Sensor, system and method for contactless sensing of a physiological parameter of a body

Abstract

A sensor 104 for contactless sensing of a physiological parameter of a body 102 is described. In an embodiment, a sensor 104 includes a waveguide including a metamaterial and configured to receive a transmit signal and propagate the transmit signal in an artificial surface plasma mode along the waveguide to generate an evanescent electromagnetic field and to provide a receive signal, where the waveguide is placed at a predetermined distance away from the body 102, the present invention relates to a method for contactless sensing of disturbances generated by physiological movement of a body 102 using an evanescent electromagnetic field, the disturbances generating a phase shift between a transmitted signal and a received signal for determining a physiological parameter of the body 102. Systems 100 and methods 200 for contactless sensing of physiological parameters of a body 102 are also described.

Description

Technical Field

[0001] The present disclosure relates to sensors, systems, and methods for contactless sensing of physiological parameters of the body. Background Art

[0002] Clinical sensing modalities for the body's vital signs typically require electrodes and sensors to be attached to the body, mostly directly to the skin, which imposes constraints on mobility and in many cases causes significant inconvenience or discomfort to the user. Therefore, there is an increasing interest in developing contactless vital sign sensing systems that do not require skin contact.

[0003] Doppler radar technology has received considerable attention in contactless vital sign monitoring. Doppler radar relies on the detection of the phase shift of reflected radio frequency (RF) waves from a moving object compared to the original transmitted wave toward the object. In this context, the detection of vital signs using radar relies on the fact that during each cardiopulmonary cycle, the physical movement of the body surface due to the deformation of the heart and lungs modulates the phase of the reflected signal, which is then measured. Since the pioneering work of JCLin in 1975, which demonstrated an X-band Doppler radar system for respiration measurement, research efforts have resulted in smaller and lighter radar systems that enable power-efficient and highly accurate vital sign sensing. However, despite these advances, such systems rely on radiated RF waves, which limits their applicability in practice because it is challenging to detect vital signs in the presence of background noise, such as generated by reflections from different parts of the body and / or other objects in the surroundings. This is even more true for environmental health monitoring applications, where RF reflections across large body areas and the movement of different body parts or other objects in the vicinity can result in background interference that is difficult to distinguish from the target signal.

[0004] It would therefore be desirable to provide a system and method for contactless sensing of physiological parameters of the body (eg, vital signs) that addresses the above-mentioned problems and / or provides a useful alternative.

[0005] Furthermore, other desirable features and characteristics will become apparent from the subsequent detailed description and the appended claims, taken in conjunction with the accompanying drawings and this background of the disclosure. Summary of the invention

[0006] Aspects of the present application relate to sensors, systems, and methods for contactless sensing of physiological parameters of the body.

[0007] According to a first aspect, a sensor for non-contact sensing of physiological parameters of a body is provided, the sensor comprising: a waveguide comprising a metamaterial and configured to receive a transmit signal, and propagate the transmit signal along the waveguide in an artificial surface plasma mode to generate an evanescent electromagnetic field, and provide a receive signal, wherein the waveguide is placed at a predetermined distance away from the body, and is used to perform non-contact sensing of disturbances generated by physiological movements of the body using the evanescent electromagnetic field, the disturbance generating a phase shift between the transmit signal and the receive signal, for determining the physiological parameters of the body.

[0008] Thus, the described embodiments provide a sensor for non-contact sensing of physiological parameters of the body. A sensor for non-contact sensing of disturbances generated by physiological movements of the body can be used to determine physiological parameters of the body by having a waveguide, the waveguide being configured to propagate a transmitted signal in an artificial surface plasma mode along the waveguide to generate an evanescent electromagnetic field, and being placed at a predetermined distance away from the body. Due to the spatial limitations of electromagnetic waves, sensing using an evanescent electromagnetic field enhances sensing sensitivity. In addition, the evanescent electromagnetic field is non-radiative, thereby minimizing background noise or clutter, which otherwise may be picked up by the waveguide due to random body movement and / or reflections from multiple objects in the periphery. Non-contact sensing of physiological parameters using the present sensor (e.g., through clothing) also enables sensing / monitoring of physiological parameters / vital signs, which is convenient and comfortable for a variety of clinical and daily life situations without the need to physically couple the sensor to the skin of the body. The use of an evanescent field for local measurement of physiological parameters of the body also allows multiplexing sensing of different parts of the body to obtain multiple physiological signals at the same time, as will be described in the exemplary embodiments described below.

