Wearable single-electrode capacitive sensor with large penetration depth for intelligent inbody- and hemorrhage monitoring

Abstract

A capacitive sensor, including a carbon nanotube paper composite (CPC) electrode; and a multi-walled carbon nanotube (MWCNT) foam disposed on the CPC electrode. Further, a method of detecting a hemorrhage with the capacitive sensor including denoising one or more capacitive signals from the capacitive sensor, deriving an offset value, an isovolumetric contraction time (IVCT), a root mean square (RMS) value of mechanical vibrations from a heart (SCG), a heart rate (HR), or a combination thereof from the one or more capacitive signals, and detecting a hemorrhage with a machine-learning algorithm based on the offset value, the IVCT, the RMS value of SCG, the HR, or a combination thereof.

Description

WEARABLE SINGLE-ELECTRODE CAPACITIVE SENSOR WITH LARGE PENETRATION DEPTH FOR INTELLIGENT INBODY- AND HEMORRHAGE MONITORINGCROSS REFERENCE TO RELATED APPLICATION

[0001] This application claims priority to U.S. Provisional Application 63 / 716,997 filed November 6, 2024, the entire disclosure of which is hereby incorporated by reference.STATEMENT OF GOVERNMENT LICENSE RIGHTS

[0002] This invention was made with government support under Grant No. W81XWH-21-DMRDP-BRISCC- DM210169, awarded by the Department of Defense (DOD). The government has certain rights in the invention.BACKGROUND

[0003] In-body monitoring is essential for sustaining and improving health outcomes, providing critical insights into internal physiological functions. Detecting internal hemorrhage and monitoring organ functions are vital for early diagnosis and timely intervention in medical emergencies. Accurate assessment of in-body biometrics, particularly for internal hemorrhage, can significantly enhance patient care by providing real-time data that aid in the timely diagnosis and management of critical health issues.

[0004] Available monitoring tools for assessing in-body biometrics range from traditional imaging techniques like computed tomography (CT) scanning to more advanced, safe, and real-time methods. Near-infrared spectroscopy (NIRS) uses light to measure tissue oxygenation and blood volume for hemorrhage assessment. However, its effectiveness is limited by the small penetration depth (XL) of light (~10 mm), restricting the ability to monitor deep tissues. Moreover, optical sensors typically require a large electric power, and their accuracy is affected by ambient light and skin characteristics. Ultrasonography uses ultrasound waves that travel through tissues and reflect upon encountering tissues with different acoustic impedances, providing in-body anatomicalinformation. With low wave absorption, it offers a larger L (-100 mm). Nevertheless, the need for seamless sensor-skin contacts poses challenges for wearable applications.

[0005] Electric methods show promise in wearable applications and offer greater XL. Impedance tomography applies a small alternating current to the tissue and measures the resulting voltage to locate hemorrhage at greater depths. However, impedance is measured between two electrodes, suffering from limited spatial resolution. Unless the spacing between two electrodes is small, the complex conductive pathways within tissues and fluids cannot be understood. Capacitance tomography generates an electric field inside the body to detect changes in dielectric constant from tissue and blood composition, aiding in hemorrhage assessment. Capacitive sensing typically features a coplanar electrode array for improved spatial resolution and lower power consumption. However, XL is also limited by the spacing between electrodes. Despite their limitations, these tools have been crucial for diagnosing hemorrhage, monitoring treatment progress, and conducting regular health check-ups.

[0006] In recent years, there has been a growing trend toward developing and adopting wearable sensors for estimating in-body biometrics. Devices equipped with photoplethysmography (PPG) and biopotential sensors, combined with intelligent algorithms, can continuously monitor vital signs such as heart rate, respiratory rate, and other critical health metrics. Leveraging intelligent algorithms renders them essential for chronic disease management and predicting early signs of illness. As a result, wearable sensors have become increasingly popular among healthcare providers and consumers due to their convenience and accessibility. However, a significant gap remains in addressing acute health issues like internal hemorrhage or lung dysfunction, where more advanced sensors are needed for profiling in-body conditions in real time. Moreover, for emergency and home use, field-deployable sensors should offer a simple, streamlined platform to be effective in critical situations.

[0007] Monitoring in-body biometrics is essential for managing critical health conditions such as internal hemorrhage. Although optical and impedance tomographytechniques offer real-time monitoring with minimal medical infrastructure, they still face challenges in accurately assessing deeper tissues in wearable formats.

[0008] Accordingly, sensors for in-body biometric measurements are needed.SUMMARY

[0009] This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This summary is not intended to identify key features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0010] In one aspect, disclosed herein is a capacitive sensor, including a carbon nanotube paper composite (CPC) electrode, and a multi-walled carbon nanotube (MWCNT) foam disposed on the CPC electrode.

[0011] In some embodiments, the CPC electrode includes a cellulose paper, and a plurality of multi-walled carbon nanotubes (MWCNTs). In some embodiments, the plurality of MWCNTs forms a plurality of cantilever shaped fibers along an edge of the CPC electrode. In some embodiments, the CPC electrode further comprises silver ink. In some embodiments, the CPC electrode is circular.

[0012] In some embodiments, the MWCNT foam is cylindrical. In some embodiments, the MWCNT foam comprises polyurethane (PU). In some embodiments, the capacitive sensor is configured to sense a change in capacitance based on pressure exerted on the MWCNT foam.

[0013] In some embodiments, the capacitive sensor is configured to detect a heart biometric, a lung biometric, an arterial biometric, or a combination thereof. In some embodiments, the capacitive sensor is configured to be positioned on a sternum of a patient. In some embodiments, the capacitive sensor is configured to measure an electrocardiogram (EKG), a femoral artery pressure, a change in capacitance, or a combination thereof.

[0014] In another aspect, disclosed herein is a sensor system comprising two capacitive sensors as described herein.

[0015] In some embodiments, a first capacitive sensor of the two capacitive sensors and a second capacitive sensor of the two capacitive sensors are configured to be positioned on a forearm of a patient. In some embodiments, the sensor system is configured to detect a global change in blood volume of the patient.

[0016] In some embodiments, a first capacitive sensor of the two capacitive sensors is configured to be positioned on a first arm of a patient and a second capacitive sensor of the two capacitive sensors is configured to be positioned on a second arm of the patient. In some embodiments, the sensor system is configured to detect a regional change in blood volume of the patient.

[0017] In yet another aspect, disclosed herein is a sensor system comprising four capacitive sensors as described herein.

[0018] In some embodiments, a first capacitive sensor of the four capacitive sensors and a second capacitive sensor of the four capacitive sensors are configured to be positioned on an intercostal mid-clavicular area of a patient.

[0019] In some embodiments, a third capacitive sensor and a fourth capacitive sensor of the four capacitive sensors are configured to be positioned on an upper pectoral area of the patient.

[0020] In some embodiments, a third capacitive sensor and a fourth capacitive sensor of the four capacitive sensors are configured to be positioned on an abdomen of the patient.

[0021] In yet another aspect, disclosed herein is a method of detecting a hemorrhage with a capacitive sensor disclosed herein, including denoising one or more capacitive signals from the capacitive sensor, deriving an offset value, an isovolumetric contraction time (IVCT), a root mean square (RMS) value of mechanical vibrations from a heart (SCG), a heart rate (HR), or a combination thereof from the one or more capacitive signals, and detecting a hemorrhage with a machine-learning algorithm based on the offset value, the IVCT, the RMS value of SCG, the HR, or a combination thereof.

[0022] In yet another aspect, disclosed herein is a method of manufacturing a capacitive sensor as disclosed herein, including depositing a first solution of MWCNTs onto the cellulose paper, screen printing silver ink onto the cellulose paper, and punching the cellulose paper with a circular punch to form the CPC electrode.

[0023] In some embodiments, the method further includes cutting a foam sheet to form a cylindrical foam and coating the cylindrical foam with a second solution of MWCNTs to form the MWCNT foam.

