CNT-COOH contents (0.01 g, 0.03 g, and 0.05 g), designated as PEG/Gly/CNT-COOH1%, PEG/Gly/CNT-COOH3%, and PEG/Gly/CNT-COOH5%, which were abbreviated as PEG/Gly/CNT1%, PEG/Gly/CNT3%, and PEG/Gly/CNT5% for brevity. Characterizations Fourier transform infrared spectroscopy (FT-IR, Nicolet iS20, ThermoFisher Scientific, USA) equipped with an attenuated total reflectance (ATR) accessory was employed to identify and analyze the characteristic functional groups of Gly/H2O solutions with different volumes over a wavenumber range of 400-4000 cm−1. Transmission electron microscopy (TEM, FEI Tecnai G2 F30, ThermoFisher Scientific, USA) was used to observe the microstructure of CNTs. X-ray photoelectron spectroscopy (XPS, Escalab 250Xi, ThermoFisher Scientific, USA) was conducted to determine the surface elemental composition and chemical states of the CNTs, with binding energies calibrated using the C1s peak at 284.8 eV. The freezing behavior of H2O, Gly, and PEG/Gly hydrogels was analyzed using differential scanning calorimetry (DSC, Q20, TA Instruments, USA). The samples were heated from -80°C to 30°C at a rate of 2 °C/min under N2 atmosphere. The low-temperature stability and antifreezing performance of Gly/H2O mixtures and PEG/Gly-based hydrogels were evaluated using a programmable low-temperature chamber. Specifically, the hydrogels were placed in the chamber at -30°C for 12 h, then removed and captured using a high-resolution digital camera. Scanning electron microscope (SEM, S4800, Hitachi, Japan) was applied to observe the CNT distribution in the PEG matrix. After freeze-drying, the xerogel was fractured in liquid N2 and coated with platinum before SEM observation. The hydrogel precursor solutions were transferred into a cylindrical mold (1 cm in diameter, 1 cm in depth). After gelation, the initial volumes were recorded as V0, and then immersed in 10 mL PBS at 37°C with shaking at 60 rpm. At predetermined intervals, the hydrogels were retrieved, and their volumes (Vt) were recorded. The swelling ratio (%) was calculated as:Swellingratio(%)=VtV0×100%In the degradation assay, the formed hydrogels were lyophilized, weighed as M0, and then immersed in 10 mL PBS at 37°C with shaking at 60 rpm. The PBS was refreshed every day, followed by lyophilization and weight (Mt). The weight remains were calculated as:Weightremains(%)=MtM0×100% Density functional theory (DFT) calculation DFT calculations were performed to investigate representative hydrogen-bonding interactions among different CNT surface functional groups. Gaussian09 software was used to perform geometry optimization calculations and vibrational frequency analyses for each structure at the B3LYP/6-311G(d) level, which has been shown to accurately describe non-covalent interactions, particularly hydrogen bonds. By comparing the binding energy (ΔE) of several initial structures, the most stable geometry could be obtained and the binding energy was calculated as follows:ΔE = EA-B-EA-EB,where EA-B, EA, and EB represented the enthalpy of the A-B complex, A, and B, respectively. The non-covalent interaction (NCI) and reduced density gradient (RDG) were performed to analyze the hydrogen-bonding strength associated with different functional groups through Multiwfn software. Visual Molecular Dynamics (VMD) software was used to produce 3D structures and color-filled isosurfaces of interaction geometries. The color-filled RDG scatter map was plotted by Gnuplot 5.3 software. Mechanical and adhesion study Mechanical properties of the hydrogels were evaluated using a universal testing machine (Instron 5848, China) at room temperature to -30°C. The tensile rate was set to 30 mm/min, and the effective dimensions of dumbbell-shaped samples were 20 mm × 5 mm × 5 mm. Elastic modulus was derived from the slope of the linear region of the stress-strain curve. The toughness was quantified by the area under the stress-strain curve. Shear adhesion strength test was performed according to ASTM F2255. The porcine skins were cut into 5 cm × 2.5 cm, then 100 μL PEG-based bioadhesives were coated on one of the porcine skins covering 2.5 cm × 2 cm, and then the two pieces were overlapped and bonded together in the opposite direction at room temperature. After 1 h, the universal testing machine was used to test the force-displacement curve in the shear direction at a speed of 5 mm min−1. Shear adhesion strength was calculated as:Shearadhesionstrength=FmaxArea Burst strength test was performed according to ASTM F2392. A 5 mm defect was created in a 3 cm cardiac tissue segment using a biopsy punch, then the PEG based bioadhesives were applied to cover the defect for 30 min at room temperature. A pressure pump was used to apply pressure at 2 mL min−1 until failure of the bioadhesive. Electrochemical study Electrical properties of the PEG, PEG/Gly, and PEG/Gly/CNT hydrogels were characterized using an electrochemical workstation (CHI660E, CH Instruments, China). Samples (10 mm × 10 mm × 1 mm) were sandwiched between two copper electrodes of equal area. A sinusoidal voltage with an amplitude of 0.01 V was applied over a frequency range of 0.1-105 Hz. Real (Z’) and imaginary (Z”) components of the impedance were recorded, and resistance (R) was obtained from the fitted equivalent circuit. Electrical conductivity of the hydrogels was calculated according to the following equation:Conductivity=L(R×A),where L was the thickness of the hydrogel, and A was contact area between the hydrogel and the electrode. To evaluate conductivity under mechanical deformation, electrochemical impedance spectroscopy (EIS) measurements were performed on stretched hydrogel samples fixed to glass pads at defined elongation distances. Cyclic voltammetry (CV) measurements were performed within 0-1.0 V at a scan rate of 100 mV/s. The enclosed area of the CV curve was proportional to the capacitance. The PEG/Gly/CNT hydrogel was cut into pieces and then rejoined for 10 min to allow self-healing. The conductivity before and after self-healing was measured using a multimeter at room temperature. Sensitivity study The hydrogel samples were cut into small pieces (10 mm × 10 mm × 1 mm) and sandwiched between two copper electrodes of equal area. The real-time resistance of the hydrogel sensors under tensile or compressive deformation was recorded using an electrochemical workstation coupled with a universal testing machine and a digital source meter (Keithley 2450, Tektronix, USA). Normalized resistance change (ΔR/R0) and gauge factor (GF) were calculated according to the following equations:ΔR/R0(%)=R‐R0R0×100%,GF=ΔR/R0ε,where R0 was the initial resistance, R was the real-time resistance under strain, and ε represented the applied strain. The sensitivity of the hydrogel-based sensor was evaluated from the linear relationship between ΔR/R0 and ε. For wearable applications, the hydrogels were sandwiched between two 3M VHB adhesive