JP7549357B2 - Soft Capacitive Pressure Sensor - Google Patents

Description

This application claims the benefit of U.S. Provisional Patent Application No. 62/757,329, filed November 8, 2018, and U.S. Provisional Patent Application No. 62/875,418, filed July 17, 2019, the specifications of which are incorporated herein by reference in their entireties.

The present invention relates to medical devices that read blood pressure. More particularly, the present invention relates to flexible, stretchable, capacitance-based pressure sensors for continuous wearable health monitoring applications. Furthermore, the present invention relates to neural network calibration of pressure sensors for continuous pressure readings.

Arterial pulses contain a wealth of cardiovascular information, including systolic and diastolic pressures, which are used clinically to quantitatively assess and monitor general health as well as cardiovascular disease. These arterial pulses can be measured at many different locations in the body, such as the brachial and radial arteries, using non-invasive blood pressure (NIBP) monitoring tools. Studies have shown that monitoring blood pressure fluctuations throughout the day can provide insight into cardiovascular health. For this reason, blood pressure should be monitored continuously to better understand blood pressure fluctuations and their impact on cardiovascular health.

Oscillometric measurements using inflatable upper arm cuffs have become widely used in clinical settings, but are intermittent, providing only one set of systolic and diastolic blood pressure values over a period of approximately 30-40 seconds. Alternatively, continuous beat-to-beat blood pressure monitoring requires detection and analysis of each cardiac cycle that pulsates the peripheral arteries. Volumetric finger cuff clamping is one method that can measure beat-to-beat blood pressure, but does not currently have a form factor that allows for ambulatory monitoring.

Arterial applanation tonometry is another method of NIBP in which an artery is locally flattened (applanated) using a pressure sensor, often a handheld probe. The degree of arterial applanation is then adjusted to measure the arterial pressure. Applanation tonometry relies heavily on the operator to stably place the pressure transducer over the artery to measure pulse pressure, which is not amenable to ambulatory continuous monitoring. Newer alternative methods use soft sensors that can improve conformability to the body. Conformity improves sensor-to-body coupling, allowing for more accurate measurement of arterial pulsation. However, currently reported soft sensors measure the transit time between two points of a pulse, rather than from the amplitude change of the pulse pressure waveform, as a means of calculating beat-to-beat blood pressure using theoretical models. Furthermore, there is a need to develop a highly sensitive pressure sensor that can detect small pressures (<5 kPa, approximately the pressure at the surface of the skin exerted by the radial pulsation from a healthy subject) and is characterized by a fast response time (approximately 100 ms, approximately the time between the diastolic and systolic points of the radial artery pulse wave).

Capacitive pressure sensors have attracted much attention due to their simple device design, fast response time, relatively small hysteresis, and low power consumption requirements, all of which are highly desirable for soft, wearable electronics applications. When soft dielectric materials are used, compression of the dielectric layer results in

where ε ₀ is the dielectric constant of a vacuum, ε _r is the relative permittivity of the dielectric material, A is the overlapping surface area of the parallel plate electrodes, and d is the distance between the parallel plate electrodes.

Capacitive sensors can be configured to "map" pressure using multiplexed data acquisition tools. For example, a grid of capacitive sensors can spatially resolve local pressures on a surface. The simple layout of capacitive sensors makes them an attractive modality for detecting mechanical pressure. However, current capacitive sensors typically feature low pressure sensitivity (below about 0.5-1 kPa ^-1 ) due to low compression of the dielectric layer, which affects arterial pulsation measurements, and they have not been used for long duration (>30 s) studies to reliably detect radial pulsation.

Previously reported capacitive sensors often use rigid substrates (i.e., polyethylene (PET)), which may limit the application of the sensor to detect radial artery pulsation. For example, the rigidity of the PET substrate may inhibit conformal contact to the human body. Also, the rigid substrate may delocalize stress over a larger area, which may attenuate the signal. This may adversely affect the spatially resolved mapping capability as well as reliable data acquisition. Another limitation is the durability of the electrodes used to develop conventional capacitive sensors. These electrodes are brittle and cannot withstand large bending/tensile/torsion strains. Other previously reported skin-like capacitive pressure sensors using elastic carbon nanotube (CNT) conductors supported on polydimethylsiloxane (PDMS) substrates have low pressure sensitivity, making it difficult to measure arterial pulsation pressure. Other types of capacitive pressure sensors incorporate PDMS micropyramid structures in the dielectric layer to significantly improve pressure sensitivity, or micro-hair-like PDMS structures to improve conformability to the skin surface and subsequently amplify the arterial pressure measurement. These sensors demonstrate that pressure applied by the sensor to the skin can produce slightly larger changes in the radial artery pressure waveform. However, these changes were minimal, perhaps suggesting that the dynamic range of the pressure sensor was not sufficient for applanation pressure measurements.

The object of the present invention is to provide a sensitive mechanical pressure sensor for measuring human physiological signals, such as pressure exerted by radial pulsation on the surface of the epidermis (skin), as specified in the independent claims, and a method for making said sensor. Furthermore, a highly accurate calibration of said sensor for continuous monitoring of pulsatile blood flow is described herein. Embodiments of the present invention are set out in the dependent claims, which may be freely combined with one another if they are not mutually exclusive.

Arterial applanation is a key component to achieving stable pulse pressure measurements by reducing the elastic resistance between the pressure sensor and the artery (i.e., the skin). To achieve arterial applanation and stable pulse pressure measurements, the pressure sensor must have high pressure sensitivity over a wide dynamic range. One approach to achieving a wide dynamic range in soft capacitive pressure sensors involves creating air gaps using micro-ridge structures.

In some aspects, the present invention features a soft capacitive pressure sensor incorporating a wrinkled metal thin film to develop a soft, stretchable electrode for radial pressure measurement applications. The capacitive pressure sensor can measure and monitor, for example, pulsatile blood flow. When the radial artery induces pressure on the sensor, the parallel plates of the sensor can move closer together. The wrinkled structure of the thin film creates robustness for the thin film to bend repeatedly. This allows for continuous arterial pressure measurement and capture of detailed pulse pressure waveforms with sufficient sensitivity over a wide dynamic range and fast response times of less than 10 ms.

The highly wrinkled thin film provides mechanical robustness against mechanical strain when supported against a soft elastomeric substrate. The softness of these electrodes allows the sensor to conform to the body and measure arterial pulsation pressure with high accuracy. To improve pressure sensitivity and dynamic range, micro-ridged structures are provided in the dielectric layer, improving pressure sensitivity by up to 5 times within a wide pressure range of 0-10 kPa. These micro-structures raise the opposing electrode and create air cavities in the dielectric layer, thereby increasing the amount by which the dielectric layer can be compressed. This in effect reduces the elastic resistance of the dielectric layer. Due to the presence of air cavities (ε _air =1), the effective dielectric constant also decreases. When the dielectric layer is compressed, the effective dielectric constant increases as the volume of the air cavities decreases, resulting in a larger change in capacitance. Continuous beat-to-beat blood pressure measurement using this soft capacitive pressure sensor is demonstrated by comparing it to an FDA-approved NIBP monitoring device.

In some examples, the capacitive pressure sensor may include a first electrode layer, a second electrode layer, a dielectric layer disposed against the first electrode layer such that the dielectric layer is between the first and second electrode layers, and one or more resilient ridges protruding from the second electrode layer toward the dielectric layer and the first electrode layer. The one or more resilient ridges create a gap separating the first and second electrode layers. When the sensor is in a rest configuration, the gap is disposed between the first and second electrode layers. The gap acts as a second dielectric layer. When the sensor is compressed, the first and second electrode layers move closer to each other, thereby decreasing the height of the gap and increasing the pressure sensitivity and capacitance of the sensor. In one example, the first electrode layer may include an elastomer layer and a conductive metal film disposed between the elastomer layer and the dielectric layer. In another embodiment, the second electrode layer may comprise an elastomer layer and a conductive metal film disposed on the elastomer layer. In a non-limiting example, the elastomer layer may include polydimethylsiloxane. The conductive metal film may include a wrinkled thin film of Au. The dielectric layer may include soft silicone rubber. In some embodiments, the pressure sensitivity of the sensor ranges from about 0.1 kPa ⁻¹ to about 0.2 kPa ⁻¹ in a pressure range between 0-10 kPa. The response time of the sensor is less than about 20 ms, preferably less than 10 ms.