[0024] In another aspect, disclosed herein is a method of detecting a central blood pressure (BP) or a peripheral blood pressure (BP) with the capacitive sensor disclosed herein, the method including positioning the capacitive sensor on skin near a target location, deriving a mean capacitance value, an isovolumetric contraction time (IV CT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from one or more capacitive signals of the capacitive sensor, and detecting the central BP or the peripheral BP with a machine-learning algorithm based on the mean capacitance value, the IVCT, the wave amplitude of SCG, the HR, the HRV, initial BP value, or a combination thereof.

[0025] In some embodiments, the capacitive sensor is one or more capacitive sensors.

[0026] In yet another aspect, disclosed herein is a method of cardiac imaging with an array of capacitive sensors comprising one or more capacitive sensors described herein, the method including positioning the array of capacitive sensors on skin near a heart, deriving a mean capacitance value, an isovolumetric contraction time (IVCT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from one or more capacitive signals from the array of capacitive sensors, and detecting cardiac function parameters with a machinelearning algorithm, wherein the cardiac function parameters comprise stroke volume, cardiac output, and ejection fraction.DESCRIPTION OF THE DRAWINGS

[0027] The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

[0028] FIGURE 1A is a wearable single-electrode capacitive sensor, in accordance with the present technology.

[0029] FIGURES 1B-1C is a schematic illustration of a capacitive sensor detecting variations in dielectric constant and pressure-induced displacement for in-body- and hemorrhage monitoring.

[0030] FIGURES 2A-2C show a fabrication method of a capacitive sensor consisting of a CPC electrode and a MWCNT foam, in accordance with the present technology.

[0031] FIGURE 3A is a SEM image of the individualized conductive fibers in a CPC electrode and real image of a capacitive sensor, in accordance with the present technology.

[0032] FIGURE 3B is a graph showing initial capacitance of the CPC electrode and capacitive sensor (N=4), in accordance with the present technology.

[0033] FIGURE. 4A is a sensing mechanism and the equivalent circuit model of a single-electrode capacitive sensor detecting in-body dielectric constant (sin-body) within the penetration depth (XL), in accordance with the present technology.

[0034] FIGURE. 4B shows a surrogate human tissue model and SNR analysis for XL characterization, and a graph showing AC of a metal object approaching CPC electrodes with various diameters in the surrogate human tissue model, in accordance with the present technology.

[0035] FIGURE. 5A shows a simulation of an electric field generated by CPC electrodes with diameters of 11 mm and 1.5 mm in a human tissue environment, in accordance with the present technology.

[0036] FIGURE. 5B shows the L of a copper electrode, a CPC electrode, and a sensor combining a CPC electrode and a MWCNT foam, in accordance with the present technology.

[0037] FIGURE. 5C is a graph showing AC of variations in dielectric constant using a copper electrode, a CPC electrode, and a CPC electrode integrated with a MWCNT foam, in accordance with the present technology.

[0038] FIGURE. 5D shows a sensor combining a CPC electrode and MWCNT foam exhibited an improved sensitivity to the pressure (S: 1.4 pF kPa-1) within the practical preload pressure range when worn on body (0.6-1.3 kPa), in accordance with the present technology.

[0039] FIGURES 5E-5F are graphs showing a comparison between the AC and real-time actual pressure under a 2 Hz-sinusoidal wave, in accordance with the present technology.

[0040] FIGURE 6A shows capacitive heart and lung models, in accordance with the present technology.

[0041] FIGURE 6B shows AC as the volume of an infusion bag increases (saline solution injection) in the heart model, in accordance with the present technology.

[0042] FIGURES 6C-6D show AC as the volume of an infusion bag increases (air injection) in the lung model, in accordance with the present technology.

[0043] FIGURES 7A-7B is a AC and spectrogram of human heart biometrics in different frequency ranges using four capacitive sensors (SI & S2 positioned above the heart; S3 & S4 located 20 cm away from the heart), in accordance with the present technology.

[0044] FIGURE 7B is a AC of human lung biometrics throughout a breathing cycle — inhale, hold, exhale, hold — using four capacitive sensors (SI and S2 placed over the chest; S3 and S4 positioned on the abdomen), in accordance with the present technology.

[0045] FIGURES 8A-8B show a model setup and schematic illustration of the relationship between BP and blood volume in arteries and veins, in accordance with the present technology.

[0046] FIGURE 8C shows a AC as the geometry of the tube increases in the artery model, in accordance with the present technology.

[0047] FIGURE 8D is a comparison of reference MBP and low-frequency AC during Valsalva and handgrip maneuvers, with SI detecting both dielectric constant and pressure changes, while S2 detects dielectric constant changes, in accordance with the present technology.

[0048] FIGURES 9A-9C shows a correlation between reference MBP and AC from two capacitive sensors during Valsalva and handgrip maneuvers, (r: Pearson correlation coefficient; p: Spearman correlation coefficient), in accordance with the present technology..

[0049] FIGURES 10A-10C show comparisons of reference MBP and low- frequency AC from two capacitive sensors placed on different arms during Valsalva, rightarm handgrip, and left-arm handgrip maneuvers, and the correlation between reference MBP and AC from two capacitive sensors during Valsalva, right-arm handgrip, and leftarm handgrip maneuvers, in accordance with the present technology.

[0050] FIGURE 11A is an experimental setup involving a capacitive sensor positioned on the skin at the mid-sternum, along with a femoral artery catheter and EKG as reference to assess arterial pressure (AP) and estimate hemorrhage, in accordance with the present technology.

[0051] FIGURE 11B shows a method for assessing hemorrhage stages using a wearable capacitive sensing system, with its performance compared to a catheter-based gold standard system, in accordance with the present technology.

[0052] FIGURES 11 C- 11 E are comparison of EKG signal, femoral arterial pulse, and high-frequency AC (l~20 Hz), and comparison of HR derived from reference EKG and AC, in accordance with the present technology.

[0053] FIGURES 12A-12F are real-time measurements of offset value, heartrate (HR), isovolumetric contraction time (IVCT), and root mean square (RMS) of SCG derived from AC, along with reference mean arterial pressure (MAP), maximum slope within one beat (dP dt- Imax), and HR during the progression of liver hemorrhage and subsequent fluid resuscitation, in accordance with the present technology.

[0054] FIGURES 12G-12H are estimations of stages of hemorrhagic shock using a random forest model based on capacitive signals (FIG. 12H) and reference signals (FIG. 12G), in accordance with the present technology. (N: normal; SI : mild shock; S2: moderate shock; S3: severe shock)

[0055] FIGURE 121 is an estimation of stages of hemorrhagic shock in a subject without reaching the S3 stage, in accordance with the present technology.

[0056] FIGURE 12J shows a correlation between sensor’s offset value and reference MAP among 7 subjects, in accordance with the present technology, (r: Pearson correlation coefficient; p: Spearman correlation coefficient; c: cross-correlation coefficient).

[0057] FIGURES 13A-13B show feature importance for reference catheterbased- (FIG. 13 A) and wearable capacitive systems (FIG. 13B) during random forest classification, in accordance with the present technology.

[0058] FIGURES 14A-14E show an example CPC array system, and measurement results, in accordance with the present technology.DETAILED DESCRIPTION

[0059] Described herein is a novel single-electrode capacitive sensor designed to measure regional body capacitance changes caused by variations in dielectric constant and pressure. The sensor features a carbon nanotube-paper composite (CPC) electrode integrated with a multi-walled carbon nanotube (MWCNT)-embedded foam. The CPC electrode, with its large surface area and high-aspect-ratio fibers, generates a high electric field for deeper tissue penetration, improving the in-body monitoring performance. Penetration depth is characterized by using surrogate tissue, heart, and lung models.Additionally, the integration of pressure-sensitive MWCNT foam significantly enhances the sensitivity, enabling precise detection of regional blood volume and tissue displacement. The novel sensing mechanism is applied to detect internal hemorrhage in a porcine model. By employing a machine learning algorithm, the sensor accurately estimates the severity of internal hemorrhage, offering a non-invasive alternative to traditional catheter-based systems. This advancement lays the foundation for a real-time wearable system that monitors critical health metrics, such as blood volume, blood pressure, and lung function.