tapes to fabricate flexible strain sensors. The assembled sensors were attached to the skin, and their real-time resistance changes under different deformation states were monitored at an applied voltage of 5 V. Biocompatibility study Hemocompatibility of the hydrogel was evaluated by hemolysis tests. Briefly, 0.1 g of hydrogel was immersed in 1 mL PBS within a 2 mL centrifuge tube and incubated in a water bath for 30 min at 37°C to obtain the extract solution. Subsequently, 900 μL of the hydrogel extract was mixed with 200 μL of fresh rabbit whole blood. As controls, 900 μL PBS mixed with 300 μL rabbit blood served as the negative control, while 900 μL of Triton X-100 mixed with 300 μL rabbit blood served as the positive control. All samples were incubated at 37°C for 1 h and centrifuged at 4000 rpm for 1 min. Then, 100 μL of supernatant from each sample was transferred to a 96-well plate, and absorbance at 540 nm was recorded using a microplate reader (Bio-Rad 680, USA). Hemolysis rate was calculated using the following equation:Hemolysisrate(%)=AP‐ANCAPC‐ANC×100%,where AP, ANC and APC represented the absorbance values of the hydrogel, negative control, and positive control, respectively. Cytocompatibility of the hydrogel was assessed using L929 mouse fibroblasts. L929 cells were seeded in 96-well plates at a density of 1 × 103 cells per well and incubated for 24 h. After removing the culture medium, 200 μL of hydrogel extract (at a ratio of surface area to medium: 6 cm2/mL) was added to each well, followed by incubation for 24 h. Subsequently, 10 μL CCK-8 solution was added to each well and incubated for 2 h. The absorbance at 450 nm was measured using a microplate reader to determine cell viability. In addition, cells were stained with calcein-AM/PI (Aladdin, China) and captured under a confocal laser scanning microscope (CLSM, TCS SP8, Leica, Germany). Cell viability was calculated as follows:Cellviability(%)=AS‐ABAC‐AB×100%,where AS, AB, and AC corresponded to absorbance of the sample, blank, and control groups, respectively. To evaluate embryonic-like repair-related gene expression (collagen type III, collagen type I, TGF-β1, TGF-β3, YAP and En1), the L929 fibroblasts were seeded into a 6-well culture plate at a density of 1 × 106 cells per well for 24 h before the PEG, PEG/Gly, and PEG/Gly/CNT matrix were placed over the cells. Because the PEG-based materials exhibited low cell-adhesive properties, an indirect coverage-based treatment protocol was adopted. After culturing for 3 days, RNA was extracted using an RNA rapid purification kit (Esunbio, China). The extracted RNA was reverse transcribed into cDNA using a reverse transcription kit (Takara, Japan). Quantitative real-time polymerase chain reaction (qPCR), with GAPDH as the internal reference, was employed to quantify gene expression. Gene expression levels were calculated using the 2−ΔΔCt method. Primers were designed and synthesized by Thermo Fisher Scientific (USA), and the details for the primers were summarized in Table S1. Wound repair and wound real-time tracking using SD rats All animal procedures were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by Tianjin Municipality and were approved by the Institutional Animal Care and Use Committee of YSY Gene Technology Co., Ltd. (Tianjin, China) (Approval No. YSY-DWLL-2024785). A total of 15 male SD rats (250-300 g) with diabetes were anesthetized with isoflurane in a sterile surgical environment. A full-thickness skin defect (10 mm in diameter) was created on the dorsal region of each rat. The animals were randomly divided into 5 groups (n = 3 per group): 1) Blank control group, without any dressing. 2) Tegaderm™ dressing group. 3) PEG bioadhesive group. 4) PEG/Gly bioadhesive group. 5) PEG/Gly/CNT bioadhesive group. For real-time wound monitoring, additional 3 rats were treated with the PEG/Gly/CNT bioadhesive sensor containing copper electrodes. The end of the electrode wires was reserved for connection to a data acquisition system. The sensor edges were fixed using medical adhesive tape to prevent detachment caused by scratching. All rats were housed individually and monitored daily for general health, food intake, and wound condition. The wound was captured and measured on postoperative days 0, 6, 9, and 12 to evaluate healing progress. On day 12, rats were sacrificed, and local wound tissues were harvested and fixed with 10% formalin for histological analysis, including hematoxylin-eosin (H&E), Masson's trichrome, and Sirius Red staining. Immunofluorescent staining was performed using antibodies against collagen type I (ab254113, Abcam), collagen type III (ab6310, Abcam), MMP-9 (ab76003, Abcam), and α-SMA (ab124964, Abcam). Major organs (heart, liver, spleen, lung, and kidney) were also subjected to H&E staining to assess systemic toxicity. Tissue sections were captured using a NanoZoomer digital slide scanner (Hamamatsu, Japan), and Sirius red-stained slices were observed under a polarized light microscope. Semi-quantitative analysis of fluorescence intensity was performed using Image Pro Plus software. On day 12, the electrical potential in the wound area was measured and analyzed using an electrometer, an oscilloscope, and platinum electrodes. The positive and negative terminals of the electrometer were connected to two platinum electrodes spaced 1 mm apart, and the potential difference across the wound was recorded to determine the average electric field strength in the region [31]. For the real-time monitoring group, impedance signals were recorded on days 0, 6, 8, 10, and 12. Before data collection, each rat was gently restrained without anesthesia to avoid stress-induced artifacts, and the reserved wire terminals of the sensor were connected to a Net Assist (v4.3.25) data acquisition system. Signal stability was verified before recording. Data were collected for 3 min using the same parameters as those applied in EIS measurements (amplitude: 0.01 V, frequency range: 0.1-105 Hz). After acquisition, the system was disconnected, and the rats were returned to their cages. All recordings were conducted in a quiet environment to minimize electromagnetic interference and ensure data accuracy. Wound repair and cardiac motion monitoring using Bama miniature pigs All animal procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by Tianjin Municipality and were approved by the Institutional Animal Care and Use Committee of YSY Gene Technology Co., Ltd. (Tianjin, China) (Approval No. YSY-DWLL-2024786). Six Bama miniature pigs (10-20 kg) were anesthetized with isoflurane in a sterile surgical environment. Three circular full-thickness skin wounds (20 mm in diameter and 5 mm in depth) were created on the dorsal region of each pig, with a lateral spacing of 5 cm and a longitudinal spacing of 3 cm between adjacent wounds. The wounds were randomly assigned to two groups: 1) Blank control group without any dressing. 