In some embodiments, the conductive layer may be any thin film metal, including but not limited to Au. Other examples include thin films of copper, silver, or aluminum. In other embodiments, the electrode layer may include a thin film of carbon nanotubes, which is a wrinkled percolating network. In some embodiments, the elastomeric layer may include a silicon-based material or a non-silicon material, such as polyurethane. In other embodiments, the silicone dielectric material may be replaced with other dielectric materials, including but not limited to lead zirconate titanate, barium titanate, polyvinylidene fluoride, or oxides of zirconia, titania, or silica.

In some aspects, the present invention also provides a method for manufacturing a stretchable and flexible electrode that can be assembled to create a soft capacitance-based sensor for continuous blood pressure monitoring applications. According to some embodiments, a method for manufacturing a capacitive pressure sensor may include attaching a conductive metal film to a silicone elastomer layer to form a first electrode layer, attaching the dielectric layer to the first electrode layer such that the conductive metal film is disposed between the silicone elastomer layer and the dielectric layer, attaching a second conductive metal film to a second silicone elastomer layer to form a second electrode layer, the silicone elastomer layer comprising one or more resilient ridges protruding from a surface of the silicone elastomer layer, and attaching the first electrode layer to the second electrode layer such that the one or more resilient ridges are directed toward the dielectric layer and the first electrode layer. The resilient ridges are configured to create an air gap between the first electrode layer and the second electrode layer. For example, the elastic ridges push or separate the dielectric layer and the first electrode layer away from the second electrode layer. However, when the sensor is compressed, for example by an arterial pulsation, the first and second electrode layers move closer together, thereby decreasing the gap height and increasing the pressure sensitivity and capacitance of the sensor.

In some embodiments, the method may include molding the silicone elastomer layer to include one or more elastic ridges protruding from a surface of the silicone elastomer layer. In other embodiments, the method may further include calibrating the capacitive pressure sensor using an artificial neural network (ANN). The calibrating step may include generating a calibration model by training the ANN using a training set of measured blood pressure signals and one or more biological input variables. Such sensors can withstand tensile/bending/torsion strains, improving durability while also maintaining soft mechanical properties for conformal contact with the human body. The soft capacitive sensors described herein also have a pressure sensitivity of approximately 0.1 kPa ⁻¹ to 0.2 kPa ⁻¹ over a pressure range of 0 to 10 kPa, and a fast response time (<20 ms).

One of the unique and inventive technical features of the present invention is the ridges that support the opposing electrodes such that an air gap is formed between the two electrode layers. In addition, the present invention has the feature that the ridges and the dielectric layer are separate components. The dielectric layer consists of both the air gap and the dielectric material used. When both the air gap and the dielectric material are present, the dielectric constant is a combination of the two. The dielectric constant of air is equal to 1, whereas a silicone elastomer such as PDMS is about 3. The dielectric constant of the sensor is therefore somewhere in between, e, depending on the percentage of the volume occupied by the air and the silicone elastomer, respectively. When the pressure sensor is compressed, the volume of the air gap decreases and the dielectric constant increases, closer to 3. While not wishing to limit the present invention to any theory or mechanism, the increase in the dielectric constant by separating the ridges from the dielectric layer contributed to a higher pressure sensitivity. The inventive technical features of the present invention surprisingly resulted in the creation of a sensor that is sensitive over a wide pressure range and allows for accurate beat-to-beat measurement of blood pressure in an arterial pulse. None of the currently known prior references or studies possess the unique and inventive technical features of the present invention.

Moreover, prior work has taught away from the present invention. For example, previous sensors have utilized dielectric materials that are shaped to house an array of structures. However, the present invention's advantageous approach of increasing the dielectric constant and increasing pressure sensitivity cannot be implemented when the dielectric layer is microstructured.

In some embodiments, having two ridges is advantageous over an array of structures because two ridges are easy to fabricate, whereas fabricating an array of structures requires etching of a silicon wafer, a time-consuming process that does not always produce high fidelity. When an array of dielectric structures is compressed, the more it is compressed, e.g. the smaller the volume of the voids, the more mechanical resistance is created and the more material needs to be compressed. Compressing two ridges requires less mechanical force and effectively eliminates the need to compress the dielectric material, improving pressure sensitivity.

According to another embodiment, the present invention provides a method for monitoring a subject's blood pressure, the method may include providing a capacitive pressure sensor, attaching the sensor to the subject's forearm or wrist where the radial artery resides, measuring a blood pressure signal of the subject using the sensor, where the sensor detects a blood pressure signal corresponding to a radial artery pulsation when the radial artery pulsation causes compression of the sensor, and using a calibration model to determine an absolute blood pressure value from the measured signal. In one aspect, blood pressure can be monitored continuously. In a further embodiment, the method includes calibrating the capacitive pressure sensor using an artificial neural network (ANN). The calibrating step may include measuring a plurality of blood pressure signals from the subject, measuring one or more biological input variables from the subject, combining the measured blood pressure signals and the biological input variables to form a training set, and training the ANN using the training set to generate a calibration model.

Thus, the present invention allows for automatic calibration of arterial blood pressure for non-invasive, continuous beat-to-beat blood pressure monitoring. In one embodiment, the capacitive sensor can be calibrated to the FDA-approved ClearSight® device. Alternatively, a neural network was developed to input various parameters to output mmHg without the ClearSight® device. In a non-limiting embodiment, the neural network can be trained using various inputs that make up a training set, such as capacitive measurements from the sensor, pressure applied by the mechanical sensor, pulse rate, EKG, accelerometer data, gyroscope data, magnetometer data, hemodynamic monitoring data, and ClearSight® data, to create a model. A training set including blood pressure signal measurements and one or more biological input variables can be obtained from multiple subjects. After this robust model is created, the input of capacitance measurements from the sensor outputs absolute pressure. This technique can also be used to subtract motion artifacts when generating a training set by moving subjects for the model in a predefined and known manner. When a calibrated sensor is used on the same person, the pulsatile waveform measured by the sensor can be correlated with arterial blood pressure. Thus, highly accurate blood pressure measurement readings can be obtained without any other input. For example, once the sensor is calibrated, no additional equipment is required by the end user. The user simply places the pressure sensor on the radial artery and absolute mmHg will be recorded.

Another unique and inventive technical feature of the present invention is the use of a neural network to calibrate a continuous mechanical blood pressure sensor. Without wishing to limit the present invention to any theory or mechanism, it is believed that this technical feature of the present invention advantageously provides non-invasive beat-to-beat monitoring of blood pressure. None of the currently known prior references possess the unique and inventive technical feature of the present invention.

Any feature or combination of features described herein is included within the scope of the present invention, unless the features included in any such combination are mutually inconsistent as would be apparent from the context, this specification, and the knowledge of one of ordinary skill in the art. Further advantages and aspects of the present invention will become apparent in the following detailed description and claims.

The features and advantages of the present invention will become apparent from a consideration of the following detailed description, taken in conjunction with the accompanying drawings.