[0060] To meet the growing demand for intelligent wearable sensing, disclosed herein is a novel single-electrode capacitive sensor made of carbon nanotube composites. The sensor is designed with a CPC electrode, layered with a MWCNT-embedded foam on top. The large surface area and high-aspect-ratio fibers of the CPC electrode generate a high-density electric field, leading to greater L compared to a copper electrode. Using surrogate models, including heart, lung, and artery, regional in-body biometrics are characterized in terms of dielectric constants and pressure changes. By incorporating a pressure-sensitive MWCNT foam, the sensor significantly improves its sensitivity to dielectric constant and pressure, enabling precise detection of in-body composition changes and tissue displacement. This advancement enabled detailed, real-time monitoring of internal hemorrhage in a porcine model. Leveraging machine learning algorithms, the sensor becomes a non-invasive and intelligent wearable device for assessing internal hemorrhage, which is characterized in comparison to invasive sensors.

[0061] FIG. 1A is a wearable single -electrode capacitive sensor, in accordance with the present technology. In one aspect, disclosed herein is a capacitive sensor including a CPC electrode, and a MWCNT foam disposed on the CPC electrode. In some embodiments, the CPC electrode includes cellulose paper and a plurality of MWCNTs. In some embodiments, the plurality of MWCNTs form a plurality of cantilever fibers along an edge of the CPC electrode (as shown in FIG. 3A). In some embodiments, the CPC electrode further includes silver ink. In some embodiments, the CPC electrode is circular.

[0062] In some embodiments, the MWCNT foam is cylindrical. In some embodiments, the MWCNT foam includes polyurethane (PU). In some embodiments, the capacitive sensor is configured to sense a change in capacitance based on pressure exerted on the MWCNT foam.

[0063] FIGs. 1B-1C are schematic illustrations of a capacitive sensor detecting variations in dielectric constant and pressure-induced displacement for in-body- and hemorrhage monitoring. In some embodiments, a single-electrode capacitive sensor monitors in-body biometrics by detecting variations in dielectric constants and pressure- induced displacement. Changes in the regional volume of tissue, air, and body fluids with distinct permittivity and geometries alter the effective dielectric constant. Additionally, fluid volume changes, particularly those associated with pulsation such as arterial pulse and seismocardiography (SCG), can induce tissue displacement and delicate pressure changes at the skin surface. The composition changes typically occur 10-50 mm beneath the skin, presenting a significant challenge for capacitive sensor design. Conventional coplanar capacitive sensors have a limited penetration depth (XL) of just a few millimeters, constrained by the fringing field between the excitation and ground electrodes, making it challenging to monitor in-body conditions. A single-electrode capacitive sensor is proposed to achieve an enhanced XL and pressure sensitivity. When a single excitation electrode is placed on the body, a floating electric ground forms in the body where the electric field strength drops below the noise level, indicating the XL. The sensing region is defined as a hemisphere with a radius of XL centered at the sensor location.

[0064] To maximize both XL and pressure sensitivity, a single-electrode sensor consists of one capacitive electrode made of a CPC electrode and MWCNT foam (FIG. 1A). Without a ground electrode, the electric field of a single-electrode sensor theoretically penetrates the body infinitely, implying / .L= / . In practice, the actual XL was limited by the electric field strength and capacitance noises. In previous reporting on liquid-level detection using a single-electrode sensor, as the liquid level exceeded 100 mm, the capacitance increased at a progressively smaller rate and eventually reached saturation at400 mm. This liquid level indicated the XL. Without raising the excitation voltage, XL could be increased by enhancing the electric field strength using high-aspect-ratio conductive fibers. Furthermore, the pressure sensitivity was enhanced through a MWCNT-embedded foam, inducing the percolation changes under deformation.

[0065] FIGs. 2A-2C show a fabrication method of a capacitive sensor consisting of a CPC electrode and a MWCNT foam, in accordance with the present technology. In some embodiments, a wet-punching fabrication protocol allows for the production of an array of circular CPC electrodes surrounded with numerous cantilever-shaped fibers. In another aspect, disclosed herein is a method of manufacturing the capacitive sensor described herein, including depositing a first solution of MWCNTs onto the cellulose paper, screen printing silver ink onto the cellulose paper, and punching the cellulose paper with a circular punch to form the CPC electrode. In some embodiments, the method further includes cutting a foam sheet to form a cylindrical foam, coating the cylindrical foam with a second solution of MWCNTs to form the MWCNT foam, and positioning the MWCNT foam onto the CPC electrode.

[0066] Based on a previously established wet-fracture technique, described in US Pat. App. No. 18 / 258776, the entire disclosure of which is hereby incoiporated by reference, a novel punching approach was developed to enhance the scalability and sensitivity. In one example, the CPC electrode with an 11 mm diameter was devised to enlarge the number of cantilever-shaped fibers and the capacitance response. First, cellulose paper (20 x 20 cm2) was coated with MWCNTs by applying an aqueous solution of 5 mg mL-l-MWCNT dispersed in a surfactant (sodium dodecyl sulfate; SDS; 1%). The resulting CPC had a sheet resistance of 295.0 ± 38.3 sq measured with a four-point probe. Silver ink was then screen-printed on the CPC for electrical connection. After wetting with water drops, the developed metal punch-die system was used to fabricate circular CPC electrodes in a 2x10 array. Finally, the CPC electrode was electrically wired and protected by a 70 pm-thick polyimide film. In the wet-punching step, the fracture along the circumference produced individual, cantilever-shaped fibers. These high-aspect-ratiofibers were aligned along the stretching, radial direction. The individualized fibers had a linear density, average length, and average thickness of 30.5 ± 3.4 nun-1, 1.3 ± 0.5 mm, and 5.8 ± 2.1 pm, respectively. The average aspect ratio was 224 (length / thickness), which allows for generating a high electric field (2.0 kV / m).

[0067] The pressure-sensitive MWCNT foam was fabricated by dip-coating a polyurethane (PU) foam in 0.4 mg mL-1 MWCNT solution (1% SDS-DI water). The foam had a diameter of 20 mm, a height of 4 mm, and a density of 2.2 x 10-5 g mm3. MWCNTs were nonspecifically coated on the cellulose paper and PU foam, as confirmed by X-ray analysis. Finally, the MWCNT foam was placed on the CPC electrode to assemble the capacitive sensor.

[0068] FIG. 3A is a SEM image of the individualized conductive fibers in a CPC electrode and real image of a capacitive sensor, in accordance with the present technology. FIG. 3B is a graph showing initial capacitance of the CPC electrode and capacitive sensor (N=4), in accordance with the present technology. Given the reproducible punching and coating methods, the initial capacitance (CO) of the sensor was 0.70 ± 0.02 pF, with a standard deviation of only 2.2 % (N=4).

[0069] FIG. 4A is a sensing mechanism and the equivalent circuit model of a single-electrode capacitive sensor detecting in-body dielectric constant (ein-body) within the penetration depth ( L), in accordance with the present technology. FIG. 4B shows a surrogate human tissue model and SNR analysis for XL characterization, and a graph showing AC of a metal object approaching CPC electrodes with various diameters in the surrogate human tissue model, in accordance with the present technology.

[0070] The single-electrode capacitive sensor, featuring a highly sensitive electrode and a pressure-sensitive foam, measures variations in dielectric constant and pressure inside the body within an extended L. The equivalent circuit model of the sensor involved three serially connected capacitances. CCPC denotes the capacitance of the CPC electrode, while Cpressureis the capacitance of the MWCNT foam that compresses under pressure. The in-body capacitance (C£) varies with changes in tissue and body fluidcomposition. When body fluid and tissue are homogeneously combined, the in-body permittivity (ein-body ) can be expressed as:Gn-body ^0 (T1T1 " ^2^2Equation 1

[0071] where 80 is vacuum permittivity (8.85 x 10-12 F m-1), el is the dielectric constant of tissue (si -30), s2 is the dielectric constant of body fluid (s2~80), vl is the volume fraction of tissue, and v2 is the volume fraction of body fluid. Ce, which is proportional to ein-body, represents a hemispherical capacitor constrained by a floating ground. The radius of the hemispherical sensing region, denoted by L, remains unchanged for small composition changes. Within the sensing region, variations in vl and v2 result in ein-body changes. For example, v2 increase leads to a higher sin-body, thus increasing Cs. Additionally, v2 changes induce pressure and displacement on the skin through the tissue, causing deformation of the MWCNT foam and subsequently altering Cpressure. The combined effect of sin-body and pressure increases the overall capacitance response.