2) PEG/Gly/CNT bioadhesive group. Each bioadhesive was fixed with medical adhesive tape to prevent displacement by scratching, and the animals were maintained under standard housing conditions. On postoperative days 7, 14, and 21, wound tissues were harvested, fixed with 10% neutral formalin, dehydrated, embedded, and sectioned for H&E, Masson's trichrome and Sirius Red staining to assess collagen deposition and tissue remodeling. In addition, one Bama miniature pig was used for a cardiac wound sealing experiment. After isoflurane anesthesia and strict aseptic preparation, the heart region was carefully exposed. A standardized cardiac injury was created using a 10 mm biopsy punch to induce bleeding, followed immediately by application of the PEG/Gly/CNT bioadhesive to the wound site for sealing and hemostasis. The pre-reserved copper wire terminals of the sensor were connected to a data acquisition system for real-time monitoring. Measurement parameters were set as amplitude 0.01 V, and frequency range of 0.1-105 Hz. Each acquisition lasted for 3 min to record cardiac motion-associated resistance signals. Seven days after the operation, the injured cardiac tissues were collected, fixed with 10% formalin, and subjected to H&E and Masson's trichrome staining to evaluate tissue integration and inflammatory response. Statistical analysis All data were expressed as mean ± standard deviation (SD). Statistical analysis was performed using one-way ANOVA test, and a P value ≤ 0.05 was considered significant. ns: not significant, ∗P ≤ 0.05, ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. Results and discussion Fabrication of the antifreezing PEG/Gly bioadhesive We first examined the phase state of Gly/H2O (G:W) mixtures with varying volume ratios at -30°C. After placing at -30°C for 12 h, solutions with G:W<3:7 became solid, while G:W4:6 exhibited partial freezing, and mixtures with G:W>4:6 remained liquid (Fig. S1a). DSC revealed a pronounced freezing point depression with increasing Gly content. The freezing point decreased to -21.96°C at G:W4:6, and further to -33.10°C at G:W5:5 (Fig. S1b), consistent with the ability of glycerol to disrupt the ordered crystallization of H2O and suppressing ice nucleation and growth via hydrogen-bonding interactions [6]. FT-IR provided additional evidence for this mechanism. In the 3200-3600 cm−1 region, the O-H stretching band increased in intensity and red-shifted with higher glycerol fractions, indicating stronger hydrogen bonding between Gly and H2O. Meanwhile, the H2O bending vibration at 1635 cm−1 progressively diminished and ultimately disappeared as Gly increased, indicating restricted water mobility and reduced structural ordering (Fig. S1c) [32]. To endow the PEG-based hydrogel with antifreezing capability, we employed PEG-succinimide glutarate (PEG-SG) and PEG-amine (PEG-NH2) as the bioadhesive hydrogel matrix [33,34], and used Gly/H2O mixtures of varying ratios as the solvent. First, we evaluated polymer solubility and gelling time. When G:W>5:5, the PEG components failed to dissolve and no gel formed (Table S2). We then assessed the antifreezing performance after 12 h at -30°C. PEG/Gly4 and PEG/Gly5 hydrogels showed no visible icing (Fig. S2a). DSC further corroborated these observations, showing that PEG/Gly4 exhibited a freezing point of -33.22°C, while PEG/Gly5 entered a glassy state under low-temperature conditions with no discernible crystallization peak (Fig. S2b), indicating complete suppression of water crystallization. Next, we compared the mechanical properties of the PEG/Gly hydrogels at room temperature and -30°C (Fig. S3). At room temperature, all hydrogels were stretchable, with elongation at break ranging from 82.49 ± 10.06% to 140.84 ± 7.44% and elastic modulus from 24.47 ± 0.50 kPa to 90.13 ± 18.22 kPa. Upon cooling to -30°C, some groups, such as PEG/Gly, PEG/Gly3 showed remarkable loss of extensibility, as elongation at break decreased by >30% compared with room temperature, alongside a sharp increase in modulus, compromising flexibility. By contrast, PEG/Gly4 displayed outstanding low-temperature stability, maintaining a tensile strength of 35.71 ± 1.38 kPa, elongation at break of 131.33 ± 16.86%, and modulus of 49.33 ± 5.13 kPa at -30°C, values that were essentially unchanged, thus preserving excellent stretchability. Although PEG/Gly5 also showed antifreezing stability, PEG/Gly4 provided the best overall balance of antifreezing behavior and mechanical robustness, and was therefore selected as the standard material system (abbreviated as PEG/Gly) for subsequent low-temperature sensing and biomedical evaluations. Preparation of the electroactive PEG/Gly/CNT bioadhesive sensor To develop a bioadhesive sensor that integrates antifreezing capability, robust tissue adhesion, and real-time monitoring, we designed a bioadhesive sensor based on an adhesive PEG matrix, Gly/H2O binary sovlent, and CNTs (Fig. 1a). To endow the PEG/Gly with electroactivity and sensing functionality, four types of functionalized CNTs bearing -OH, -COOH, -NH2, or -SO3H groups were incorporated into the network. TEM confirmed no significant differences in size among the four CNTs, while XPS verified successful functionalization with comparable degrees of surface modification among groups (Fig. S4, Fig. S5, Table S3). These results ruled out size effects and disparities in functionalization degree as sources of performance variation in the bioadhesive sensors.Fig. 1Fabrication and characterization for the antifreezing and electroactive bioadhesive sensor. (a) Schematic illustration of the fabrication process using PEG-SG, PEG-NH2, Gly/H2O binary solvent, and different functionalized CNTs to prepare the bioadhesive sensor. (b) Tensile stress-strain curves measured at -30°C (inset images show the macroscopic appearance of the bioadhesive sensor before and after stretching). (c) Toughness of the bioadhesive sensor with different CNT contents. (d) Calculated binding energies and hydrogen-bond lengths for representative interaction models involving different CNT surface functional groups. (e) Tensile stress-strain curves of PEG/Gly hydrogels containing different amounts of CNT-COOH at -30°C. (f) Toughness and shear adhesion strength of PEG/Gly hydrogels containing varied amounts of CNT-COOH. (g) Tissue adhesion with various tissues. (h) Live/dead staining of L929 cells and quantitative analysis of cell viability after 24 h of culture. Dotted line indicates the acceptable line for biomedical application. Data were presented as mean ± SD (n = 3). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. Fabrication and characterization for the antifreezing and electroactive bioadhesive sensor. (a) Schematic illustration of the fabrication process using PEG-SG, PEG-NH2, Gly/H2O binary solvent, and different functionalized CNTs to prepare the bioadhesive sensor. (b) Tensile stress-strain curves measured at -30°C (inset images show the macroscopic appearance of the bioadhesive sensor before and after stretching). (c) Toughness of the bioadhesive sensor with different CNT contents. (d) Calculated binding energies and hydrogen-bond lengths for representative interaction models involving different CNT surface functional groups. (e) Tensile stress-strain curves of PEG/Gly hydrogels containing different amounts of CNT-COOH at -30°C. (f) Toughness and shear adhesion strength of PEG/Gly hydrogels containing varied amounts of CNT-COOH. (g) Tissue adhesion with various tissues. (h) Live/dead staining of L929 cells and quantitative analysis of cell viability after 24 h of culture. Dotted line indicates the acceptable line for biomedical application. Data were presented as mean ± SD (n = 3). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. Tensile tests at -30°C revealed that incorporation of some CNTs (e.g., CNT-OH, CNT-SO3H) reduced extensibility relative to the PEG/Gly, whereas addition of CNT-COOH exhibited the opposite trend, achieving an elongation at break of 273.56 ± 4.68%, 1.94-fold to the PEG/Gly, demonstrating superior stretchability under cold conditions (Fig. 1b, Fig. S6). Consistent with the tensile results, toughness measurements also showed that CNT-COOH produced the most pronounced reinforcement (Fig. 1c). At -30°C, the PEG/Gly/CNT-COOH reached a toughness of 60.06 ± 3.74 kJ/m3, 2.01-fold to the PEG/Gly and markedly outperforming the other functionalized CNT groups. To uncover the molecular origin of this optimal reinforcement, we employed DFT calculations to elucidate the interaction mechanisms. To gain qualitative insight into the functional-group-dependent local interaction preference, we employed simplified DFT calculations focusing on representative hydrogen-bonding motifs. Although this model did not fully reproduce the complexity of the PEG/Gly/CNT network, it provided a useful basis for comparing the relative interaction tendencies of different CNT surface groups. Starting from four representative hydrogen-bonding motifs involving -COOH, -NH2, -SO3H, and -OH groups, we computed the interaction enthalpy (ΔH). The calculated ΔH values were -70.91, -51.04, -62.70, and -30.95 kJ/mol, respectively (Fig. 1d). In accordance with the minimum energy principle, the CNT-COOH-related motif exhibited the lowest ΔH and the shortest hydrogen bond length (2.66 Å). These results suggested that CNT-COOH possessed the strongest local interaction tendency among the functional groups, which might contribute to the experimentally observed enhancement in network stability and mechanical performance [35]. To further quantify noncovalent interactions, we performed RDG analysis and visualization. Likewise, the CNT-COOH system displayed prominent blue isosurfaces, evidencing strong attractive hydrogen bonding between carboxyl groups. Correspondingly, the sign (λ2) ρ scatter plots showed pronounced peaks in the highly negative region, corroborating the high interaction strength of the COOH···COOH hydrogen bonds (Fig. S7). Such robust hydrogen bonding might promote conformational extension and more ordered, directional assembly of CNT-COOH within the hydrogel matrix. Considering both structural stability and the experimentally observed mechanical reinforcement, CNT-COOH was identified as the most suitable conductive filler to provide a resilient supporting network for antifreezing bioadhesive sensors. To refine the mechanical reinforcement imparted by CNT-COOH, we evaluated the hydrogels containing 1-5 wt% CNT-COOH at -30°C (Fig. 1e, f). Tensile tests showed a pronounced toughness enhancement upon CNT-COOH addition. The PEG/Gly hydrogel exhibited a toughness of 27.65 ± 5.70 kJ/m3, which rose to 51.04 ± 3.02 kJ/m3 and 60.06 ± 2.75 kJ/m3 upon addition of 1 wt% and 3 wt% CNT-COOH. For 5 wt% CNT-COOH, the toughness dropped to 29.59 ± 7.06 kJ/m3, which was attributed to filler aggregation. Elongation at break followed a similar trend (Fig. S8). Adhesion testing identified 3 wt% CNT-COOH as optimal, exhibiting a shear adhesion strength of 41.50 ± 1.62 kPa, which was considered sufficient to support cardiac wound sealing and vascular closure [36,37], whereas the 5 wt% CNT-COOH group showed a concomitant decline in adhesion, consistent with the toughness decrease. This phenomenon could be attributed to the coupled effects of toughening, cohesion enhancement, and adhesion reinforcement mediated by CNT-COOH. At low loadings (1-3 wt%), CNT-COOH served as an efficient toughener, the carboxyl groups formed dynamic crosslinks with PEG chains via hydrogen bonding, while uniformly dispersed rigid fillers restricted chain slippage, thereby increasing cohesive strength and macroscopic toughness, which in turn indirectly strengthened interfacial adhesion [38]. Concurrently, specific hydrogen bonding between carboxyl groups and substrate functionalities, nanoscale mechanical interlocking, and improved stress transfer within the denser network further enhanced adhesion. At 5 wt%, however, the high surface energy drove filler aggregation, disrupting network homogeneity and impeding intimate contact at the tissue/material interface, which together degraded adhesion (Fig. S9). Collectively, 3 wt% was identified as the optimal CNT-COOH loading to simultaneously achieve high toughness and strong adhesion at -30°C. PEG/Gly/CNT3% was abbreviated as PEG/Gly/CNT and uesed for further study. We evaluated the mechanical properties of the PEG/Gly/CNT hydrogel at different temperatures ranging from room temperature to -30°C, the results showed a slight change in tensile strength and elongation at break, indicating its stretchability and ductility as a flexible wearable sensor in extreme low-temperature environments (Fig. S10). The PEG/Gly/CNT exhibited strong adhesion across diverse native tissues including porcine skin, gallbladder, heart, lung, and liver, demonstrating broad bioadhesive applicability across tissues with complex surface topographies (Fig. 1g), and exhibited a higher burst pressure than PEG and PEG/Gly (Fig. S11). Moreover, the PEG/Gly/CNT showed a self-healing property. After cutting and kneading, fragmented pieces rapidly coalesced into an intact construct after cutting and kneading, and the healed bioadhesive withstood external handling (e.g., lifting with tweezers) without failure, as well as displayed similar conductivity to the intact state (Fig. S12). This behavior was attributed to a dynamically reversible physical crosslinking network imparted by CNT-COOH, wherein intermolecular hydrogen bonding rapidly re-formed after mechanical disruption. Concurrently, interdiffusion and entanglement of flexible PEG chains across the freshly created interfaces enabled efficient reconnection of the cut surfaces. These features afforded unique advantages for conformal integration on