A cross-sectional view of a capacitive pressure sensor of the present invention, in which a dielectric layer is disposed between two wrinkled Au thin film (wAu) layers, with ridges along the electrode trace edges forming an air gap. Figure 1 shows a cross-sectional view of a capacitive pressure sensor of the present invention. A dielectric layer is disposed between two wrinkled Au thin film (wAu) layers. When compressed, the area of the void decreases, thereby changing the effective dielectric constant. The distance between the electrodes also decreases, thereby increasing the capacitance. FIG. 1 shows a non-limiting example of a manufacturing process for making a flexible capacitive pressure sensor. In step a), a laser cutter is used to etch the shadow mask and the PS substrate. In step b), Au is sputtered onto the PS substrate through the shadow mask. In step c), the shadow mask is removed, and in step d), the PS substrate is heated to 140° C. to promote biaxial shrinkage. In step e), the wAu is treated with MPTMS, and then a silicone elastomer is spin-coated onto the substrate, thus defining the etched grooves. In step f), the substrate is placed in an organic solvent to lift the wAu off the PS. In step g), the electrical interconnects are attached, and then the electrode and dielectric layers are plasma-bonded together, and in step h), the final device is fabricated. Schematic of a pressure sensor when placed on the wrist, over the radial artery. As blood pulsates through the radial artery, the pressure sensor deforms. A screw is used to apply incremental pressure to applanate the radial artery. Schematic of a pressure sensor when placed on the wrist, over the radial artery. As blood pulsates through the radial artery, the pressure sensor deforms. A screw is used to apply incremental pressure to applanate the radial artery. 1 is a scanning electron microscope (SEM) image of an example capacitive pressure sensor and wAu. 1 is a SEM image of a cross section of a micro-ridge structure. 1 is a SEM image of a gap that exists between electrodes. FIG. 1 shows the height profile of a PDMS elastic ridge formed during a PDMS molding process, where the aspect ratio of the ridge increases during the shrinkage process. FIG. 14 shows pressure sensitivity curves from 0 to 10 kPa for a capacitive pressure sensor with ridges (red), a capacitive pressure sensor without ridges (blue), and a sensor made with flat Au electrodes (green). 1 shows the pressure sensitivity curves of a pressure sensor with ridges versus without ridges, where the black line shows the loaded condition and the colored line shows the unloaded condition. The amount of load applied before unloading is shown in Newtons by the colored line. FIG. 7B is an inset showing the electromechanical response of the pressure sensor after loading with a force of 1 N (about 40 kPa) and after unloading. 1 is a graph of cyclic loading of 25 kPa over 5000 cycles. 1 shows pressure sensitivity (PS) curves of soft capacitive pressure sensors. PS curves of seven different capacitive sensors with micro-ridge structures (height 85 μm, width 100 μm) are shown. 10A shows the pressure sensitivity (PS) curves of a soft capacitive pressure sensor, from 0 to 10 kPa, inset of the PS curves of FIG. 1 shows pressure sensitivity (PS) curves for soft capacitive pressure sensors. PS curves for three different sensors with larger micro-ridges (height 190 μm, width 600 μm) are shown. 10C shows the pressure sensitivity (PS) curves of a soft capacitive pressure sensor, from 0 to 10 kPa, inset of the PS curves of FIG. 1 shows pressure sensitivity (PS) curves of soft capacitive pressure sensors, showing the curvature PS of three different sensors without micro-ridge structures. 10B shows the pressure sensitivity (PS) curves of a soft capacitive pressure sensor, from 0 to 10 kPa, inset of the PS curves of FIG. 10E. Figure 1 shows the signal response after inducing a strain shock on the pressure sensor. A small probe with a diameter of about 2 mm was attached to an Arduino-controlled linear actuator to induce the strain shock (<1 ms). Figure 1 shows the results when a small probe, approximately 2 mm in diameter, is attached to an Arduino-controlled linear actuator. The experiment started with a pre-strain on the sensor. Distance was measured as the probe was moved away from the sensor as the distance increased (shorter distances result in larger changes in capacitance). Figure 2 shows the signal response from the pressure sensor when a periodic strain is induced. Figure 1 shows the results when a small probe, approximately 2 mm in diameter, is attached to an Arduino-controlled linear actuator. The experiment started with a pre-applied strain on the sensor. Distance was measured as the probe was moved away from the sensor as the distance increased (shorter distances result in larger changes in capacitance). Figure 2 shows the Fast Fourier transform (FFT) of the applied cyclic strain and pressure sensor signal response measured at approximately 11 Hz. FIG. 1 is a diagram of the setup showing how a pressure sensor is attached to the wrist to measure radial artery pulsation. FIG. 13B shows the pressure sensor and wrist mounted components of the setup of FIG. 13A. 1 is a photographic image of where the sensor was placed to measure beat-to-beat blood pressure. FIG. 1 shows an example of an arterial pulsatile waveform measured with a capacitive pressure sensor (top) and a ClearSight® device (bottom). FIG. 15B shows a single beat waveform from the highlighted section of FIG. 15A, illustrating cardiovascular features. FIG. 1 shows an example of four 70-beat intervals from subject 1 used to compare capacitive pressure sensors with ClearSight®. The arterial pulse waveform is shown in black, and the SBP and DBP are highlighted in red. FIG. 13 shows linear regression analysis of SBP, DBP, and MAP between pressure sensor and ClearSight®. FIG. 13 shows an example of a pressure sensor calibration model from subject 1 for SBP. FIG. 13 shows an example of a calibration model of the pressure sensor from subject 1 for DBP. FIG. 13 shows an example of a calibration model of the pressure sensor from subject 1 for MAP. Figure 1 shows the combined Bland-Altman plot for all subjects. The data includes the different sensors used on subject 1 for a total of 9 separate tests. The dashed lines show the two standard deviations and the solid line shows the mean bias. FIG. 1 shows a method for measuring pulse transit time (PTT) using ECG (mV) and a pressure sensor (pF) placed in the radial artery. FIG. 10 shows that PTT ⁻¹ (solid circle line) measured using ECG has a similar trend in blood pressure fluctuations measured with a pressure sensor. FIG. 13 is a diagram showing the correlation of PTT ^-1 with a pressure sensor. FIG. 1 shows the correlation of PTT ^-1 with ClearSight®. Figure 1 shows 60 seconds of pressure sensor data measured at various times and then calibrated using the PTT model generated by ClearSight®. Bland-Altman analysis of the systolic peak shows mean error and standard deviation within the ISO specifications for FDA approval. From left to right, a plasma-bonded capacitive pressure sensor, individual electrodes, and a 4x4 sensing grid of capacitive pressure sensors. FIG. 13 shows pulsatile measurement using two sensing areas. 10 demonstrates the ability of the sensor to spatially resolve pressure at different locations on the same sensor. Data acquisition was performed using a Texas Instruments FDC2214EVM.

"Explanation of symbols" is a list of elements that correspond to specific elements referenced in this specification.

1A-1B, in some embodiments, the present invention features a capacitive pressure sensor (100). The sensor (100) may include a first electrode layer (110), a second electrode layer (120), and a dielectric layer (130) juxtaposed between the first electrode layer (110) and the second electrode layer (120). When the sensor (100) is in a resting configuration as shown in FIG. 1A, a gap (140) is disposed between the first electrode layer (110) and the second electrode layer (120). The gap (140) may act as a second dielectric layer. In a preferred embodiment, the sensor (100) is bendable, compressible, and flexible. As shown in FIG. 1B, when the sensor (100) is compressed, the height of the gap decreases, which brings the first electrode layer (110) and the second electrode layer (120) closer together, increasing the pressure sensitivity and capacitance of the sensor. Without wishing to limit the invention to any theory or mechanism, it is believed that the present invention advantageously provides a pressure sensor with high electromechanical reliability, high pressure sensitivity, fast response time, and low energy consumption. None of the currently known prior references or studies have the unique and inventive technical features of the present invention.

In some embodiments, the first electrode layer (110) may comprise an elastomer layer (112) and a conductive film (114) disposed on the elastomer layer (112). A dielectric layer (130) may be disposed on the conductive film (114) of the first electrode layer (110). In other embodiments, the second electrode layer (120) may comprise an elastomer layer (122) and a conductive film (124) disposed on the silicone elastomer layer (122). In further embodiments, the second electrode layer (120) may comprise a resilient ridge (125) protruding from the second electrode layer (120) towards the first electrode layer (110), the ridge creating an air gap separating the first electrode layer (110) and the second electrode layer (120) in a rest configuration. In one embodiment, the conductive film (124) may be disposed between resilient ridges (125). The resilient ridges (125) may preferably bend to reduce the height of the gap when the sensor (100) is compressed. As used herein, the height of the gap may refer to the distance between the dielectric layer (130) and the conductive film (124) of the second electrode layer (120).