[0009] The waveguide may include a sensing layer on a sensing side of the waveguide suitable for detecting disturbances generated by physiological movements of the body, a ground layer on an opposite side of the sensing side, and a non-conductive layer sandwiched between the sensing layer and the ground layer, wherein the ground layer is configured to confine the evanescent electromagnetic field to the sensing side of the waveguide.

[0010] The sensing layer may include a comb-like rectangular strip having an elongated base and a plurality of teeth extending along and from the elongated base, wherein adjacent teeth of the plurality of teeth are separated by a gap.

[0011] The height of the teeth measured from the elongated base can be adapted to vary the degree of wavelength confinement of the artificial surface plasmon mode.

[0012] According to a second aspect, there is provided a system for contactless sensing of a physiological parameter of a body, the system comprising one or more sensors as described above and a software defined radio (SDR) system configured to provide a transmit signal and receive a receive signal.

[0013] The SDR system may include a digital-to-analog converter (DAC), and the SDR system may be configured to: generate a digital complex baseband signal; convert the digital complex baseband signal using the DAC to form an analog baseband signal; modulate the analog baseband signal with a carrier signal to provide a transmit signal; demodulate the received signal to obtain in-phase and quadrature (IQ) components associated with the digital complex baseband signal; and digitize the obtained IQ components.

[0014] The SDR system may be configured to perform a complex conjugate multiplication of the digital complex baseband signal and the digitized IQ component to determine a phase shift signal associated with a phase shift between the transmit signal and the receive signal.

[0015] The SDR system may be configured to filter the phase shift signal with a low pass filter and downsample the filtered phase shift signal to form a decimated phase shift signal. The cutoff frequency of the low pass filter may be greater than 5 Hertz (Hz) and less than 200 Hz.

[0016] The SDR system may be adapted to perform inverse tangent demodulation and unwrap the extracted phase shifted signal to obtain a time-varying phase signal associated with the phase shift between the transmitted signal and the received signal.

[0017] Physiological movement can be associated with more than one physiological parameter, and the system may include a processor and a data storage device, the data storage device storing computer program instructions, the computer program instructions being operable to cause the processor to: process the time-varying phase signal with a bandpass filter to separate the time-varying phase signal into separate components, the separate components being associated with each of the more than one physiological parameters.

[0018] Among them, one or more sensors may include a first sensor and a second sensor, the first sensor is provided on the back of the body and is suitable for detecting a respiratory signal and a cardiac signal associated with the body, and the second sensor is provided on the wrist of the body and is suitable for detecting a radial artery pulse signal associated with the body. The system may include a processor and a data storage device, the data storage device stores computer program instructions, and the computer program instructions are operable to enable the processor to: process a first time-varying phase signal associated with the first sensor with a bandpass filter to separate the first time-varying phase signal into a time-varying respiratory phase signal and a time-varying cardiac phase signal; and process a second time-varying phase signal associated with the second sensor with a bandpass filter to obtain a time-varying radial artery pulse phase signal.

[0019] The data storage device may store computer program instructions operable to cause the processor to: perform a fast Fourier transform on the first 15 seconds of each signal segment of the time-varying respiratory phase signal to estimate a respiratory period; calculate a moving average curve by taking the mean of the time-varying respiratory phase signal over a time window equal to twice the respiratory period to generate each data point of the moving average curve; calculate an intercept between the moving average curve and the time-varying respiratory phase signal; use the calculated intercept to identify peaks on the time-varying respiratory phase signal, wherein each of the peaks is identified as the maximum value between an intercept having a positive slope and a subsequent intercept having a negative slope; and calculate a respiratory cycle as the duration time between two adjacent peaks.

[0020] The data storage device may store computer program instructions operable to cause the processor to: calculate a first order time derivative waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; set all negative values ​​of the first order time derivative waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to zero to form a composite waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; square the composite waveform associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a squared signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; filter the squared signal using a moving average filter having a predetermined time window to generate an integrated signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; detect the first order time derivative waveform associated with the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; The invention relates to a method for detecting peaks in an integrated signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; detecting peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; using the detected peaks in the integrated signal to verify the detected peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; calculating beat positions in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal, each of the beat positions being the nearest preceding positive zero intercept relative to each verified peak of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; and calculating a beat-to-beat interval associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal, the beat-to-beat interval being the time interval between consecutive beat positions, wherein the beat-to-beat interval associated with the time-varying cardiac phase signal is related to the heart rate, and the beat-to-beat interval associated with the time-varying radial artery pulse phase signal is related to the radial artery pulse rate.