complex curvatures and on-demand, cut-to-shape fabrication, providing a versatile design space for next-generation bioelectronic devices. In accordance with ISO10993, we conducted hemolysis testing and L929 fibroblast culture assays. All PEG-based hydrogels exhibited hemolysis rates <5% (Fig. S13), meeting biomedical safety requirements [39]. In CCK-8 assays and live/dead staining of L929 cells, the viabilities of L929 cells cultured with PEG, PEG/Gly, and PEG/Gly/CNT were 95.89 ± 1.54%, 95.84 ± 1.33%, and 84.64 ± 2.07%, respectively, within acceptable ranges for medical-device applications [40]. Fluorescent images revealed spindle-shaped fibroblastic morphology with good spreading across all hydrogels (Fig. 1h), indicating excellent cytocompatibility and supporting their potential for biomedical applications. A pivotal challenge for hydrogel deployment is long-term stability, as water loss markedly degrades mechanical and electrical performance, largely confining their use to humid environments [41,42]. Leveraging the synergistic water-retention effect of the Gly/H2O binary solvent, the PEG/Gly/CNT hydrogel exhibited excellent moisture-retention behavior. The swelling profiles showed that the swelling ratios of both PEG and PEG/Gly/CNT hydrogels increased during the first 360 min and reached equilibrium at 720 min, wheras the PEG/Gly/CNT exhibited a lower swelling ratio than the PEG (Fig. S14), which was attributed to the incorporation of CNTs, which provided additional hydrogen-bonding interactions and a rigid reinforcing framework [43,44]. Room-temperature drying tests were performed to compare solvent retention between PEG and PEG/Gly/CNT hydrogels. The PEG hydrogel nearly completely dehydrated by day 15, whereas the PEG/Gly/CNT hydrogel maintained ∼90% of its initial mass by day 25 (Fig. S15a), highlighting its superior moisture retention. Consistently, resistance measurements remained stable over 25 days for PEG/Gly/CNT, while PEG showed a continuous increase in resistance due to water loss (Fig. S15b). Collectively, the PEG/Gly/CNT hydrogel maintained stable performance across a broad humidity range, overcoming the moisture-dependence constraint and expanding the application scope of hydrogel-based biosensors. Electrochemical performance of the PEG/Gly/CNT bioadhesive sensor The core function of an antifreezing, conductive biosensor is to perceive mechanical inputs (e.g., wound strain, organ motion), and transduce them into electrical outputs, and thus electrical conductivity underpins efficient mechanical-to-electrical signal transduction, while conventional hydrogel sensors fail to maintain efficient electrical transduction because their conductive pathways become blocked (Fig. 2a). We systematically characterized PEG, PEG/Gly, and PEG/Gly/CNT bioadhesive sensors by EIS and CV, focusing on performance at room temperature and -30°C. At room temperature, the three hydrogels displayed distinct conductivities. All EIS Nyquist plots showed non-linear behavior, indicating that charge-transfer resistance dominated the system (typical Faradaic behavior) [45]. Notably, PEG/Gly/CNT showed a markedly lower impedance than PEG and PEG/Gly, suggesting that CNT-COOH substantially reduced charge transfer barriers. Quantitatively, the conductivities of PEG, PEG/Gly, and PEG/Gly/CNT were 1.26 ± 0.21 S/m, 0.97 ± 0.10 S/m, and 2.15 ± 0.08 S/m, respectively. CV further revealed charge-storage capabilities of 1.032 mF/cm2 (PEG), 0.776 mF/cm2 (PEG/Gly), and 1.379 mF/cm2 (PEG/Gly/CNT), confirming the highest charge-transfer efficiency for PEG/Gly/CNT at room temperature (Fig. S16(a and b)). When cooled to -30°C, the PEG and PEG/Gly exhibited sharply increased impedances (Fig. 2b), and the conductivity of PEG/Gly decreased to 1.07 ± 0.14 S/m (Fig. 2c), indicating severely hindered charge transfer and unstable transduction. In contrast, the PEG/Gly/CNT maintained a robust conductive network, with its conductivity remaining at 2.07 ± 0.20 S/m, and its capacitance remained at 1.047 mF/cm2 (Fig. 2d). These results demonstrated that the PEG/Gly/CNT maintained efficient charge transfer under low-temperature conditions.Fig. 2Electrochemical characterization for the antifreezing and electroactive bioadhesive sensor. (a) Schematic illustration of the bioadhesive sensor showing maintained ion and electron transport under low-temperature conditions. (b) EIS plots measured at -30°C (inset shows the equivalent circuit model). (c) Comparison of electrical conductivity at room temperature and -30°C. (d) CV curves measure at -30°C. (e) LED-lighting performance of PEG/Gly and PEG/Gly/CNT bioadhesive sensors under different conditions. (f) Comparison of conductivity as a function of temperature between the present bioadhesive sensor and previously reported systems. Data were presented as mean ± SD (n = 3). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. Electrochemical characterization for the antifreezing and electroactive bioadhesive sensor. (a) Schematic illustration of the bioadhesive sensor showing maintained ion and electron transport under low-temperature conditions. (b) EIS plots measured at -30°C (inset shows the equivalent circuit model). (c) Comparison of electrical conductivity at room temperature and -30°C. (d) CV curves measure at -30°C. (e) LED-lighting performance of PEG/Gly and PEG/Gly/CNT bioadhesive sensors under different conditions. (f) Comparison of conductivity as a function of temperature between the present bioadhesive sensor and previously reported systems. Data were presented as mean ± SD (n = 3). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. To visually compare conductivity at room temperature and -30°C, we integrated the PEG/Gly/CNT into a direct circuit (DC) as the conductive medium to light up a light-emitting diode (LED). The PEG/Gly hydrogel could complete the circuit at room temperature but produced only a weak glow. At -30°C, it barely sustained conduction, and stretching further increased impedance, dimming the LED. In contrast, the PEG/Gly/CNT remained brightly and stably lit at both room temperature and -30°C. During tensile deformation, its resistance changed minimally and the LED brightness remained close to the unstrained state (Fig. 2e), underscoring the combined low-temperature conductive stability and mechanical tolerance of PEG/Gly/CNT. Compared with previously reported systems (Fig. 2f), PEG/Gly/CNT showed superior low-temperature operability. At -30°C, it exhibited not only a higher conductivity than that of reported systems but also preserved the conductive continuity and structural integrity required for sustained biosensor function, overcoming the typical performance drop or failure of conventional hydrogels in the cold. This capability provided a strong materials-level basis for flexible biosensors in extreme-cold scenarios, such as polar exploration and cold-chain monitoring. Sensor