In one embodiment, the elastic ridge (125) may comprise two ridges. The ridges may be parallel to each other on the second electrode layer (120). In other embodiments, the elastic ridge (125) may comprise three or more ridges. In some embodiments, the elastic ridge may be an elongated strip having a triangular profile or cross-section, as shown in the figures. The pointed distal end of the triangular ridge may contact the dielectric layer of the opposing electrode layer. In other embodiments, the elastic ridge may have an upside-down "U" profile. In some other embodiments, the elastic ridge may be a pyramidal structure, e.g., two pyramidal structures.

In some embodiments, the elastomeric layers (112, 122) may include a silicone, such as polydimethylsiloxane. In other embodiments, the elastomeric layers (112, 122) may include a non-silicone elastomer, such as polyurethane.

In some other embodiments, the conductive film (114, 124) is a metal film. The metal film may be wrinkled, such as a wrinkled Au thin film. Other conductive materials that may be used in accordance with the present invention include, but are not limited to, metals such as copper, silver, or aluminum thin films. Alternatively, a percolating network of conductive nanomaterials may be used as the conductive electrodes, such as carbon nanotubes.

In yet another embodiment, the dielectric layer (130) may include a soft silicone rubber such as Ecoflex. Thus, when the sensor is in a resting configuration, the dielectric constant of the sensor is about 1. When the sensor is compressed, the dielectric constant may be about 3. In some other embodiments, the dielectric layer (130) may include a dielectric material having a higher dielectric constant. For example, the dielectric constant of the dielectric material may be greater than 3. Examples of dielectrics having a higher dielectric constant include, but are not limited to, oxides such as zirconia, titania, or silica, and piezoelectric materials such as lead zirconate titanate (PZT), barium titanate (BaTiO3), and polyvinylidene fluoride (PVDF). While not wishing to limit the invention to a particular theory or mechanism, the dielectric material is selected such that the dielectric constant of the material is much greater than 1, which is the dielectric constant of air. This may allow for a larger change in the dielectric constant when the sensor is compressed, thus improving pressure sensitivity.

In one embodiment, the pressure sensitivity of the sensor may be from about 0.1 kPa ⁻¹ to about 0.2 kPa ⁻¹ over a pressure range of 0 to 10 kPa. In another embodiment, the response time of the sensor is less than about 20 ms, preferably less than 10 ms.

According to some embodiments, the invention features a method of monitoring a subject's blood pressure. The method may include providing any one of the capacitive pressure sensors (100) described herein, attaching the sensor to the subject's forearm or wrist where the radial artery resides, and operably connecting the sensor to a pressure gauge. Without wishing to limit the invention to any particular theory or mechanism, pulsation of the radial artery causes compression of the sensor, which detects a signal corresponding to the pulsation of the radial artery and transmits the signal to the pressure gauge. The pressure gauge can then determine, e.g., calculate, the blood pressure from the signal.

In yet another embodiment, a method for monitoring a subject's blood pressure may include attaching a capacitive pressure sensor (100) to the subject's forearm or wrist where the radial artery is located; measuring a blood pressure signal of the subject using the sensor (100), where pulsation of the radial artery causes compression of the sensor and the sensor (100) detects a blood pressure signal corresponding to the pulsation of the radial artery; and determining an absolute blood pressure value from the measured signal using a calibration model. In a further embodiment, the monitoring method may further include calibrating the capacitive pressure sensor (100) using an artificial neural network (ANN). The calibrating step may include measuring a plurality of blood pressure signals from the subject, measuring one or more biological input variables from the subject, combining the measured blood pressure signals and the biological input variables to form a training set, and training the ANN using the training set to generate a calibration model. The calibration model is used to calibrate the sensor (100) and allows for determining an absolute blood pressure value from the measured signal. In other examples, the calibrating step may further include subtracting motion artifacts by including motion data in the training set. In some examples, the training set may include blood pressure signal measurements from multiple subjects and one or more biological input variables. In other examples, the one or more biological input variables may include sensor pressure, pulse rate, EKG data, accelerometer data, gyroscope data, magnetometer data, or hemodynamic monitoring data.

In one example, the calibration model can utilize additional physiological parameters, including pulse transit time (PTT). PTT is the time it takes for a pulse of blood to travel from one point to another. This can be measured using an electrocardiogram (ECG) and a proximal sensor, such as a pressure sensor on the radial artery. These two components can detect the start of a pulse and the time it takes for that pulse to reach the radial artery. PTT can correlate with blood pressure, since higher pressure allows the pulse to be sent through the body faster. PTT can help calibrate a capacitive pressure sensor to a reference value. Thus, the PTT and capacitive sensor information can be used in combination to generate an ANN for calibrating the pressure sensor.

In one embodiment, the sensor can be used to continuously monitor the subject's blood pressure, for example beat-to-beat. In another embodiment, the sensor can also be used to perform a single blood pressure measurement. The sensor can preferably be used to monitor blood pressure non-invasively. In some embodiments, the sensor (100) can be attached to the forearm or wrist by adhesive tape or a cuff, such as an adjustable or elastic band.

According to another embodiment, the invention features a method of manufacturing a capacitive pressure sensor (100). The method may include attaching a conductive metal film (114) to an elastomeric layer (112) to form a first electrode layer (110), attaching a dielectric layer (130) to the first electrode layer (110) such that the conductive metal film (114) is disposed between the elastomeric layer (112) and the dielectric layer (130), attaching a second conductive metal film (124) to a second elastomeric layer (122) having a resilient ridge (120) protruding from a surface of the second elastomeric layer to form a second electrode layer (120), and attaching the first electrode layer (110) to the second electrode layer (120). The elastic ridges (120) are preferably oriented toward the first electrode layer (110), and the dielectric layer (130) is juxtaposed between the first electrode layer (110) and the second electrode layer (120). The elastic ridges (125) thus create an air gap (140) between the first electrode layer (110) and the second electrode layer (120). When the sensor (100) is compressed, the height of the air gap decreases, thereby bringing the first electrode layer (110) and the second electrode layer (120) closer together, increasing the pressure sensitivity and capacitance of the sensor.

In an alternative embodiment, the present invention may have the ability to sense different areas by using a grid of such sensors. This is important to be able to spatially resolve the optimal location of the radial artery. This may be achieved by introducing additional electrodes into the manufacturing process. It should be understood that the present invention is not limited to a 4x4 grid sensor. In other embodiments, the amount of sensing "pixels" may be any size, e.g., 1x1, 2x2, 3x3, 5x5, 10x10, 100x100, etc. Alternatively or in combination, multiple sensors may be used together to enable mapping capabilities (spatial resolution). For example, 2-5 sensors may be operatively coupled to each other to sense different areas.

According to some other embodiments, the invention features a method for calibrating a mechanical continuous blood pressure monitor using an artificial neural network (ANN). By way of non-limiting example, the method may include providing a mechanical continuous blood pressure monitor, using the blood pressure monitor on a patient, measuring a plurality of mechanical blood pressure signals of the patient, measuring one or more additional biological input variables of the patient, combining the measured mechanical blood pressure signals and the additional biological input variables to form a training set, training the ANN using the training set to generate a calibration model, and calibrating the blood pressure monitor using the calibration model such that the blood pressure monitor can be used to continuously measure the patient's blood pressure.

In some embodiments, the blood pressure monitor may include any of the capacitive pressure sensors described herein. By way of non-limiting example, the capacitive pressure sensor may include a first electrode layer, a second electrode layer, and a dielectric layer juxtaposed between the first and second electrode layers. In a preferred embodiment, when the sensor is in a resting configuration, an air gap is disposed between the first and second electrode layers, and when the sensor is compressed, the height of the air gap decreases, thereby bringing the first and second electrode layers closer together and increasing the pressure sensitivity and capacitance of the sensor.

In some embodiments, the blood pressure monitor may be used on an artery of the subject, such as the radial artery or another artery. In some embodiments, the blood pressure monitor may be calibrated to measure beat-to-beat blood pressure. In some embodiments, the training set may include measurements from multiple patients. By way of non-limiting example, the training set may include measurements from tens, hundreds, thousands, tens of thousands, or more patients. According to one embodiment, the calibrated blood pressure monitor may be used on a new patient with high accuracy without the need for recalibration.