[0021] The data storage device may store computer program instructions operable to cause the processor to: receive a time-varying cardiac phase signal and a time-varying radial artery pulse phase signal; process the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a processed time-varying cardiac phase signal and a processed time-varying radial artery pulse phase signal; generate an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal based on the processed time-varying cardiac phase signal and the processed time-varying radial artery pulse phase signal using a trained machine learning model, wherein a peak of the aligned time-varying cardiac phase signal corresponds to an electrocardiogram (ECG) R wave peak, and a peak of the aligned time-varying radial artery pulse phase signal corresponds to a photoplethysmography (PPG) maximum first derivative (MFD) point; calculate a pulse transit time (PPT) as a time delay between one of the peaks of the aligned time-varying cardiac phase signal and a corresponding one of the peaks of the aligned time-varying radial artery pulse phase signal; and convert the calculated PPT into systolic and diastolic pressure values. Converting the calculated PTT to systolic and diastolic blood pressure values ​​may include using a linear PPT-blood pressure relationship.

[0022] The data storage device may store computer program instructions operable to cause the processor to: search for a local maximum of the aligned time-varying radial artery pulse phase signal within a subsequent time window of 0.15 to 0.4 seconds (s) of one of the peaks of the aligned time-varying cardiac phase signal, the local maximum being a corresponding one of the peaks used to calculate the PPT.

[0023] The data storage device may store computer program instructions operable to cause the processor to: receive training data, the training data comprising a training time-varying cardiac phase signal and a training time-varying radial artery pulse phase signal; process the training data to form a training processed time-varying cardiac phase signal and a training processed time-varying radial artery pulse phase signal; and train a machine learning model to form a trained machine learning model, wherein the data storage device stores computer program instructions operable to cause the processor to train the machine learning model, and the data storage device may store computer program instructions operable to cause the processor to: use the machine learning model to generate a training time-varying cardiac phase signal output and a training time-varying radial artery pulse phase signal output based on the training processed time-varying cardiac phase signal and the training processed time-varying radial artery pulse phase signal; and use a regression layer to minimize the mean square error (MSE) between each of the training time-varying cardiac phase signal output and the training time-varying radial artery pulse phase signal output and the corresponding target time-varying cardiac phase signal and time-varying radial artery pulse phase signal for forming a trained machine learning model.

[0024] The trained machine learning model may include a long short-term memory (LSTM) network followed by a fully connected (FC) layer for each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal.

[0025] The data storage stores computer program instructions operable to cause the processor to process the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal. The data storage may store computer program instructions operable to cause the processor to: shift the time-varying cardiac phase signal left by a predetermined amount of time to form a processed time-varying cardiac phase signal; and differentiate the time-varying radial artery pulse phase signal with respect to time to generate a time derivative of the time-varying radial artery pulse phase signal to form a processed time-varying radial artery pulse phase signal.

[0026] The data storage device may store computer program instructions operable to cause the processor to: identify epochs for blood pressure sensing, each of the identified epochs being a time window having a predetermined time period during which both the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal are present; calculate a mean heart rate and a mean pulse rate for the identified epochs; select one or more candidate epochs from the identified epochs, wherein an absolute difference between the mean heart rate and the mean pulse rate for each of the one or more candidate epochs is less than two beats per minute; and calculate a signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in each of the one or more candidate epochs. e ) is calculated as:

[0027]

[0028] Where N is the number of detected beats of the time-varying cardiac signal or the time-varying radial artery pulse phase signal in each of the one or more candidate epochs, t e is the length of each of the corresponding one or more candidate epochs in minutes, and I is the beat-to-beat interval in seconds from each consecutive pair of detected beats; and one or more detection epochs are selected from the one or more candidate epochs for continuous blood pressure detection, wherein the signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in the one or more detection epochs is e ) is greater than 0.5.