[0072] FIG. 5A shows a simulation of an electric field generated by CPC electrodes with diameters of 11 mm and 1.5 mm in a human tissue environment, in accordance with the present technology. FIG. 5B shows the XL of a copper electrode, a CPC electrode, and a sensor combining a CPC electrode and a MWCNT foam, in accordance with the present technology. FIG. 5C is a graph showing AC of variations in dielectric constant using a copper electrode, a CPC electrode, and a CPC electrode integrated with a MWCNT foam, in accordance with the present technology. FIG. 5D shows a sensor combining a CPC electrode and MWCNT foam exhibited an improved sensitivity to the pressure (S: 1.4 pF kPa-1) within the practical preload pressure range when worn on body (0.6-1.3 kPa), in accordance with the present technology. FIGs. 5E- 5F is a graph showing a comparison between the AC and real-time actual pressure under a 2 Hz-sinusoidal wave, in accordance with the present technology.

[0073] In the single-electrode configuration, the CPC electrode and MWCNT foam increase the surface area and electric field, thereby extending XL. To evaluate XL, asurrogate human tissue model was developed. NaCl solution ( 12 L at 0.9 g L-l), simulating an extracellular environment, was prepared in a plastic tank. A capacitive sensor was mounted on one side of the tank, while the liquid and tank were electrically grounded on the opposite side. Since the human body acts as a floating ground, the electric ground was applied to ensure the electric field converged into the solution. Without the electric ground, a much larger volume of liquid (> 40 L) would be necessary to achieve the same effect, which was impractical for constructing a surrogate model. A moving metal object was used to simulate the displacement of a floating conductor in the body, such as a blood vessel, and to determine XL. Capacitance changes (AC= Ci - CO) were measured when the metal object approached the sensor. CO and Ci denote the initial capacitance and the capacitance with the moving metal object at time i, respectively. XL was defined when AC exceeded the noise level of the capacitance measurement setup in the model. The noise level was determined as the standard deviation (o) of the initial capacitance signal in the model.

[0074] For XL characterization in the surrogate model, the CPC electrodes with 1.5, 3, 7, and 11 mm-diameters were prepared using the wet-punching methods and then mounted at the side of a tank. As the diameter increased, XL increased due to the larger CO, resulting in greater AC. XL of CPC electrodes with different diameters ranged from 41 to 107 mm. The relationship between surface area and electric field was numerically computed using Ansys 3D Maxwell. At the same distance, the 11 mm-diameter electrode exhibited the largest electric field strength and the smaller gradient, indicating the largest XL.

[0075] When the surface area was further increased by adding a MWCNT foam (diameter: 20 mm), XL increased to 124 mm, 1.2 times that of a CPC electrode alone (Figure 2e). The XL of a CPC electrode combined with MWCNT foam was 1.3 times that of a copper electrode with an 11 mm diameter. At a 50 mm distance, the signal-to-noise ratio (SNR) of a sensor with a CPC electrode and MWCNT foam (SNR: 10.8) was 2.0 and 3.5 times that of a CPC electrode and a copper electrode alone, respectively. The higher SNR resulted from both larger capacitance and a stronger electric field.

[0076] In addition to the sensor’s initial capacitance, L was influenced by the electrical properties of a medium. As the larger permittivity and conductivity of liquid media increased, XL decreased. Given the human body was composed of an inhomogeneous matrix and complex fluids, XL was expected to vary across different body regions. In the human body, the extended XL of 124 mm based on the general tissue model was at least 2.4 times the clinically relevant depth of 10-50 mm, making it suitable for detecting in-body biometrics in most body regions.

[0077] Axial resolution was also characterized in the surrogate model and calculated as the ratio of noise level to sensitivity within the clinically relevant region (10-50 mm). The sensor achieved an axial resolution of 1.5 mm in depth. Both axial and lateral resolutions can be enhanced by employing an array of single-electrode capacitive sensors, as the higher sensor density allows finer detection of lateral variations. The enhanced XL and axial resolution of the single-electrode capacitive sensor shows significant potential for monitoring in-body biometrics at greater depth compared to previously reported sensors (Table 1).Table 1. Performance of wearable sensors for inbody biometrics monitoringAxialReference Method Setup / .i [mm]a)resolution[mm] iZhao,W r earableA„ _ „ , _Li ■g ,ht . Devi .ces Optical ArrayJ0.5 -0.15Wang,ConformalUltrasonic Ultrasound Array 40 0.4DeviceLo Presti,SkinBioimpedance Impedance Array -10 -10SensorHuang,ElectricalCapacitance _.1.. omograp ,hy Capacitance ArrayJ-60 -2 sensorPresent SingleApplication Capacitance electrode 124 1.5;i,Penetration depth.

[0078] The developed sensor detects the changes in dielectric constants with high sensitivity. In one example, AC in response to dielectric constant changes was tested using four different media to simulate the physiological range of dielectric constants (a: 20-80). Liquid media, including isopropyl alcohol (IP A), deionized (DI) water, NaCl solution (0.9 g L-l), and polyurethane (PU) foam in NaCl solution (0.9 g L-l ) were prepared in a 10 mL glass beaker. The sensor was positioned beneath the beaker, separated by a thin, rigid spacer. Compared to copper and CPC electrodes, the combination of CPC electrode and MWCNT foam exhibited the largest AC (Figure 2f). It achieved a sensitivity of 0.45 fF s- 1, 1.7 and 2.5 times those of a CPC electrode and a copper electrode alone, respectively (Figure S10).

[0079] The developed sensor also demonstrated an enhanced sensitivity to pressure. When pressure was applied using an electrically grounded metal plate fixed to a stepper actuator, the percolation changes in the MWCNT foam increased the pressure sensitivity. According to the pressure characteristics, AC exhibited three distinctsensitivities: 0.5, 1.4, and 0.7 pF kPa-1 in the low-pressure region (0-0.5 kPa), intermediate-pressure region (0.5-2 kPa), and high-pressure region (2-6 kPa), respectively. The sensor demonstrated the highest sensitivity within the practical preload pressure range when attached to the body (0.6- 1.3 kPa). It was found that the nonlinear stiffnesses of the MWCNT foam resulted in different sensitivity regions. The sensitivity (1.4 pF kPa-1) was 9.6 times that of a CPC electrode alone due to the percolation changes caused by MWCNT foam deformation. The detection limit (DL) for pressure was evaluated by SNR analysis. DL was defined when SNR was 3. DL in the physiological pressure range was 6 Pa. The pressure sensing performance was reproducible (N=4), showing an error below 4.6 %.

[0080] The developed sensor exhibits a robust response under dynamic loading conditions. Under a 2 Hz-cyclic compression (0.6-0.8 kPa), a developed sensor exhibited a response time of 30 ms and a recovery time of 40 ms. The real-time pressure was computed from the displacement of the actuator, obtained by taking the double integral of the measured acceleration (accelerometer, ISM330DHCX, SparkFun). These response and recovery times enabled the developed sensor to capture cyclic motion up to 25 Hz (1 / 40 ms-1), covering SCG frequency ranges (1-20 Hz). Furthermore, a 1500-cyclic pressure test under 0.5-1 kPa showed a consistent response within 0.2 % AC changes after 1500 cycles (Figure SI 4).

[0081] In another aspect, disclosed herein is a sensor system comprising four capacitive sensors as described herein. In some embodiments, a first capacitive sensor of the four capacitive sensors and a second capacitive sensor of the four capacitive sensors are configured to be positioned on an intercostal mid-clavicular area of a patient.