In some embodiments, the mechanical blood pressure signal may include a pulsatile waveform. In some embodiments, the additional biological input variables include sensor pressure, pulse rate, EKG data, accelerometer data, gyroscope data, magnetometer data, or hemodynamic monitoring data. In some embodiments, the method may further include subtracting motion artifacts. As a non-limiting example, the motion artifacts may be subtracted by including motion data in the training set. As a non-limiting example, the motion data may be included in the training set by moving the patient in a predetermined manner.

In one embodiment, the invention features a calibrated, mechanical, continuous blood pressure monitor. By way of non-limiting example, the monitor may include a capacitive pressure sensor configured to enable continuous monitoring of beat-to-beat blood pressure, and a calibration model generated by an artificial neural network (ANN) using multiple measurements from the capacitive pressure sensor. In some embodiments, the calibrated blood pressure monitor may be configured to provide high accuracy blood pressure measurements without additional data input.

In some embodiments, the multiple measurements may be from a single patient or from multiple patients. In some embodiments, the capacitive pressure sensor may be disposable or reusable. In some embodiments, the sensor may be configured to be attached to a skin surface of the patient. As a non-limiting example, the sensor may be attached above an artery of the patient. In a preferred embodiment, the monitor may include a single sensor. Alternatively, multiple sensors may be used in combination.

In one embodiment, the invention features a method for continuously monitoring beat-to-beat blood pressure of a subject in need of blood pressure monitoring. By way of non-limiting example, the method may include providing a calibrated mechanical continuous blood pressure monitor, attaching the monitor to the subject's forearm or wrist where the subject's radial artery resides, measuring the subject's mechanical blood pressure signal using the monitor, and interpreting the measured signal using a calibration model to determine an absolute blood pressure value. In some embodiments, the monitor may include a capacitive pressure sensor. In some embodiments, the method may be non-invasive. In some embodiments, calibration of the monitor may be performed using an artificial neural network (ANN). In some embodiments, the monitor is attached to the subject's skin surface by adhesive tape or a cuff.

In other embodiments, the pressure sensor of the present invention can be used on other parts of the body and is not limited to use on the forearm or wrist. For example, the sensor can be attached to the neck, upper arm, chest, leg, etc.

In yet other embodiments, the present invention may be implemented in other applications requiring pressure measurements, flow rates, etc. As a non-limiting example, pressure sensors may be used in structures such as pipes.
Real-life examples

The following are non-limiting examples of the present invention. It should be understood that the examples are not intended to limit the present invention in any manner. Equivalents or alternatives are within the scope of the present invention.

experiment

Manufacturing of capacitance-based pressure sensors

The sensor was constructed with four layers: a wrinkled electrode, a dielectric layer, an air gap, and a wrinkled electrode with etched ridges. With reference to FIG. 2, the process for fabricating both electrodes began with rinsing a pre-stressed polystyrene (PS) substrate (Grafix Shrink Film KSF50-C, Grafix Arts, OH) with 70% ethanol. The sensor design was developed using AutoCAD (Autodesk, Inc., CA) and patterned using a laser cutter (Universal Laser, AZ) onto a shadow mask made of an adhesive polymer film (Grafix Friskeet Film, Grafix Arts, OH). The design was created such that the final electrode dimensions after shrinkage were 2 mm wide and 15 mm long. The Friskette film was first placed on top of the PS substrate, and then the sensor design was patterned onto the masked PS substrate. For the ridge-etched electrodes, higher power settings (0.5% power, 3% speed, 1000 PPI) were used to etch into the PS substrate along the sides of the electrodes. After both substrates were masked and patterned, 15 nm of Au was deposited onto each PS substrate using a magnetron sputter coater (Q150R, Quorum Technologies, UK).

The shadow mask was then removed and the sputtered substrate was placed in a convection oven at 140 °C to induce biaxial shrinkage and wrinkling of the Au film. The shrunken sample was treated with 5 mM 3-mercaptopropyltrimethoxysilane (95% MPTMS, Sigma Aldrich) in pure ethanol for 1 h. The sample was then rinsed with ethanol and dried with an air gun. Polydimethylsiloxane (PDMS) (Sylgard 184 Silicon Elastomer Base, Dow Corning, MI) was then poured onto the MPTMS-treated Au film and spin-coated at 300 rpm for 30 s to give a final substrate thickness of approximately 0.5 mm. The sample was placed in vacuum for 30 min to remove air bubbles and cured overnight in a convection oven at 60 °C. The cured sample was then placed in a 75° acetone bath for 15 minutes to dissolve the PS substrate and lift the PDMS and wrinkled Au film off the PS. Residual PS was then removed from the wrinkled thin film by immersing the sensor in toluene for 2 minutes and rinsing with acetone. The sample was then air-dried overnight.

Colloidal silver liquid (Pelco Colloidal Silver Liquid, Ted Pella, CA) was used to interconnect the Au electrodes with the wires, and after drying, resin was used to seal the interconnected contacts. After the resin dried, a polyimide tape was applied over the resin to further secure the interconnects with the electrodes. Ecoflex® can only be plasma bonded to other silicones if it is cured on the PDMS substrate or chemically modified with other silicone elastomers. Therefore, silicone elastomer Ecoflex® 0030 (Smooth-On, PA) was first spin-coated onto the cured PDMS layer. The application step was performed at 1000 rpm for 10 seconds and the spin-coating step was performed at 3000 RPM for 30 seconds, resulting in a final thickness of 15 μm. The elastomer was then placed in a 60°C convection oven for 2 hours to cure. The cured elastomer was then plasma bonded (PE-50, Plasma Etch, NV) to a flat electrode at 120 mTorr in open air for 40 seconds and then placed in a 60°C convection oven to promote chemical bonding. After removing the PDMS layer from the Ecoflex® layer, a flat electrode with a dielectric layer was then plasma bonded to the etched electrode to form the final capacitance-based sensor. The sensor was placed in a 60°C convection oven to promote chemical bonding.

Flat Au electrodes (which served as the control) were fabricated by patterning 90 nm of Au on a PS substrate. The Au was treated with 5 mM MPTMS in pure ethanol (95% MPTMS) for 1 hour. PDMS was then spin-coated onto the substrate at 300 RPM for 30 seconds and cured in a convection oven at 60° C. for 2 hours. A drop of acetone was placed on the PDMS to lift the Au off the PS substrate. After attaching the electrical interconnects, Ecoflex® 0030 was spin-coated onto one electrode at 3000 RPM for 30 seconds and then cured in a convection oven at 60° C. for 2 hours. After curing, the opposing electrode was placed on top of the electrode with an Ecoflex® dielectric layer. The surface area overlap of the electrodes was ^2x2 mm2.

Characteristics evaluation

A scanning electron microscope (SEM) (FEI Magellan 400XHR) was used to characterize the wrinkle structure of the Au film. Pressure sensitivity was tested using a force gauge (Force Gauge Series 5, Mark-10, NY) connected to a test stand (ESM303, Mark-10, NY). The force gauge was placed slightly above the sensor and then moved at a descent rate of 1.1 mm per minute using a 6 mm diameter force probe. A glass fiber probe was used to reduce fringing effects. The capacitance change was collected during the application of force and measured using an LCR meter (300 kHz) (E4980AL Precision LCR Meter, Keysight, CA). Data was collected using LabView and then processed using Matlab. An impedance analyzer (1 MHz, 500 mV) (4291B, Agilent, CA) was used to measure the signal response of the pressure sensor due to repeated static loads applied using a custom-built linear actuator controlled by an Arduino. Distance information from the custom-built linear actuator was recorded with a linear potentiometer (Spectra Symbol, UT) using a National Instrument data acquisition system (USB-6003, TX).

Beat-by-beat blood pressure method

The sensor was assembled on a custom Velcro® band using a screw to apply pressure gradually. Prior to attaching the sensor to the body, a Tegaderm® (3M Health Care, MN) strip was placed on the left wrist to promote affinity of the sensor to the skin. The sensor was then attached and tightly bound with a custom Velcro® band. A ClearSight® finger cuff was attached to the index finger of the subject's right hand. Measurements of the pressure sensor and the ClearSight® device were measured simultaneously and subsequently analyzed in Matlab.