[0029] According to a third aspect, a method for non-contact sensing of a physiological parameter of a body using one or more sensors is provided, wherein each of the one or more sensors includes a waveguide, the waveguide includes a metamaterial and is configured to propagate a transmission signal in an artificial surface plasma mode along the waveguide to generate an evanescent electromagnetic field, and provide a reception signal, the evanescent electromagnetic field is used for non-contact sensing of disturbances generated by physiological movements of the body, the method comprising: (i) placing the waveguide at a predetermined distance away from the body for non-contact sensing of the disturbance; (ii) providing the transmission signal to the waveguide; (iii) receiving the reception signal from the waveguide; and (iv) processing the reception signal and the transmission signal to determine a phase shift between the transmission signal and the reception signal caused by the disturbance, for determining the physiological parameter of the body.

[0030] The waveguide can be made of a flexible material. This allows the waveguide to be easily implemented in clothing, for example, for sensing or monitoring physiological parameters of the body.

[0031] The waveguide may include a sensing layer on a sensing side of the waveguide adapted to detect disturbances generated by physiological movements of the body, a ground layer on an opposite side of the sensing side, and a non-conductive layer sandwiched between the sensing layer and the ground layer, wherein the ground layer may be configured to confine the evanescent electromagnetic field to the sensing side of the waveguide. This reduces measurement noise picked up by the waveguide and improves the sensitivity of the waveguide.

[0032] The sensing layer may include a comb-like rectangular strip having an elongated base and a plurality of teeth extending along and from the elongated base, wherein adjacent teeth of the plurality of teeth are separated by a gap.

[0033] The method may comprise providing a transmit signal having a frequency range of 400 MHz to 200 GHz.In an embodiment, a carrier signal of the transmit signal may have a frequency range of 400 MHz to 200 GHz.

[0034] The method may include: generating a digital complex baseband signal; converting the digital complex baseband signal using a digital-to-analog converter (DAC) to form an analog baseband signal; modulating the analog baseband signal with a carrier signal to provide a transmit signal; demodulating the received signal to obtain in-phase and quadrature (IQ) components associated with the digital complex baseband signal; and digitizing the obtained IQ components.

[0035] The method may include performing a complex conjugate multiplication of the digital complex baseband signal and the digitized IQ component; and determining a phase shift signal associated with a phase shift between the transmit signal and the receive signal.

[0036] The method may include filtering the phase-shifted signal with a low pass filter; and down-sampling the filtered phase-shifted signal to form a decimated phase-shifted signal.

[0037] The method may include performing arctangent demodulation and unwrapping on the extracted phase-shifted signal to obtain a time-varying phase signal associated with a phase shift between the transmitted signal and the received signal.

[0038] Among them, one or more sensors may include a first sensor and a second sensor, the first sensor is provided on the back of the body and is suitable for detecting a respiratory signal and a cardiac signal associated with the body, and the second sensor is provided on the wrist of the body and is suitable for detecting a radial artery pulse signal associated with the body. The method may include: processing a first time-varying phase signal associated with the first sensor with a bandpass filter to separate the first time-varying phase signal into a time-varying respiratory phase signal and a time-varying phase cardiac phase signal, and processing a second time-varying phase signal associated with the second sensor with a bandpass filter to obtain a time-varying radial artery pulse phase signal.

[0039] The method may include: receiving a time-varying cardiac phase signal and a time-varying radial artery pulse phase signal; processing the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a processed time-varying cardiac phase signal and a processed time-varying radial artery pulse phase signal; using a trained machine learning model to generate an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal based on the processed time-varying cardiac phase signal and the processed time-varying radial artery pulse phase signal, wherein a peak of the aligned time-varying cardiac phase signal corresponds to an electrocardiogram (ECG) R wave peak, and a peak of the aligned time-varying radial artery pulse phase signal corresponds to a photoplethysmography (PPG) maximum first derivative (MFD) point; calculating a pulse transit time (PPT) as a time delay between one of the peaks of the aligned time-varying cardiac phase signal and a corresponding one of the peaks of the aligned time-varying radial artery pulse phase signal; and converting the calculated PPT into a systolic pressure value and a diastolic pressure value. Converting the calculated PTT to systolic and diastolic blood pressure values ​​may include using a linear PPT-blood pressure relationship.