[0082] FIG. 6A shows capacitive heart and lung models, in accordance with the present technology. FIG. 6B shows AC as the volume of an infusion bag increases (saline solution injection) in the heart model, in accordance with the present technology. FIGs. 6C- 6D show AC as the volume of an infusion bag increases (air injection) in the lung model, in accordance with the present technology.

[0083] FIG. 7A-7B is a AC and spectrogram of human heart biometrics in different frequency ranges using four capacitive sensors (SI & S2 positioned above the heart; S3 & S4 located 20 cm away from the heart), in accordance with the present technology. FIG. 7B is a AC of human lung biometrics throughout a breathing cycle — inhale, hold, exhale, hold — using four capacitive sensors (SI and S2 placed over the chest; S3 and S4 positioned on the abdomen), in accordance with the present technology.

[0084] The large XL and high sensitivity of the developed sensor enable the detection of in-body biometrics, validated in simplified heart and lung models. Capacitive heart and lung models were developed to simulate AC to blood- and air volume changes in these organs. Two PU foams soaked in a 0.9 g L1NaCl solution were used as surrogates for human tissue. The top of the foams was electrically grounded. An infusion bag (120x80x60 mm3) connected to a syringe was placed between two foams to simulate the heart or lung. Either saline solution (s~78) or air (e~I) was injected into the bag via the syringe to mimic the increased volume of the heart or lung, respectively. The sensor was positioned beneath the foam with a 30 mm distance to the infusion bag.

[0085] Two different constraints for a sensor were tested to characterize the effect of dielectric constant and pressure changes. In the first constraint, the sensor was physically constrained under the foam by medical tape. The volume changes in the bag induced variations in the dielectric constant and exerted pressure on the sensor due to physical expansion or contraction (ACs+pressure). In the second constraint, a sensor was located through a rigid spacer without a constraint, and the pressure effect on the sensor was eliminated, isolating the effect of dielectric constant-induced AC (i.e. ACs). This corresponded to a wearable sensor placed with a rigid spacer between the sensor and the skin, secured with double-sided tape.

[0086] In the heart model, as the volume of saline solution in the bag increased from 50 niL to 300 mL, both ACe and ACs+pressure increased. This was due to the higher effective dielectric constant (E2 > si) and the compression of the MWCNT foam caused by the bag’s expansion. In the lung model, as the air volume in the bag increased, theeffective dielectric constant (s2 < si) decreased, leading to a reduction in ACs. However, ACs+pressure still increased due to the greater AC from the compression of the MWCNT foam. The simulated ACs showed similar trends in the numerical models for both heart and lung.

[0087] Based on a human subject study, the developed sensor detects the heart and lung biometrics by measuring the changes in dielectric constant and pressure. Four sensors were attached to the skin: sensors 1 and 3 (SI & S3) detected both dielectric constant and pressure, physically constrained by a medical tape, while sensors 2 and 4 (S2 & S4) measured only dielectric constant changes, with a rigid spacer placed between the sensors and skin, attached using a double-sided tape. In some embodiments, a third capacitive sensor and a fourth capacitive sensor of the four capacitive sensors are configured to be positioned on an upper pectoral area of the patient. For heart monitoring, sensors S 1 and S2 were positioned on the skin over the left 4-5th intercostal mid-clavicular area, while sensors S3 and S4 were placed on the skin over the right upper pectoral area about 20 cm away from the position of SI and S2. In the low-frequency range (0.5-2 Hz), both SI and S2 detected heartbeat frequency signals resulting from blood volume changes in the thorax, including the heart and major arteries. In contrast, S3 and S4 showed minimal responses in this range. In the high-frequency range (2-35 Hz), SI successfully detected mechanical vibrations from the heart (SCG), while the signals from S2, S3, and S4 were indistinguishable from noise (Figure SI 6).

[0088] In some embodiments, a third capacitive sensor and a fourth capacitive sensor of the four capacitive sensors are configured to be positioned on an abdomen of the patient. For lung monitoring, S 1 and S2 were positioned on the skin over the chest in the right 4-5th intercostal mid-clavicular area, while S3 and S4 were placed over the abdomen (Figure 3e). Both SI and S3 captured the physical effort involved in the breathing cycle — inhale, hold, exhale, and hold — because of the constraint imposed by the medical tape. Given the low air permittivity compared to tissue, S2 detected lung volume changes, with AC decreasing as more air filled the lungs. However, AC at SI increased during inhalationas the expanding lungs and thorax compressed the sensor. Note that the absolute AC at SI was ~10 times that at S2. With the constraint of a medical tape, AC was primarily influenced by lung and thorax expansion rather than the air-induced dielectric decrease. These findings highlight the potential of the developed sensor for monitoring blood and air volumes in the thorax through the low-frequency AC signal.

[0089] In another aspect, disclosed herein is a sensor system including two capacitive sensors as described herein. The developed sensor’s high sensitivity allows for the detection of blood volume in distal extremity regions. In general, increased blood pressure (BP) leads to a higher regional volume of blood in arteries, capillaries, veins, and interstitial fluid. This increased volume results in a higher effective dielectric constant and tissue expansion. A simplified capacitive artery model was developed to simulate AC for blood volume fluctuations in distal extremity regions. Using a setup similar to the previous section, a plastic tube was placed 10 mm above the sensor to mimic an arterial vessel. Three tubes with varying outer and inner diameters were filled with saline solution to test the AC. AC increased as more saline solution filled the tube, resulting in a higher effective dielectric constant.

[0090] FIGs. 8A-8B show a model setup and schematic illustration of the relationship between BP and blood volume in arteries and veins, in accordance with the present technology. FIG. 8C shows a AC as the geometry of the tube increases in the artery model, in accordance with the present technology. FIG. 8D is a comparison of reference MBP and low-frequency AC during Valsalva and handgrip maneuvers, with SI detecting both dielectric constant and pressure changes, while S2 detects dielectric constant changes, in accordance with the present technology.

[0091] In some embodiments, a first capacitive sensor of the two capacitive sensors and a second capacitive sensor of the two capacitive sensors are configured to be positioned on a forearm of a patient. In some embodiments, the sensor system is configured to detect a global change in blood volume of the patient.

[0092] The sensor’s capability to detect changes in blood volume in distal extremities while minimizing muscle artifacts was tested on healthy human volunteers. Sensor 1 (SI) was physically constrained by a medical tape, and sensor 2 (S2) had a rigid spacer placed between the sensor and the skin. Both sensors were attached to the forearm, alongside a continuous BP monitoring finger cuff (VitalStream, Caretaker Medical, VA, USA) worn on the middle finger as a reference. To examine the correlation between AC and reference BP while reducing muscle artifacts during handgrip with the forefinger and thumb, sensors SI and S2 were strategically positioned in areas with minimal muscle activity. Handgrip was performed, demonstrating muscle movements in the forearm. However, near the elbow, SI and S2 did not detect any noticeable muscle activity during the handgrip, as evidenced by the AC and spectrogram.

[0093] FIGs. 9A-9C shows a correlation between reference MBP and AC from two capacitive sensors during Valsalva and handgrip maneuvers, (r: Pearson correlation coefficient; p: Spearman correlation coefficient), in accordance with the present technology.

[0094] FIGs. 10A-10C show comparisons of reference MBP and low-frequency AC from two capacitive sensors placed on different arms during Valsalva, right-arm handgrip, and left-arm handgrip maneuvers, and the correlation between reference MBP and AC from two capacitive sensors during Valsalva, right-arm handgrip, and left-arm handgrip maneuvers, in accordance with the present technology.