performance of the PEG/Gly/CNT bioadhesive sensor To ensure measurement accuracy and minimize artifacts from fluctuating contact resistance, we optimized the electrode-bioadhesive interface with a multilayer stack consisting of 3M insulating film, a copper-tape electrode, and the bioadhesive sensor (Fig. S17), providing a stable platform for precise sensing. At -30°C, the PEG/Gly/CNT sensor exhibited excellent performance. The relative resistance change (ΔR/R0) increased linearly to ∼90% as strain rose from 0% to 150%, with a GF of 2.61, comparable to its room-temperature behavior (Fig. 3a). Notably, the device displayed a two-stage sensitivity profile, showing differentiated responses in the low-strain (0-50%) and high-strain (50-150%) regimes, which was advantageous for wide-range dynamic monitoring. The sensor also responded rapidly (response time: 258 ms; recovery time: 290 ms, Fig. 3b), enabling real-time capture of transient mechanical signals. Under cyclic loading at 1.0 Hz, the ΔR/R0 waveforms were regular and highly reproducible (Fig. 3c), confirming stable and precise readout during repeated deformation.Fig. 3Sensor characterization for the antifreezing and electroactive bioadhesive sensor. (a) Relationship between relative resistance change and strain of the bioadhesive sensor. (b) Response and recovery times of the bioadhesive sensor at -30°C. (c) Resistance change under different cyclic strains at a frequency of 1.0 Hz. (d) Real-time monitoring of finger bending at different angles using the bioadhesive sensor at -30°C. (e) Relative resistance change during repeated loading and unloading of 130% strain over 300 cycles. (f) Relative resistance changes of the bioadhesive sensor during wrist bending at room temperature and -30°C. (g) Relative resistance changes of the bioadhesive sensor during elbow bending in the initial state and after storage under ambient conditions for one week. (h) Comparison of gauge factor as a function of temperature between the present bioadhesive sensor and previously reported systems. Sensor characterization for the antifreezing and electroactive bioadhesive sensor. (a) Relationship between relative resistance change and strain of the bioadhesive sensor. (b) Response and recovery times of the bioadhesive sensor at -30°C. (c) Resistance change under different cyclic strains at a frequency of 1.0 Hz. (d) Real-time monitoring of finger bending at different angles using the bioadhesive sensor at -30°C. (e) Relative resistance change during repeated loading and unloading of 130% strain over 300 cycles. (f) Relative resistance changes of the bioadhesive sensor during wrist bending at room temperature and -30°C. (g) Relative resistance changes of the bioadhesive sensor during elbow bending in the initial state and after storage under ambient conditions for one week. (h) Comparison of gauge factor as a function of temperature between the present bioadhesive sensor and previously reported systems. To validate the biosensing potential, the PEG/Gly/CNT sensor was mounted on key body joints involved in motion. During finger bending tests, the output ΔR/R0 increased linearly with bending angle and remained stable throughout the measurement at both room temperature and -30°C (Fig. S18, Fig. 3d), demonstrating excellent instantaneous responsiveness and enabling precise quantification of hand motions. The sensor also captured both high-frequency, small-amplitude knee swings during casual jogging and large-amplitude flexion-extension during high-knee running with high real-time fidelity (Fig. S19). Under 130% strain, the device maintained outstanding signal repeatability after 300 loading-unloading cycles (Fig. 3e), consistent with the long-term resistance stability (Fig. S15b), highlighting long-term durability and structural robustness. The sensor discriminated wrist extension and flexion, and its response profiles at room temperature and -30°C showed similar patterns without discernible signal attenuation (Fig. 3f), indicating that high strain sensitivity was preserved even in extreme cold. When deployed for real-time elbow motion monitoring in ambient conditions, the stored device retained essentially the same signal magnitude after one week (Fig. 3g), indicating excellent anti-dehydration behavior and electrical stability. Benchmarking against previous reports further demonstrated a notable improvement in GF (Fig. 3h), while the overall reliability, longevity, and damage tolerance met the demands of stretchable electronics. Coupled with its stable low-temperature performance, this platform opened new avenues for wearable health monitoring, intelligent human-machine interaction, and biomedical diagnostics, demonstrating substantial technological promise. Embryonic regeneration and real-time monitoring of the PEG/Gly/CNT bioadhesive sensor In a diabetic full-thickness wound model, different materials were applied to the wound to evaluate their wound-healing efficacy (Fig. 4a). Based on wound-potential measurements performed on day 12 after treatment, the PEG/Gly/CNT group showed the greatest increase in wound potential among all groups (Fig. 4b), with a wound-potential increment of 182.00 ± 3.27 mV, which was significantly higher than the control, Tegaderm, and PEG/Gly groups (Fig. S20), indicating its effectiveness in facilitating endogenous bioelectrical signal transmission. The PEG/Gly/CNT exhibited a faster wound recovery rate by day 12, the wound recovery reached 99.63 ± 0.41%, significantly exceeding the Tegaderm and PEG/Gly groups (Fig. S21, Fig. 4c), and was also higher than healing rates reported in several recent studies (Fig. 4d). For real-time monitoring, the PEG/Gly/CNT sensor applied to the wound was connected to a data acquisition system to record resistance changes during healing. In the immediate postoperative phase, the resistance increased, which correlated with the early inflammatory exudative stage typical of diabetic wounds. Protein- and cytokine-rich exudates could hinder ion migration within the hydrogel and dilute the conductive medium, thereby elevating the charge-transfer impedance [46]. As healing progressed with inflammation subsiding, exudate diminishing, and tighter apposition between newly formed tissue and the hydrogel, the conductive pathways progressively recovered, leading to a decrease in resistance. The three rats displayed highly reproducible resistance-time curves. Regression analysis further revealed a significant correlation between resistance and wound-recovery rate, indicating that the impedance signal captured the dynamic progression of wound repair (Figs. S22 and S23).Fig. 4Embryonic wound regeneration for the bioadhesive sensor. (a) Schematic illustration of diabetic rat full-thickness skin wound modeling and wound-repair evaluation. (b) Wound potential distribution of different treatments on day 12. (c) Wound-recovery rates