Statistical analysis

Sensor capacitance readings were taken from the LCR meter at a sampling frequency of approximately 56 Hz (data points were time-stamped with millisecond accuracy). All data sets were post-processed in Matlab and linearly interpolated to 200 Hz to match the ClearSight® (200 Hz) data set. The interpolated data was then smoothed with a moving average filter of 5 data points.

As mentioned above, ClearSight® segments beat-by-beat blood pressure measurements by the number of cardiac cycles detected. Intervals of 70 beats were considered the most accurate and precise for blood pressure measurements, although each of these intervals contained only 69 complete cardiac cycles that were used for analysis. Different breathing maneuvers were performed in four consecutive intervals. Linear regressions were obtained between the pressure sensor and ClearSight® for SBP, DBP, and MAP (n=69) for each subject and individual interval. The four intervals per subject were then combined and analyzed by linear regression (n=276).

A Bland-Altman analysis was used to investigate the agreement between the ClearSight® and the pressure sensor. Data for each subject were randomly split into two separate sets, with 75% of the data used to create the model and 25% used to test the model. A linear regression model was constructed to predict SBP, DBP, and MAP values, respectively (n=207). The remaining data for each subject (n=69) were then calibrated with the linear regression model. The mean error and standard deviation were then calculated for all nine subjects' tests combined for each SBP, DBP, and MAP parameter (n=621). Three consecutive cardiac cycles were averaged and analyzed similarly (n=207). The mean error and standard deviation were calculated for all SBP, DBP, and MAP parameters combined without averaging of cardiac cycles (n=1863) and also with averaging of three consecutive cardiac cycles (n=621).

Results

How sensors are manufactured

A soft capacitive pressure sensor consisting of two conductive parallel plates separated by a dielectric layer was fabricated using a thermally induced shrink fabrication process (Fig. 2). When shrunk on a polystyrene (PS) substrate, highly wrinkled structures of Au thin films (wAu) were formed. When these wrinkled thin film structures were transferred onto a silicone elastomer substrate, the wAu thin films demonstrated stretchability up to 200% before electrical failure. Because wAu electrodes are mechanically robust, they were used as the parallel plates (2x2 ^mm2 ) of a capacitive pressure sensor. These wrinkled structures significantly improve the mechanical robustness and allow the integration of 15 nm thick wAu thin films onto soft substrates to withstand thousands of cycles. These electrodes can be supported on soft silicone substrates, including polydimethylsiloxane (PDMS), which is especially important for measuring localized pressure. Many reported capacitive pressure sensors use rigid substrates such as polyethylene terephthalate (PET, E~2.5 GPa), which are not as mechanically compatible with the human body as soft silicone elastomers such as PDMS (E~1 MPa), and rigid substrates can potentially hinder spatial resolution of local stresses and negatively affect arterial pulsation measurements.

A soft silicone elastomer, Ecoflex® (15 μm), was used as the soft dielectric layer between the parallel plates. Ecoflex® was used instead of PDMS (Shore hardness A-48) due to its soft mechanical properties (Shore hardness 00-30) to reduce the elastic resistance between the parallel plate electrodes and improve pressure sensitivity. In addition, microstructured ridges were created in PDMS by imitating the laser cut etched grooves in the PS substrate. The microstructured ridges were about 85 μm high and about 100 μm wide, and were located adjacent to the edges of the wrinkled thin film electrodes and spaced about 2 mm apart from each other, as can be seen in Figure 5A. The size of the gap between the electrodes was about 130 μm high, as can be seen in Figure 5B. The ridges supported the opposing electrodes and created air cavities in the dielectric layer.

When pressure was applied, the parallel plates moved closer together and the capacitance increased. Compression of the air cavity also increased the effective dielectric constant, closer to that of ε _eco , resulting in a larger change in capacitance. This effect can be explained by Equation 2:

Here, _εr0 is the relative dielectric constant before pressure is applied, and _εr is the relative dielectric constant after pressure is applied.

Electromechanical property evaluation

The electromechanical performance of the soft capacitive pressure sensor was evaluated by measuring the pressure sensitivity, which is defined in Equation 3.

where ΔC is the change in capacitance, _C0 is the initial capacitance, and P is the pressure. The pressure sensitivity of the sensor was measured by applying a force and measuring the change in capacitance. The force was applied to the tip of a 6 mm glass fiber probe attached to a force gauge. The glass fiber probe was used to reduce any fringe capacitance interference with the capacitive pressure sensor.

The measured pressure sensitivity of the sensor with micro-ridges was 0.148 kPa ^-1 between 0 and 10 kPa (Figure 7A). A control group of wrinkled electrodes without micro-ridges was also tested. The capacitive pressure sensor without micro-ridges exhibited a lower pressure sensitivity of 0.029 kPa ^-1 between 3 and 10 kPa. However, the pressure sensitivity of the pressure sensor without ridges was about 0.12 kPa ^-1 in the range of 0 to 2 kPa, which is comparable to the pressure sensor with micro-ridges. In this case, there are no micro-ridges to create air cavities, but small voids still exist due to the unevenness of the wrinkled Au electrode. However, these voids are fully compressed in the low pressure region, so applying more pressure results in a lower pressure sensitivity. Therefore, the microstructured ridges enable the soft capacitive pressure sensor to achieve high pressure sensitivity over a wide dynamic range. However, as the size of the microridges increased (height 190 μm, width 600 μm), the pressure sensitivity decreased (FIG. 8). This decrease in pressure sensitivity may be due to the larger features that need to be compressed.

A capacitive pressure sensor with flat Au electrodes was fabricated for comparison with wAu electrodes. 90 nm of Au was used instead of 15 nm because the brittleness of the Au thin film on the PDMS substrate made it impossible to transfer 15 nm. The capacitive pressure sensor with flat Au electrodes (with little or no voids) showed a pressure sensitivity of about 0 kPa ^-1 from 0 to 10 kPa. This suggests that the wAu electrodes provide the mechanical robustness to allow the attachment of films as thin as 15 nm, and also show a significantly higher pressure sensitivity compared to flat Au electrodes.

A larger pressure range was also investigated and compared between sensors with and without microstructured ridges. A mechanical load was first applied at 1 N (Figure 8) and then removed, showing signs of hysteresis. This was then repeated with increasing loads from 1 to 20 N, as can be seen in Figure 7B. The electromechanical response in both conditions shows characteristics of the Mullins effect, indicating that the stress-strain curve varies with the magnitude of the previously applied stress. As can be seen in Figure 7B, the electromechanical response of the pressure sensor with microridges showed higher pressure sensitivity over a wider dynamic range compared to the pressure sensor without microridges.

In addition, the electromechanical response to cyclic loading was analyzed as shown in Figure 9. The pressure sensor was subjected to approximately 25 kPa for 5,000 cycles. The pressure sensor demonstrated mechanical robustness to the stresses of multiple cyclic loadings. This demonstrates the durability to cyclic mechanical loading required for continuous arterial pressure measurement.

It is clear that the incorporation of micro-ridges improved the pressure sensitivity by up to 5 times within a wide dynamic range between 0 and 10 kPa. In addition, the mechanical robustness of the wrinkled structure allows the deposition of significantly thinner electrodes, which are easier and faster to fabricate. The reproducibility of such sensors is shown in Figures 10A-10F. In addition, the response and relaxation times of the pressure sensors were measured. A probe with a diameter of about 2 mm was attached to an Arduino-controlled linear actuator. The pressure sensor showed a response time (<10 ms) and relaxation time (<17 ms) when subjected to a strain impulse (<1 ms) (Figure 11). The pressure sensor was also able to measure periodic strains up to 10 Hz (Figures 12A-12B). The sampling measurement rate used for the dynamic mechanical tests was about 130 Hz (4291B, Agilent, CA).