[0095] The low-frequency capacitance signal reflects the hemodynamic signals, including blood volume and BP changes, while mitigating the impact of noise, muscle artifacts, and irrelevant physiological signals. Two BP-perturbing maneuvers, Valsalva and sustained isometric handgrip, were performed to alter the BP in the forearm. Given the low- frequency nature of hemodynamics (0.01-0.4 Hz), low-frequency AC (< 0.2 Hz) was analyzed to estimate the blood volume changes induced by the maneuvers and to mitigate the effect of regular respiratory activity (0.2-0.4 Hz). Both SI and S2 showed the trends consistent with the reference mean blood pressure (MBP). During these maneuvers, bothMBP and AC initially decreased due to disrupted blood flow. However, they increased as the sympathetic nervous system activated to raise BP. The correlation between MBP and AC was evaluated using Pearson (r) and Spearman (p) correlation coefficients. Given the nonlinear correlation between BP and blood volume, AC at SI exhibited a higher correlation (r: 0.50 and p: 0.49) by detecting changes in the dielectric constant and pressure from the expansion of arteries, veins, and interstitial fluid. Additionally, the absence of a rigid spacer mitigated the capacitance coupling effect, resulting in a more pronounced AC of SI than S 2.

[0096] FIG. 11A is an experimental setup involving a capacitive sensor positioned on the skin at the mid-sternum, along with a femoral artery catheter and EKG as reference to assess arterial pressure (AP) and estimate hemorrhage, in accordance with the present technology.

[0097] FIG. 11B shows a method for assessing hemorrhage stages using a wearable capacitive sensing system, with its performance compared to a catheter-based gold standard system, in accordance with the present technology. FIGs. 11C-11E are comparison of EKG signal, femoral arterial pulse, and high-frequency AC ( 1—20 Hz), and comparison of HR derived from reference EKG and AC, in accordance with the present technology.

[0098] In some embodiments, a first capacitive sensor of the two capacitive sensors is configured to be positioned on a first arm of a patient and a second capacitive sensor of the two capacitive sensors is configured to be positioned on a second arm of the patient. In some embodiments, the sensor system is configured to detect a regional change in blood volume of the patient.

[0099] The developed sensor demonstrates the capability to estimate regional blood volume. To further determine the sensor’s sensitivity to regional vs. global changes in blood volume, one sensor was attached to the right forearm (SR) and the other to the left forearm (SL), physically constrained by medical tapes. A continuous BP monitoring finger cuff was worn on the right hand. Valsalva, right- arm sustained isometric handgrip, and left-arm sustained isometric handgrip were sequentially performed. During Valsalva, both SR and SL demonstrated a similar trend with the reference MBP. However, only SR and MBP exhibited significant responses during the right-arm handgrip. This response was due to changes in regional hemodynamics in the right arm caused by the right-arm handgrip. In contrast, during the left-arm handgrip, SL exhibited a significant response, while SR and MBP did not, due to different regional hemodynamic conditions. SR showed good agreement with the MBP (r: 0.72 and p: 0.38), whereas SL had poor agreement (r: -0.15 and p: 0.03).

[0100] In addition, similar responses and correlations were observed among different subjects (N=3). In event-based analysis, SR showed good correlations ranging from 0.5 to 0.9 among different maneuvers. In contrast, SL also displayed a good correlation (~0.6) during the Valsalva but negative correlations ranging from -0.2 to -0.8 during handgrips. This difference arose because BP changed centrally during Valsalva, while during handgrips, BP first altered regionally before becoming central. SR measured regional BP on the same arm with the reference, whereas SL assessed regional BP on the opposite arm. These results highlighted the sensor’s potential for estimating regional hemodynamics and blood volume qualitatively.

[0101] In yet another aspect, disclosed herein is a method of detecting a hemorrhage with the capacitive sensors disclosed herein, the method including denoising one or more capacitive signals from the capacitive sensor, deriving an offset value, an isovolumetric contraction time (IVCT), a root mean square (RMS) value of mechanical vibrations from a heart (SCG), a heart rate (HR), or a combination thereof from the one or more capacitive signals, and detecting a hemorrhage with a machine-learning algorithm based on the offset value, the IVCT, the RMS value of SCG, the HR, or a combination thereof.

[0102] The developed sensor, aided by a machine learning method, enabled the assessment of stages of hemorrhagic shock from liver injury. Hemorrhage tests were carried out using a porcine model that included general anesthesia, mechanical ventilation,and instrumentation for continuous arterial pressure (AP) monitoring. Hemorrhagic shock was induced by direct liver laceration, followed by a 10-minute period of free bleeding into the abdomen. Animals then received intravenous fluids to replace blood loss and increase BP. A developed sensor, positioned on the skin at the mid-sternum, was attached by a medical tape and covered with a waterproof polyurethane to prevent liquid from entering the sensor and to mitigate the humidity effect. The sensor used a low-frequency AC offset signal (< 0.2 Hz) to estimate cardiac blood volume and a high-frequency AC signal (l~20 Hz) to discern cardiac function, including cardiac contraction time, via SCG. A catheterbased gold standard system (BIOPAC Systems Inc., CA, USA) continuously measured electrocardiogram (EKG) and femoral arterial pressure as reference signals.

[0103] FIGs. 12A-12F are real-time measurements of offset value, heartrate (HR), isovolumetric contraction time (IVCT), and root mean square (RMS) of SCG derived from AC, along with reference mean arterial pressure (MAP), maximum slope within one beat (dP dt-lmax), and HR during the progression of liver hemorrhage and subsequent fluid resuscitation, in accordance with the present technology.

[0104] A machine learning method was applied to assess the degree of hemorrhage and severity of hemorrhagic shock in the porcine model. First, the capacitive signals were denoised using wavelet decomposition (db4, level 2) and Savitzky-Golay filtering in MATLAB. The offset value, isovolumetric contraction time (IVCT), root mean square (RMS) value of SCG, and heart rate (HR) were derived from the AC (Figure SI 9). Mean arterial pressure (MAP), maximum derivative of arterial pressure (dP dt-lmax) during each pulsation, and HR were derived from the reference signals. Engineering features, including mean, standard deviation, and gradient values, were extracted from every 5-second signal window. These features were then used to train a random forest model to detect the severity of hemorrhage.

[0105] FIGs. 12G-12H are estimations of stages of hemorrhagic shock using a random forest model based on capacitive signals (FIG. 12H) and reference signals (FIG.12G), in accordance with the present technology. (N: normal; S 1 : mild shock; S2: moderate shock; S3: severe shock)

[0106] The developed sensor accurately captures the prominent peaks (aortic valve opening) in the SCG signal, following the reference EKG and leading to the arterial pulse. The HR derived from the SCG signal showed good agreement with that computed from the reference EKG, with a high correlation (r: 0.97). Additionally, the sensor demonstrated a low mean absolute error (MAE) of 2.9 beats per minute (BPM) and a low standard deviation of absolute error (SDAE) of 2.3 BPM compared to the EKG.

[0107] Liver hemorrhage tests were performed on 7 pigs. After establishing baseline measurements, each subject underwent free bleeding from liver laceration, followed by intravenous fluid resuscitation.

[0108] FIG. 121 is an estimation of stages of hemorrhagic shock in a subject without reaching the S3 stage, in accordance with the present technology.

[0109] FIG. 12J shows a correlation between sensor’s offset value and reference MAP among 7 subjects, in accordance with the present technology, (r: Pearson correlation coefficient; p: Spearman correlation coefficient; c: cross-correlation coefficient).

[0110] Four stages of shock were established to demonstrate the progression of MAP changes in response to blood loss: normal (N), mild shock (SI), moderate shock (S2), and severe shock (S3). A maximum 30% decrease in MAP from baseline was classified as the SI stage. Greater than 30% decreased MAP from baseline was classified as the S2 stage. If the MAP remained low after fluid resuscitation, it was classified as the S3 stage.

[0111] The clinically important features derived from AC, trained in a random forest model, estimate shock stages with an accuracy of over 99%, which is comparable to that of reference invasive arterial pressure sensors. A random forest model with 50 trees (MATLAB), validated through 5-fold cross-validation, was used to predict shock stages using a small dataset (N=7). The model achieved an accuracy of 0.994 and an Fl score of 0.975 using the sensor signals. These results were comparable to the accuracy (0.995) and Fl score (0.979) using the reference signals. Furthermore, in a survival case, the modelsuccessfully detected that the animal survived the liver injury without reaching the S3 stage, with MAP and offset values showing a slight drop that returned to normal after fluid resuscitation.