[0040] The method may include searching for a local maximum of the aligned time-varying radial artery pulse phase signal within a subsequent time window of 0.15s to 0.4s of one of the peaks of the aligned time-varying cardiac phase signal, the local maximum being a corresponding one of the peaks used to calculate the PPT.

[0041] The method may include: identifying epochs for blood pressure sensing, each of the identified epochs being a time window having a predetermined time period during which both a time-varying cardiac phase signal and a time-varying radial artery pulse signal are present; calculating a heart rate mean and a pulse rate mean for the identified epochs; selecting one or more candidate epochs from the identified epochs, wherein an absolute difference between the heart rate mean and the pulse rate mean for each of the one or more candidate epochs is less than two beats per minute; and calculating a signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in each of the one or more candidate epochs. e ) is calculated as:

[0042]

[0043] Where N is the number of detected beats of the time-varying cardiac phase signal or the time-varying radial artery pulse phase signal in each of the one or more candidate epochs, t e is the length of each of the corresponding one or more candidate epochs in minutes, and I is the beat-to-beat interval in seconds from each consecutive pair of detected beats; and one or more detection epochs are selected from the one or more candidate epochs for continuous blood pressure detection, wherein the signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in the one or more candidate epochs is e ) is greater than 0.5.

[0044] Therefore, embodiments provide sensors, systems and methods for non-contact sensing of physiological parameters of the body. By using a sensor including a waveguide, and placing the waveguide at a predetermined distance away from the body, the sensor provides non-contact sensing of disturbances generated by physiological movements of the body for determining physiological parameters of the body, and the waveguide is configured to propagate a transmitted signal in an artificial surface plasma mode along the waveguide to generate an evanescent electromagnetic field. Due to the spatial confinement of electromagnetic energy in the evanescent electromagnetic field, sensing using the evanescent electromagnetic field enhances sensing sensitivity. In addition, the evanescent electromagnetic field is non-radiative, thereby minimizing background noise or clutter, which may otherwise be picked up by the waveguide due to random body movement and / or reflections from multiple objects in the periphery. Non-contact sensing of physiological parameters using the present system and method (e.g., through clothing) also enables sensing / monitoring of physiological parameters / vital signs, which is convenient and comfortable for a variety of clinical and daily life situations. The use of non-radiative local sensing by evanescent electromagnetic fields allows multiplexed sensing of different parts of the body simultaneously to obtain multiple physiological signals, such as respiration rate, heart rate, and radial artery pulse rate. In an embodiment, cuffless blood pressure monitoring can be implemented using heart rate and radial artery pulse rate obtained from the back and wrist of the body, respectively, to obtain environmental vital sign monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Embodiments will now be described, by way of example only, with reference to the following drawings, in which:

[0046] Figure 1A and Figure 1B A system for contactless sensing of a physiological parameter of a body and a diagram of the body are shown according to an embodiment, wherein Figure 1A A schematic diagram of a system including a sensor, a software defined radio system and a computer is shown, and Figure 1B Shows the description based on Figure 1A Diagram of non-contact sensing of the sensor;

[0047] Figure 2 The use of the embodiment is shown Figure 1A A flowchart of the steps of a method for performing non-contact sensing of physiological parameters of a body by a system;

[0048] Figure 3A and Figure 3B It shows that according to the embodiment Figure 1A An image of a portion of a system where Figure 3A An image is shown illustrating the integration of a sensor including a spoof surface plasmon (SSP) metamaterial waveguide with a software defined radio (SDR) system via a coaxial cable (SDR system not shown), and Figure 3BShows instructions for use Figure 3A The system is used as an image of a seat-integrated health monitoring system, where the heartbeat signal and breathing signal can be measured from the back of the body;

[0049] Figure 4A and Figure 4B A diagram showing simulation results of an SSP waveguide according to an embodiment, wherein Figure 4A The simulation|S is shown to illustrate the case where there is no human body and the case where there is a human body. 21 |The resulting graph, and Figure 4B A diagram illustrating the electric field of the waveguide in the xz plane when interfaced with a transmit signal is shown to illustrate the interaction between the electric field and the human body for sensing (white arrows indicate the direction of wave propagation);

[0050] Figure 5 shows a schematic diagram for explaining the integration of a metamaterial fabric sensor with a software defined radio (SDR) system according to an embodiment;