of different treatment groups after application of the bioadhesive sensor. (d) Comparison of wound-recovery rates achieved by the bioadhesive sensor with those reported in previous studies (red area indicates the diabetic wound model, and gray area indicates the normal wound model). (e) Histological staining using H&E, Masson's trichrome, and Sirius Red staining to assess collagen remodeling on day 12. (f) Immunofluorescence staining of collagen type III and collagen type I, and of MMP-9 and α-SMA, together with quantitative analysis of fluorescence intensity. (g) Schematic illustration of porcine wound repair promoted by facilitated endogenous bioelectrical signal transmission. (h) Sirius Red staining at the porcine wound defects. Data were presented as mean ± SD (n = 3). ∗P ≤ 0.05, ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. Embryonic wound regeneration for the bioadhesive sensor. (a) Schematic illustration of diabetic rat full-thickness skin wound modeling and wound-repair evaluation. (b) Wound potential distribution of different treatments on day 12. (c) Wound-recovery rates of different treatment groups after application of the bioadhesive sensor. (d) Comparison of wound-recovery rates achieved by the bioadhesive sensor with those reported in previous studies (red area indicates the diabetic wound model, and gray area indicates the normal wound model). (e) Histological staining using H&E, Masson's trichrome, and Sirius Red staining to assess collagen remodeling on day 12. (f) Immunofluorescence staining of collagen type III and collagen type I, and of MMP-9 and α-SMA, together with quantitative analysis of fluorescence intensity. (g) Schematic illustration of porcine wound repair promoted by facilitated endogenous bioelectrical signal transmission. (h) Sirius Red staining at the porcine wound defects. Data were presented as mean ± SD (n = 3). ∗P ≤ 0.05, ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001. H&E staining showed a markedly narrower granulation tissue width in the PEG/Gly/CNT group (∼1.28 mm) than in the control (4.30 mm), Tegaderm (4.83 mm), and PEG/Gly (2.04 mm). Masson's trichrome further revealed richer and more orderly collagen deposition (Fig. S24). Consistently, Sirius Red staining displayed pronounced collagen type III (yellow signal) in PEG/Gly/CNT (Fig. 4e), together with qPCR assays showing elevated collagen type III, TGF-β3, YAP gene expression levels (Fig. S25), suggesting an embryonic-like repair healing phenotype [47]. Immunofluorescence confirmed this trend. In PEG/Gly/CNT, collagen type III expression was more pronounced than collagen type I, yielding a col-III/col-I ratio of 5.43 ± 1.17, significantly higher than in other groups, where collagen type I predominated. Likewise, MMP-9 expression exceeded α-SMA, with a MMP-9/α-SMA ratio of 5.78 ± 0.77 in PEG/Gly/CNT (Fig. 4f, Fig. S26, Fig. S27). Given that MMP-9 is a associated with matrix remodeling in embryonic healing, whereas α-SMA is commonly associated with fibrotic/scar-type repair [[48], [49], [50]], these results indicated that PEG/Gly/CNT promoted tissue regeneration by facilitating endogenous bioelectrical signal transmission and supporting an embryonic-like healing program. Importantly, no abnormal histopathology was observed in major organs (Fig. S28), supporting the in vivo biosafety of the PEG/Gly/CNT bioadhesive sensor. Consistently, in a Bama miniature pig wound model, PEG/Gly/CNT accelerated closure compared with control (Fig. S29). Histology showed keratinized tissue formation by postoperative day 14 and a mature stratum corneum on day 21 at the wound site (Fig. S30). We proposed the following mechanism: The electroactive bioadhesive sensor facilitated endogenous bioelectrical signal transmission at the injured wound site, thereby potentially guiding keratinocytes and fibroblasts to regulate collagen deposition and remodeling. This microenvironmental reprogramming likely activated the Wnt/β-catenin pathway [51,52], promoting scar-reduced, embryonic-like tissue regeneration (Fig. 4g), as suggested by Sirius Red staining, which showed that a more ordered deposition of collagen type III in the PEG/Gly/CNT than in the control (Fig. 4h). It should be noted that the significance of subzero tolerance was not to enable wound healing at -30°C, but to ensure that the hydrogel sensor remained mechanically compliant, adhesive, and conductive during storage, transport, and immediate deployment in extreme-cold environments. Cardiac monitor and performance evaluation of the PEG/Gly/CNT bioadhesive sensor To demonstrate the feasibility and translational potential of the bioadhesive sensor for wound monitoring, we employed a Bama miniature pig cardiac wound model and evaluated the experimental workflow, sensing performance, and biocompatibility (Fig. 5a). As illustrated schematically (Fig. 5b), cardiac contraction generates 5-12% cyclic strain [53], which the sensor transduced into resistance changes, enabling indirect monitoring of cardiac motion. To emulate clinical needs for dynamic, postoperative cardiac surveillance, we established a stepwise protocol comprising wound creation, sensor placement, adhesion verification, and cardiac-motion signal monitoring. A standardized perforation was first introduced on the heart surface, after which the bioadhesive sensor was conformally applied to the wound. Secure adhesion was visually validated by forceps lifting, demonstrating stable attachment even under continuous myocardial motion during the observation period. Finally, the bioadhesive device enabled real-time monitoring of beating-associated mechanical signals from the heart (Fig. 5c), thereby supporting dynamic assessment of postoperative cardiac wounds. This integrated procedure underscores the sensor's conformal adaptability to complex, highly mobile tissue interfaces and validates its reliable adhesion in a dynamic wound milieu, both of which were prerequisites for stable monitoring on dynamic tissue interfaces.Fig. 5Wound monitor function of the bioadhesive sensor. (a) Schematic illustration of sensor application and cardiac-motion signal monitoring at the wound site of a Bama miniature pig heart. (b) Schematic illustration of beating-associated mechanical signal monitoring, showing cardiac pulsation and sensor readout. (c) Application of the bioadhesive sensor to seal the cardiac wound and monitor cardiac motion through an external readout circuit. (d) Resistance change-strain relationship curve of the bioadhesive sensor. (e) Real-time cardiac motion signals recorded by the bioadhesive sensor. (f) Comparison of peak amplitudes of the beating-associated resistance signals on day 0 and day 7 after implantation. (g) Histological staining of the miniature porcine heart interface after application of the PEG/Gly/CNT bioadhesive sensor for 7 days. The asterisk indicates the implanted material. Wound monitor