[0112] Among 7 porcine subjects, the offset value of AC showed moderate to high correlations with invasively monitored MAP. The Pearson (r), Spearman (p), and cross-correlation (c) coefficients were 0.58, 0.62, and 0.76, respectively. This highlights the sensor's ability to estimate regional blood volume using the offset value.

[0113] FIGs. 13A-13B show feature importance for reference catheter-based- (FIG. 13 A) and wearable capacitive systems (FIG. 13B) during random forest classification, in accordance with the present technology.

[0114] Feature importance for predicting the shock stages was analyzed in MATLAB. For reference signals, the mean value of MAP was the most significant feature in the prediction. In capacitive signals, the mean values of offset, RMS, and HR were key features contributing to the highly accurate prediction of the stages of shock.

[0115] In discussion, the extended XL and enhanced sensitivity of the singleelectrode capacitive sensor enabled qualitative detection of blood volume changes and subtle heart vibrations. Combined with a machine learning algorithm, this wearable sensor demonstrated high accuracy in predicting the severity of hemorrhagic shock, comparable to an invasive catheter-based system. Several wearable sensing systems have demonstrated potential in real-time hemorrhage monitoring with minimal medical instruments. Mechanical sensing methods, including seismocardiography (SCG) and phonocardiography (PCG), are used to measure heart vibrations or sounds to assess cardiac function and blood pressure during hemorrhage. These systems, enhanced by machine learning algorithms, can predict the severity of blood loss. While these systems exhibit low error rates of ~3% when estimating hypovolemia compared to invasive catheter-based systems, they still rely on indirect estimation of blood volume, limiting their ability to assess hemodynamics. In contrast, wearable ultrasound patches show more direct blood volume measurement. However, their complex fabrication methods and need for seamlesscontact make them expensive and cumbersome. This real-time, wearable, and non-invasive single-electrode capacitive sensor offers a promising alternative to advanced hemorrhage and in-body biometrics monitoring.

[0116] Disclosed herein is a novel single-electrode capacitive sensor comprising a CPC electrode and a MWCNT-embedded foam. The high-aspect-ratio fibers in the CPC electrode intensified the electric field, enabling highly sensitive detection of dielectric constant and pressure related to the tissue and body fluid compositions, as well as tissue displacement. With the integration of the pressure-sensitive MWCNT foam, the composite sensor achieved a pressure sensitivity of 1.4 pF kPa-1 and a detection limit of 6 Pa. The large surface area of the CPC electrode and MWCNT foam extended the penetration depth (XL) in the surrogate tissue model to 124 mm, with an axial resolution of 1.5 mm. The enhanced sensing performance enabled the detection of blood or air volume changes in the heart, arteries, and lungs. Further, mean blood pressure (MBP) could be qualitatively estimated based on the offset value changes using the capacitive sensor. In human subject tests involving hand grips and the Valsava maneuver, MBP showed good correlations ranging from 0.5 to 0.9. In a porcine model, the sensor, incorporating a machine learning algorithm, accurately assessed hemorrhage stages, achieving an accuracy of 0.994 and an Fl score of 0.975. These results were comparable to the accuracy (0.995) and Fl score (0.979) obtained from an invasive catheter-based system installed in the femoral artery. Compared to other wearable sensing systems capable of real-time hemorrhage monitoring using minimal medical instruments, the single-electrode capacitive sensor shows significant promise for advanced monitoring of hemorrhage and in-body biometrics, offering direct measurement.

[0117] FIGs. 14A-14E show an example CPC array system, and measurement results, in accordance with the present technology. In some embodiments, the system may be used for cardiac imaging.

[0118] In some embodiments, the system may be further configured to monitoring and / or measuring central blood pressure (such as of the heart and major arteries) and / or peripheral blood pressure (such as of the branchial and radial arteries).

[0119] As shown in FIG. 14A, in one aspect, an array of CPC sensors was developed. In some embodiments, the array may be a 4x4 sensor array patch. This sensor patch may be used to visualize the heart (i.e., for cardiac imaging).

[0120] FIGs. 14B-14E show the visualized arterial and ventricular movement of the heart. Specifically shown is the right artery (RA), left artery (LA), right ventricle (RV), and left ventricle (LV) of the imaged heart.

[0121] Accordingly, in another aspect, disclosed herein is a method of detecting a central or peripheral blood pressure (BP) with the capacitive sensor disclosed herein. In some embodiments, the method includes positioning one or more capacitive sensors at skin near a target location, deriving a mean capacitance value, an isovolumetric contraction time (IVCT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from the one or more capacitive signals, and detecting a BP with a machine-learning algorithm based on the mean capacitance value, the IVCT, the wave amplitude of SCG, the HR, the HRV, initial BP value, or a combination thereof.

[0122] In another aspect, disclosed herein is a method of cardiac imaging with an array of capacitive sensors as described herein. In some embodiments, the method includes positioning more capacitive sensors at skin near the heart, deriving a mean capacitance value, an isovolumetric contraction time (IVCT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from the one or more capacitive signals, and detecting cardiac function parameters: stroke volume, cardiac output, and ejection fraction with a machinelearning algorithm.

[0123] EXAMPLES

[0124] Fabrication of CPC Electrodes and MWCNT foam: A suspension of MWCNT (5 mg mL-1, Cheap Tubes Inc., VT, USA) in aqueous sodium dodecyl sulfate (SDS, 1%, Sigma-Aldrich, MO, USA) was deposited on a cellulose paper (Scott hard roll paper towels, Kimberly-Clark, 20 cm x 10 cm) twice. The CPC was then oven-dried at 80°C for 10 minutes. Silver ink (MG Chemicals, USA) was screen-printed onto the CPC (Figure lb). Drops of deionized water were applied to CPC to aid in wetting. An array of circular CPC electrodes was fabricated by punching the wet paper through the circular punches (Figure 1c). The 100 pm gap between the punch and die caused tensional fractures, resulting in numerous conductive fibers along the circular edges. A polyimide tape was used to package the CPC electrode, followed by wire connections. A 4 mm-high polyurethane (PU) foam sheet (density: 2.2 x 10-5 g mm3, Dream Solutions USA) was cut into a 20 mm diameter using a hole saw drill bit. MWCNT was coated on the PU foam through dip-coating with a 0.4 mg mL-1 MWCNT aqueous SDS solution.

[0125] Experimental setup for sensor characterization: Capacitance value was measured using a commercial chip (AD7747, Analog Devices Inc.) (Figure S5a). For penetration depth analysis, 12 L of 0.9 g L-l NaCl (Spectrum Chemical Mfg. Corp., NJ, USA) solution was prepared in a plastic tank (45 x 29 x 16 cm3) to simulate the physiological buffer. A sensor was mounted on one side of the tank, with the liquid and tank electrically grounded on the opposite side. A wire was used to connect the liquid and tank to the earth ground. A metal block (15 x 8 x 1 cm3) was placed inside the tank and moved towards the sensor to observe the capacitance response. For permittivity sensitivity analysis, a 10 mL glass beaker containing the dielectric medium was used (Figure S5b). A sensor was placed beneath the beaker with a thin polylactic acid spacer in between. For pressure sensitivity analysis, a thin, flat, electrically grounded metal plate was affixed to the stepper motor actuator to control the capacitive area and apply compression accurately (Figure S5c). The force applied to a sensor was measured with a load cell (DYMH-103 10kg, CALT sensor). Response / recovery time- and fatigue tests were conducted using the same setup and controller.