[0051] Figure 6 A flow chart showing a method for processing physiological signals to detect heartbeat and pulse beat according to an embodiment;

[0052] Fig. 7A and Figure 7B shows a graph of measured vital sign waveforms according to an embodiment, wherein Fig. 7A shows a graph of the breathing signal, and Figure 7B shows a graph of a heartbeat signal with detected beat locations (shown as circles);

[0053] Fig. 8A , Figure 8B and Figure 8C The example for comparison is shown in FIG. Figure 3A A graph of a heartbeat measured by an SSP metamaterial sensor and a heartbeat measured using a reference electrocardiogram (ECG), wherein Fig. 8A shows a graph comparing the measured beat-to-beat interval (BBI) with the reference electrocardiogram (ECG) R-peak interval in milliseconds, Figure 8B A graph showing the linear correlation between BBI measured using the SSP metamaterial sensor and the RR interval measured by ECG, and Figure 8C Bland-Altman plots showing pairwise comparisons between ECG RR intervals and SSP BBI are shown;

[0054] Fig. 9 An example of using the embodiment is shown. Figure 1AFigure 3. System for multi-point tracking of physiological signals from different body parts;

[0055] Fig. 10A and Fig. 10B A diagram showing a full wave simulation of the electric field at 2.4 GHz for sensing the radial artery according to an embodiment, wherein Fig. 10A A diagram showing a full wave simulation of the electric field using conventional radiation sensing, and Fig. 10B A diagram showing a full wave simulation of the electric field using non-radiative SSP surface wave sensing;

[0056] Fig.11 shows a diagram of an arm placed over a metamaterial fabric waveguide according to an embodiment;

[0057] Fig. 12A and Fig. 12B FIG. 4 shows a graph of a geometric parameter h of a metamaterial fabric waveguide according to an embodiment, wherein Fig. 12A The plasma wave number β is shown for different fabric substrates. p as a function of the geometric parameter h, and Fig. 12B shows a graph of transmission loss as a function of fabric conductivity at 2.4 GHz for different geometric parameters h;

[0058] Fig.13 shows dispersion curves of metamaterial fabric waveguides with different geometric parameters h according to an embodiment;

[0059] Fig.14A and Fig. 14B A diagram is shown to illustrate the design of a planar matching section of a metamaterial fabric waveguide according to an embodiment, wherein: Fig.14A A diagram is shown for explaining the structure of the plane matching section, and Fig. 14B shows the relationship between the transmission coefficient and the geometric parameter h for different numbers of matching units N;

[0060] Fig.15A and Fig. 15B A diagram showing a full wave simulation during sensing according to an embodiment, wherein Fig.15A a diagram showing a full wave simulation of a cross section of an arm during radial artery pulse sensing, and Fig. 15B A diagram showing a full wave simulation of a cross section of a human torso during heartbeat sensing;

[0061] Fig.16 Graphs showing electric field distribution in the yz plane (top) and the xz plane (bottom) of a conventional microstrip sensor and a metamaterial fabric sensor of the present disclosure according to an embodiment;

[0062] Fig.17A and Fig. 17B shows simulation |S without interaction with the body and due to interaction with different body parts according to an embodiment 21 |A plot of the transmission spectrum, where Fig.17A Shows the simulation of the SSP metamaterial fabric sensor |S 21 | Plot of the transmission spectrum, Fig. 17B shows the simulation of the microstrip fabric sensor |S 21 | Graph of the transmission spectrum.

[0063] Fig.18A and Fig.18B Graphs showing simulated phase changes ΔΦ obtained from conventional microstrip sensors and metamaterial fabric sensors of the present disclosure according to embodiments are shown, wherein Fig.18A shows a graph of simulated phase change ΔΦ as a result of the radial artery pulse during the cardiac cycle, and Fig.18B shows a graph of simulated phase changes ΔΦ as a result of heartbeats within a cardiac cycle;

[0064] Fig.19 shows a graph of simulated phase change ΔΦ as a function of separation distance d at the back of the body and the wrist between a conventional microstrip sensor and a metamaterial fabric sensor of the present disclosure according to an embodiment;

[0065] Fig. 20A and Fig. 20B FIG. 4 is a graph showing transmission loss of an SSP metamaterial fabric sensor as a function of bending radius according to an embodiment, wherein Fig. 20A shows a graph of the transmission loss of the SSP metamaterial fabric sensor bent in the xz plane, and Fig. 20B A graph showing the transmission loss of the SSP metamaterial fabric sensor when bent in the xy plane;