function of the bioadhesive sensor. (a) Schematic illustration of sensor application and cardiac-motion signal monitoring at the wound site of a Bama miniature pig heart. (b) Schematic illustration of beating-associated mechanical signal monitoring, showing cardiac pulsation and sensor readout. (c) Application of the bioadhesive sensor to seal the cardiac wound and monitor cardiac motion through an external readout circuit. (d) Resistance change-strain relationship curve of the bioadhesive sensor. (e) Real-time cardiac motion signals recorded by the bioadhesive sensor. (f) Comparison of peak amplitudes of the beating-associated resistance signals on day 0 and day 7 after implantation. (g) Histological staining of the miniature porcine heart interface after application of the PEG/Gly/CNT bioadhesive sensor for 7 days. The asterisk indicates the implanted material. To quantify the fidelity of this strain-to-resistance conversion, we established the ΔR/R0-strain calibration curve (Fig. 5d). The ΔR/R0 exhibited a highly linear dependence on strain, with a fitted relation of y = 0.5344x-0.021 (R2 = 0.9867), demonstrating excellent linear response and precise strain quantification. Building on this calibration, real-time recordings of cardiac activity displayed regular oscillatory waveforms that were fully synchronized with the heartbeat cycle (Fig. 5e), confirming accurate capture of dynamic cardiac mechanical activity through electrical resistance readout. The average peak amplitudes of the beating-associated resistance signals recorded by the bioadhesive sensor showed no significant difference between day 0 and day 7 after implantation (Fig. 5f). The combination of real-time responsiveness and quantitative precision supports the use of this platform as an integrated tool for postoperative monitoring of cardiac mechanical motion at wound sites, offering potential support for the early detection of complications such as abnormal regional wall motion. Biocompatibility is pivotal for clinical translation. Histological assessment of the sensor-myocardium interface revealed tight conformal contact between the implanted PEG/Gly/CNT bioadhesive sensor and cardiac tissue, without evident inflammatory cell infiltration or tissue necrosis (Fig. 5g), indicating excellent in vivo compatibility. This supported the short-term in vivo safety of the sensor at cardiac wound sites and and highlighted its potential for future translation to cardiovascular wound monitoring. In the Bama miniature pig cardiac wound model, the bioadhesive sensor demonstrated precise strain-resistance transduction, real-time cardiac motion monitoring, and superior biocompatibility. Beyond offering an innovative tool for postoperative, dynamic assessment of cardiac wounds, the platform extended the application of conductive hydrogels to cardiovascular monitoring. Importantly, the present system should not be simply regarded as a combination of antifreezing, adhesive, and conductive elements. Instead, the key advance lay in establishing an integrated material platform in which each component contributed to a unified monitoring-to-repair function. These results provided theoretical and experimental support for next-generation devices that integrated mechanical sensing, mechano-electrical readout of cardiac pulsation, and favorable in vivo biocompatibility. Despite the promising results, several limitations should be noted. First, although the PEG/Gly/CNT hydrogel showed high solvent retention and stable electrical performance, glycerol retention under continuous exposure to wound exudate was not directly quantified, and its relationship with the preservation of subzero flexibility after prolonged immersion remained to be clarified. Second, while the minimal in vitro resistance drift suggested that the in vivo impedance evolution mainly reflected wound-state changes rather than sensor dehydration alone, these two factors were not fully decoupled and warrant further investigation under controlled exudate-mimicking conditions. Third, the porcine cardiac study relied on a transthoracic wired connection for signal acquisition, which represented a proof-of-concept setup rather than a clinically applicable long-term implant configuration. Future efforts should therefore focus on wireless, fully implantable, and mechanically robust device architectures to further advance the translational potential of this platform. Conclusions We report a PEG/Gly/CNT bioadhesive sensor with antifreezing capability, strong adhesion, and electroactivity. Using PEG as the adhesive hydrogel matrix and a Gly/H2O binary solvent to confer subzero tolerance (-30°C), the incorporation of 3 wt% CNT-COOH formed a stable conductive network at -30°C (GF = 2.61, conductivity = 2.07 ± 0.20 S/m). The bioadhesive sensor exhibited outstanding mechanical properties and adhesion, wide-range linear strain response with rapid, stable readout at low temperature, and practical application in LED driving and human joint motion monitoring. In a diabetic rat wound model, the bioadhesive sensor device enabled real-time electrical monitoring and remarkably accelerates healing, with increased collagen type III and elevated MMP-9/α-SMA ratios, indicating an embryonic-like healing phenotype. In Bama miniature pigs, the bioadhesive sensor achieved conformal adhesion on heart wounds and real-time recording of cardiac motion-associated resistance signals. Hence, the antifreezing performance broadened the environmental applicability of the bioadhesive sensor, while supporting its wound-monitoring and repair-promoting functions under physiological conditions. This platform offered a promising solution for intelligent wound management under extreme cold and cardiovascular monitoring. CRediT authorship contribution statement Linhao Hou: Writing – original draft, Investigation, Conceptualization. Wei Jing: Investigation, Data curation. Xueying Zhang: Methodology, Investigation. Pengfei Wei: Validation. Fanqi Hu: Validation. Dingfei Qian: Validation. Wenchao Wang: Validation. Bo Zhao: Conceptualization. Yiqian Huang: Writing – review & editing, Project administration, Conceptualization. Xuesong Zhang: Writing – review & editing, Supervision. Xinyu Liu: Visualization, Methodology. Yuguo Chen: Writing – review & editing, Supervision. Ethics approval and consent to participate All animal procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by Tianjin Municipality and were approved by the Institutional Animal Care and Use Committee of YSY Gene Technology Co., Ltd. (Tianjin, China) (Approval No. YSY-DWLL-2024785, YSY-DWLL-2024786). Declaration of competing interest The authors declare the following personal relationships which may be considered as potential competing interests: Wei Jing, Xueying Zhang, Pengfei Wei, Bo Zhao and Yiqian Huang are currently employed by Beijing Biosis Healing Biological Technology Co. Ltd.