[0126] Experimental setup for capacitive heart, lung, and artery models: For the heart and lung models, two 30 mm-high PU foam sheets were soaked in a 9 g L1NaCl solution. An infusion bag made of parafilm (Amcor, USA) was placed in the middle. When fully inflated, the bag measured 120 x 80 x 60 mm3. A syringe was connected to the bag to inject liquid or air. For the artery model, a 10 mm-high PU foam sheet and a 30 mm- high PU foam sheet, both soaked in a 9 g L1NaCl solution, were used. Plastic tubes were inserted between the foam sheets and connected to the syringe for liquid injection. The capacitive sensor was positioned beneath the models, with the top of the wet foam sheet grounded electrically through a connection to the earth ground via a wire and a rectangular aluminum film (60 x 60 mm2).

[0127] Human subject test: The study related to HR and respiratory effort monitoring was approved by the Institutional Review Board (IRB) at University of Washington (UW) (IRB ID: STUDY00018275). The study for breathing tests and sensor characterization was approved by UW IRB (STUDY00013761). The study related to blood pressure (BP) monitoring with BP-perturbing maneuvers (Valsalva and handgrip) was approved by UW IRB (IRB ID: STUDY00019217). Valsalva involves inhaling deeply and then forcefully exhaling with a closed mouth and pinched nose for 20 sec. Handgrip exercise consists of squeezing a sponge cube between the forefinger and thumb for 30 sec.

[0128] Experimental setup and protocol for hemorrhage in porcine model: The animal test and protocol were approved by the Institutional Animal Care and Use Committee at UW (#:4329-02) and federal Animal Care and Use Review Office (#:DM210169.e002). Animals were held in a local vivarium and provided food and water ad libitum until the evening before experimentation. Intramuscular ketamine and xylazine were used for anesthesia induction, and general anesthesia was initiated and maintained with inhaled isoflurane throughout the procedure. The left femoral artery and vein were cannulated to continuously monitor arterial blood pressure and administer resuscitationfluids, respectively. Following a 10-minute baseline measurement of arterial pressure, a liver laceration was created by sharply dissecting the left lobe of the liver with a scalpel using a standardized template. The liver was placed back in the abdomen, the abdominal incision was closed, and the animals were allowed to bleed freely for 10 minutes. Animals were given intravenous fluids to replace blood loss and monitored for up to 3 hours. Survivors were euthanized under anesthesia with a pentobarbital overdose. Animals were euthanized earlier if the arterial pressure waveform was lost, indicating cardiac arrest and imminent death.

[0129] While illustrative embodiments have been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.

Claims

CLAIMSWhat is claimed is:

1. A capacitive sensor, comprising: a carbon nanotube paper composite (CPC) electrode; and a multi-walled carbon nanotube (MWCNT) foam disposed on the CPC electrode.

2. The capacitive sensor of Claim 1, wherein the CPC electrode comprises: a cellulose paper; and a plurality of multi-walled carbon nanotubes (MWCNTs).

3. The capacitive sensor of Claim 2, wherein the plurality of MWCNTs forms a plurality of cantilever-shaped fibers along an edge of the CPC electrode.

4. The capacitive sensor of Claim 1, wherein the CPC electrode further comprises silver ink.

5. The capacitive sensor of Claim 1, wherein the CPC electrode is circular.

6. The capacitive sensor of Claim 1, wherein the MWCNT foam is cylindrical.

7. The capacitive sensor of Claim 1, wherein the MWCNT foam comprises polyurethane (PU).

8. The capacitive sensor of Claim 1, wherein the capacitive sensor is configured to sense a change in capacitance based on pressure, permittivity, or a combination thereof exerted through the MWCNT foam.

9. The capacitive sensor of Claim 1 , wherein the CPC electrode is an excitation electrode, configured to apply an excitation voltage.

10. The capacitive sensor of Claim 9, wherein blood vessels of a patient, main arteries of a patient, veins of a patient, capillaries of a patient, or a combination thereof are a virtual electric ground.1 1. The capacitive sensor of Claim 1 , wherein the capacitive sensor is configured to detect biometrics of a heart, a lung, arteries, muscles, a brain, a liver, a kidney, a stomach, intestines, a pancreas, a spleen, skin, or a combination thereof.

12. The capacitive sensor of Claim 1, wherein the capacitive sensor is configured to be positioned on a sternum of a patient.

13. The capacitive sensor of Claim 12, wherein the capacitive sensor is configured to measure capacitance changes resulting from variations in effective tissue permittivity, displacement of a floating ground within blood, seismocardiography (SCG), or a combination thereof.

14. A sensor system comprising an array of capacitive sensors according to Claim 1.

15. The sensor system of Claim 14, wherein a first capacitive sensor of the array of capacitive sensors is configured to be positioned on a target location on a patient and a second capacitive sensor of the array of capacitive sensors is configured to be positioned on a control of the patient.

16. The sensor system of Claim 15, wherein the sensor system is configured to detect a global change in blood volume of the patient.

17. The sensor system of Claim 15, wherein the sensor system is configured to detect a regional change in blood volume of the patient.

18. The sensor system of Claim 14, wherein a first capacitive sensor of the array of capacitive sensors and a second capacitive sensor of the array of capacitive sensors are configured to be positioned on an intercostal mid-clavicular area of a patient.

19. The sensor system of Claim 18, wherein a third capacitive sensor and a fourth capacitive sensor of the array of capacitive sensors are configured to be positioned on an upper pectoral area of the patient.

20. The sensor system of Claim 18, wherein a third capacitive sensor and a fourth capacitive sensor of the array of capacitive sensors are configured to be positioned on an abdomen of the patient.

21. A method of detecting a hemorrhage with the capacitive sensor of Claim 1, the method comprising: positioning one or more capacitive sensors at skin near a target location; denoising one or more capacitive signals from the capacitive sensor; deriving an offset value, an isovolumetric contraction time (IVCT). a root mean square (RMS) value of mechanical vibrations from a heart (SCG), a heart rate (HR), or a combination thereof from the one or more capacitive signals; and detecting a hemorrhage with a machine-learning algorithm based on the offset value, the IVCT, the RMS value of SCG, the HR, or a combination thereof.

22. A method of detecting healing of a surgical site or wound with the capacitive sensor of Claim 1, the method comprising: attaching one or more capacitive sensors to a bandage; monitoring capacitance changes with the one or more capacitive sensors; determining an offset value, a root mean square (RMS) value, or a combination thereof; and detecting a hemorrhage with a machine-learning algorithm based on the offset value, the root mean square (RMS) value, or a combination thereof.

23. A method of manufacturing the capacitive sensor of Claim 1, the method comprising: depositing a first solution of MWCNTs onto the cellulose paper; screen printing silver ink onto the cellulose paper; and punching the cellulose paper with a circular punch to form the CPC electrode.

24. The method of Claim 22, further comprising: cutting a foam sheet to form a cylindrical foam; and coating the cylindrical foam with a second solution of MWCNTs to form the MWCNT foam.

25. A method of detecting a central blood pressure (BP) or a peripheral blood pressure (BP) with the capacitive sensor of Claim 1, the method comprising: positioning the capacitive sensor on skin near a target location; deriving a mean capacitance value, an isovolumetric contraction time (IVCT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from one or more capacitive signals of the capacitive sensor; combining physiological parameters, including age, sexes, weight, height, and body mass index (BMI); and detecting the central BP or the peripheral BP with a machine-learning algorithm based on the mean capacitance value, the IVCT, the wave amplitude of SCG, the HR, the HRV, initial BP value, or a combination thereof.

26. The method of Claim 25, wherein the capacitive sensor is one or more capacitive sensors.

27. A method of cardiac imaging with an array of capacitive sensors comprising one or more capacitive sensors of Claim 1, the method comprising: positioning the array of capacitive sensors on skin near a heart;deriving a mean capacitance value, an isovolumetric contraction time (IVCT), a wave amplitude of mechanical vibrations from a heart (SCG), a heart rate (HR), a HR variability (HRV), initial BP value, or a combination thereof from one or more capacitive signals from the array of capacitive sensors: combining physiological parameters, including age, sexes, weight, height, and body mass index (BMI); and detecting cardiac function parameters with a machine-learning algorithm, wherein the cardiac function parameters comprise left ventricle volume, aorta valve volume, stroke volume, cardiac output, and ejection fraction.

Images