Beat-to-beat blood pressure monitoring

Experimental setup for NIBP

As previously mentioned, arterial tonometry is a method of quantifying arterial pressure using a pressure sensor. Additionally, soft capacitive pressure sensors can be used to monitor arterial pressure. Highly accurate and precise measurements of radial artery pressure can then be converted to beat-to-beat blood pressure for NIBP wearable applications.

To demonstrate beat-to-beat blood pressure monitoring, sensors were used on healthy subjects under approval from the University of California Institutional Review Board (IRB number 2016-2924). One soft capacitive pressure was tested on a total of seven subjects to demonstrate robustness. Two additional soft capacitive sensors were tested on subject 1 to demonstrate repeatability. The pressure sensor was attached to the wrist above the radial artery. The subject was then instructed to maintain a slightly overextended hand with the palm facing up to help expose the radial artery at the skin surface. During this measurement, the subject was seated with the pressure sensor close to heart level. No allergic reactions or pain were reported by the subjects tested.

For arterial pulsation measurements, the pressure sensor was mounted on an acrylic backing plate using a Velcro® strap. A screw was attached to the acrylic backing plate so that the acrylic backing plate could apply pressure increments to applanate the radial artery. The pressure increments increased the baseline capacitance of the capacitive pressure sensor. A schematic of the pressure sensor device can be seen in Figures 3A-3B. To improve the contact between the pressure sensor and the human skin, medical tape was also attached to the wrist. Finally, a PDMS spacer (250 μm) was also used between the pressure sensor and the epidermis to further compress the tissue and amplify the radial artery pulsation. When blood pressure increases in the radial artery, the radial artery expands, deforming the surrounding tissue, which in turn deforms the pressure sensor. This pressure can be related to the arterial blood pressure as long as the pressure sensor is kept in stable contact with the body.

To evaluate the ability of the capacitive pressure sensor to measure beat-by-beat blood pressure, the pressure sensor was compared to the ClearSight® (Edwards Lifesciences, Irvine, Calif.), an FDA-approved finger volume clamping device. The ClearSight® was attached to the index finger of the subject's right hand. A photographic image showing where the device was attached can be seen in FIG. 14. Measurements were taken simultaneously measuring the pressure applied from the radial artery with the pressure sensor and measuring the brachial artery pressure with the ClearSight®. An example of the measured radial artery pulsatile waveforms with the pressure sensor and ClearSight® is presented in FIG. 16A. As seen in FIG. 16B, the fast response time and pressure sensitivity allowed for detection of unique features of the radial artery pulsatile waveform, including the late systolic peak that is not easily discernible in the ClearSight® signal.

The parameters investigated included systolic (SBP), diastolic (DBP), and mean arterial pressure (MAP), which are the most common parameters when assessing a person's cardiovascular health. SBP is the blood pressure against the arterial wall when the heart contracts, DBP is the blood pressure against the arterial wall when the heart relaxes, and MAP is the average pressure throughout one cardiac cycle and can be calculated using Equation 4:

where PP is the pulse pressure, which is equal to SBP minus DBP.

Arterial pulsations can be palpated at many different locations in the human body. These arterial pulsations are of different waveforms as they propagate to different areas of the cardiovascular tree. As blood is pumped from the heart to the peripheral arteries, the narrowing of the blood vessels amplifies the arterial pulsation waveform. By measuring these arterial pressures, the information can then be used proxies to estimate arterial pressure at different locations in the cardiovascular tree.

The ClearSight® measures arterial pressure in the finger to estimate brachial blood pressure. Extensive studies have evaluated the performance of the ClearSight® device in a wide range of people and have shown satisfactory results in measuring brachial artery pressure with high accuracy and precision when compared to radial artery catheters, making the ClearSight® device the gold standard for measuring beat-to-beat blood pressure. However, it is important to note that these studies also show variable results among patients, especially when measuring systolic blood pressure.

Blood pressure data analysis for each heartbeat

When the ClearSight® starts measuring, it measures 10 cardiac cycles before starting the calibration step. After self-assessment of accuracy, the ClearSight® measures 20 cardiac cycles and repeats the calibration step. The ClearSight® will continue to measure more cardiac cycles until it reaches 70 cardiac cycles, which is considered the most accurate and precise in blood pressure measurement. In this critical region, the capacitive pressure sensor was compared with the ClearSight®. In addition, after each subsequent 70 heartbeat interval, the subject was asked to alternate between deep and normal breathing. Deep breathing can increase blood pressure due to slight cardiac compression caused by lung expansion. The subject was asked to breathe deeply to evaluate the ability of the soft capacitive pressure sensor to detect larger changes in blood pressure.

16A-16B show data collected for one subject. In FIG. 16A, qualitative analysis shows that the two devices measured similar trends in blood pressure. This is evident during deep breathing periods, where infrequent blood pressure changes are reflected by both the pressure sensor and ClearSight®. SDP, DBP, and MAP were then plotted against each other and analyzed using linear regression, as can be seen in FIG. 16B. The agreement between the pressure sensor and the ClearSight® device showed a strong correlation, with ^R2 = .765 for SDP, ^R2 = .902 for DBP, and ^R2 = .839 for MAP. As mentioned above, the ClearSight® device has difficulty in measuring SBP values with high accuracy and precision, which may explain the lower ^R2 between the pressure sensor and ClearSight®.

Data for the remaining subjects can be found in Tables 1 to 3.

To further evaluate the accuracy and precision of the pressure sensor's beat-to-beat blood pressure monitoring capabilities, the pressure sensor was calibrated to ClearSight® and a model of the pressure sensor was created and cross-validated. To create the model, three consecutive cardiac cycles were first averaged together. After averaging, 75% of the data was randomly selected to generate a linear regression model of the pressure sensor. The remaining retained data set from the pressure sensor was converted to millimeters of mercury (mmHg), a unit of blood pressure. An example of this calibration from one subject is shown in Figures 17A-C.

A Bland-Altman analysis was then used to assess the agreement between the pressure sensor and the ClearSight® in measuring blood pressure. Bland-Altman examines the difference in simultaneously measured blood pressure plotted against the mean value of simultaneously measured blood pressure. A larger difference indicates a greater degree of disagreement between the two devices. As shown in FIG. 17D, all seven subjects are combined in one Bland-Altman plot, including the data set of subject 1, who was tested with two additional sensors. The calculated mean error and standard deviation were -0.054 ± 2.09 mmHg. ISO 81060-2, set by AAMI, indicates that if the mean error is less than 5 mmHg and the standard deviation is less than 8 mmHg, the NIBP is considered to be replaceable with an arterial catheter. The Bland-Altman analysis here indicates that the mean error and standard deviation are well below the requirements set forth in the ISO standard. Tables 4-5 also show the mean error and standard deviation calculated without cardiac cycle averaging, which again shows that the pressure sensor is well within the ISO standard. This suggests that when calibrated to the ClearSight® device, the pressure sensor is highly accurate and precise in measuring blood pressure.

The soft capacitive pressure sensor of the present invention can be utilized for radial artery pressure measurement applications. Stable measurement of arterial pressure allowed tracking and detection of subtle changes in SBP, DBP, and MAP pressures. Correlation with the ClearSight® device shows promising results toward potential ambulatory beat-to-beat NIBP monitoring. The electromechanical nature of the pressure sensor enables the capability to monitor a wide range of pressures with high accuracy. In addition, the fast response time and wide dynamic range of the capacitive pressure sensor allows high fidelity detection of the radial artery pulsatile waveform, enabling high accuracy and precise measurement of blood pressure, which is essential for monitoring acute cardiovascular events.

In a further embodiment, the present invention provides a facile method for developing a soft capacitive pressure sensor with favorable electromechanical properties for measuring pulsatile arterial blood flow. By utilizing soft, highly wrinkled thin film electrodes, a soft pressure sensor with a wide dynamic range could be fabricated that can be effectively coupled to the human body to quantify localized, subtle pressures. The sensor also demonstrated sufficient pressure sensitivity, fast response time, as well as mechanical robustness against cyclic loading.

Blood pressure calibration using pulse transit time (PTT)

Pulse pressure measurements obtained using the pressure sensor of the present invention are highly correlated with blood pressure measurements obtained using ClearSight®. However, there is still a need to calibrate the pulse pressure measurements obtained with the pressure sensor to absolute blood pressure without continued reliance on ClearSight®.

There are two notable methods that can be used to calibrate a pressure sensor to measure absolute blood pressure. One method involves the use of a traditional oscillometric arm cuff to calibrate the systolic and diastolic blood pressure. The oscillometric arm cuff calculates the systolic and diastolic blood pressure over a set period during inflation and deflation of the arm cuff. This measurement does not provide beat-to-beat blood pressure, but the values calculated by the oscillometric arm cuff can be used to calibrate the pressure sensor to a reference value for blood pressure.

Another method involves the use of pulse transit time (PTT). PTT is the time that indicates how quickly the pulse wave travels from one arterial location to another. The time that the pulse wave travels is directly related to blood pressure. For example, the faster the pulse wave travels, the higher the blood pressure and vice versa. PTT can be measured using an ECG and a pressure sensor placed in a peripheral artery (i.e., radial artery). The ECG provides the moment the heart contracts, and the pressure sensor provides information about how long it took for the pulse to travel to the peripheral artery. More specifically, PTT is the time between the R peak of the ECG and the corresponding diastolic peak measured by the pressure sensor (Figure 18). PTT can also be measured by using two peripheral sensors to measure the time it takes for the pulse to travel from one sensor to the next. Studies have shown that measuring the peripheral arterial pulse wave rather than using an ECG provides a better correlation with blood pressure. This can be achieved by measuring with two pressure sensors, one placed in the radial artery and one in the brachial artery. Note that PTT can only provide absolute blood pressure values, and does not provide any other information about the pulse wave (e.g., augmentation index) (Figure 19).

PTT is a subject-dependent parameter and requires calibration to blood pressure. This can be accomplished by calibrating to a traditional arm cuff as described above (PTT can also be calibrated using ClearSight®). The PTT model can be used for months to years before the PTT model needs to be recalibrated. Once the PTT model is calibrated, it can then be used to subsequently calibrate the pressure sensor (Figures 20A-C). The advantage of calibrating the pressure sensor to the PTT model is the feedback mechanism that calibrates the pressure sensor to blood pressure after large movements or when the pressure sensor is switched on and off.

As used herein, the term "about" refers to plus or minus 10% of the referenced number.

While preferred embodiments of the present invention have been shown and described, it will be readily apparent to one of ordinary skill in the art that modifications may be made thereto that do not exceed the scope of the appended claims. Accordingly, the scope of the present invention is to be limited only by the following claims. In some embodiments, the figures presented in this patent application are drawn to scale, including angles, dimensional ratios, etc. In some embodiments, the figures are merely representative, and the claims are not limited by the dimensions of the figures. In some embodiments, the description of the invention set forth herein using the phrase "comprising" includes embodiments that can be described as "consisting essentially of" or "consisting of," and thus the requirement of a written description to claim one or more embodiments of the invention using the phrase "consisting essentially of" or "consisting of" is satisfied.

The reference numerals recited in the claims below are merely for ease of review of this patent application, are exemplary, and are not intended to limit the scope of the claims in any way to the particular features having corresponding reference numerals in the drawings.

REFERENCE SIGNS LIST 100 Capacitive pressure sensor 110 First electrode layer 120 Second electrode layer 112, 122 Elastomer layers 114, 124 Conductive film 125 Ridge 130 Dielectric layer 140 Air gap

Claims (17)

A soft capacitive pressure sensor, comprising:
a. a first electrode layer;
b. a second electrode layer;
c. a dielectric layer disposed on the first electrode layer such that the dielectric layer is between the first and second electrode layers;
d. a plurality of resilient structures protruding from the second electrode layer toward the dielectric layer and the first electrode layer, the plurality of resilient structures creating a gap separating the first electrode layer and the second electrode layer;
a first conductive metal film disposed between the plurality of resilient structures , the second electrode layer comprising the first conductive metal film, wherein when the capacitive pressure sensor is in a resting configuration, the air gap is disposed between the first electrode layer and the second electrode layer, the air gap acts as a second dielectric layer, and when the capacitive pressure sensor is compressed, the first electrode layer and the second electrode layer move closer to one another, thereby decreasing a height of the air gap and increasing pressure sensitivity and capacitance of the capacitive pressure sensor. The capacitive pressure sensor of claim 1, wherein the first electrode layer comprises a first elastomer layer and a second conductive metal film disposed on the first elastomer layer, the second conductive metal film being between the first elastomer layer and the dielectric layer. The capacitive pressure sensor of claim 2, wherein the second electrode layer comprises a second elastomer layer, and the first conductive metal film is disposed on the second elastomer layer. The capacitive pressure sensor of claim 3, wherein the first and second elastomer layers include polydimethylsiloxane or polyurethane. The capacitive pressure sensor of claim 4, wherein the first and second conductive metal films include carbon nanotubes or wrinkled thin films of gold, silver, copper, or aluminum. The capacitive pressure sensor of claim 1, wherein the dielectric layer comprises a silicone elastomer, lead zirconate titanate, barium titanate, polyvinylidene fluoride, or an oxide of zirconia, titania, or silica. 1. A method for monitoring blood pressure in a subject, the method comprising:
a. providing a capacitive pressure sensor as claimed in claim 1;
b. attaching the capacitive pressure sensor to the subject's wrist where the radial artery is located;
c. measuring a blood pressure signal of the subject using the capacitive pressure sensor, wherein pulsation of the radial artery causes compression of the capacitive pressure sensor, and the capacitive pressure sensor detects the blood pressure signal corresponding to pulsation of the radial artery;
and d. determining an absolute blood pressure value from said measured signal using a calibration model. The method of claim 7, wherein the blood pressure is monitored beat-by-beat. calibrating the capacitive pressure sensor using an artificial neural network (ANN);
a. measuring a plurality of blood pressure signals from the subject;
b. measuring one or more biological input variables from said subject;
c. combining the measured blood pressure signals and the biological input variables to form a training set;
and d. training the ANN using the training set to generate the calibration model. The method of claim 9, wherein the training set includes blood pressure signal measurements from multiple subjects and one or more biological input variables. The method of claim 9, wherein the one or more biological input variables include sensor pressure, pulse rate, EKG data, accelerometer data, gyroscope data, magnetometer data, or hemodynamic monitoring data. The method of claim 9, further comprising the step of subtracting motion artifacts by including motion data in the training set. 1. A method of manufacturing a capacitive pressure sensor, the method comprising:
a. attaching a first conductive metal film to a first elastomeric layer to form a first electrode layer;
b. attaching the dielectric layer to the first electrode layer such that the first conductive metal film is disposed between the first elastomeric layer and the dielectric layer;
c. molding a second elastomeric layer having a plurality of resilient structures protruding from a surface of the second elastomeric layer;
d. attaching a second conductive metal film to the second elastomeric layer to form a second electrode layer such that the second conductive metal film is disposed between the plurality of elastic structures;
e. attaching the first electrode layer to the second electrode layer such that the plurality of resilient structures are oriented toward the dielectric layer and the first electrode layer, the plurality of resilient structures creating an air gap between the first electrode layer and the second electrode layer, the air gap acting as a second dielectric layer;
When the capacitive pressure sensor is compressed, the first electrode layer and the second electrode layer move closer to each other, thereby decreasing the height of the gap and increasing the pressure sensitivity and capacitance of the capacitive pressure sensor. The method of claim 13, wherein the first and second elastomeric layers comprise polydimethylsiloxane or polyurethane. The method of claim 13, wherein the first and second conductive metal films comprise carbon nanotubes or wrinkled thin films of gold, silver, copper, or aluminum. The method of claim 13, wherein the dielectric layer comprises soft silicone rubber, lead zirconate titanate, barium titanate, polyvinylidene fluoride, or oxides of zirconia, titania, or silica. The method of claim 13, further comprising calibrating the capacitive pressure sensor using an artificial neural network (ANN), the calibration comprising generating a calibration model by training the ANN using a training set of measured blood pressure signals and one or more biological input variables.

Images