[0066] Fig.21A and Fig. 21B A diagram showing two SSP metamaterial fabric sensor designs according to an embodiment, wherein Fig.21A shows a U-shaped sensor design for a chair, Fig. 21B A straight sensor design for use on a table is shown;

[0067] Fig. 22 The use of the embodiment is shown Figure 1A Illustration of a system's vital signs data collection setup;

[0068] Fig.23 The use of the embodiment is shown Fig.23A graph of sample data collected from two sensor channels CH1 and CH2 by the vital signs data collection setup;

[0069] Fig.24A , Fig. 24B and Fig.24C FIG. 4 shows a Bland-Altman plot of hop-by-hop intervals obtained using reference measurements and measurements from a metamaterial fabric sensor of the present disclosure, according to an embodiment, wherein Fig.24A shows a Bland-Altman plot for respiratory rate, Fig. 24B shows a Bland-Altman plot for heart rate, Fig.24C A Bland-Altman plot for radial artery pulse rate is shown;

[0070] Fig.25 A diagram illustrating a machine learning model according to an embodiment, the machine learning model including a long short-term memory (LSTM) network and a fully-connected (FC) layer for performing alignment of a heart signal (CH1) and a pulse signal (CH2) to obtain a pulse transit time (PTT) for blood pressure measurement;

[0071] Fig.26 shows a Bland-Altman plot of an estimated PTT obtained from test data and an estimated PTT obtained from data of a metamaterial fabric sensor using a sensing system according to an embodiment;

[0072] Fig. 27 is a flow chart illustrating steps of a computer-implemented method for generating an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal to calculate pulse transit time (PPT) for conversion to systolic and diastolic pressure values, according to an embodiment;

[0073] Fig.28 A diagram is shown illustrating Fig.25 The architecture and training of one of the LSTM aligner channels of the machine learning model;

[0074] Fig.29A and Fig.29B A graph showing sample data used to verify pulse signal alignment and cardiac signal alignment according to an embodiment Fig.25 A machine learning model, where Fig.29A shows a graph of the average detection error relative to the subject ID for pulse signal alignment, and Fig.29B showing a graph of mean detection error relative to subject ID for cardiac signal alignment;

[0075] Fig.30 A photograph showing multiplexed health monitoring using a dual-channel metamaterial fabric sensor system in an office environment according to an embodiment;

[0076] Fig.31 shows normalized outputs from both channels CH1 and CH2, and graphs of detected vital signs including heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP), and diastolic blood pressure (DBP) according to an embodiment;

[0077] Fig.32A and Fig.32B A graph showing a linear regression for estimating blood pressure using calculated PTT values ​​according to an embodiment is shown, wherein Fig.32A shows a graph of linear regression for estimating systolic blood pressure, and Fig.32B A graph showing linear regression for estimating diastolic blood pressure is shown;

[0078] Fig.33A and Fig.33B A graph showing a candidate period for selecting continuous blood pressure detection according to an embodiment is shown, wherein Fig.33A shows a graph for identifying candidate epochs for blood pressure sensing using the absolute difference between the mean heart rate (HR) and the mean pulse rate (PR) for each epoch, and Fig.33B The use of signal quality metrics (Q e )choose Fig.33A a graph of candidate periods for continuous blood pressure monitoring identified in;

[0079] Fig.34 shows long-term health monitoring data of normalized outputs from both channels CH1 and CH2, and graphs of detected vital signs including heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP), and diastolic blood pressure (DBP) according to an embodiment;

[0080] Fig.35 A photograph showing a passenger health monitoring system including a metamaterial fabric sensor according to an embodiment;

[0081] Fig.36 The use of the embodiment is shown Fig.35 A graph of normalized output and detected heart rate measured by a passenger health monitoring system, and a ... a graph of your reference heart rate as measured by the Watch; and

[0082] Fig.37 is a diagram showing a method according to an embodiment of the present invention. Figure 1A Block diagram of the technical architecture of a computer. DETAILED DESCRIPTION

[0083] Exemplary embodiments are described that relate to sensors, systems, and methods for contactless sensing of physiological parameters of the body.

[0084] Seamlessly embedded contactless sensors in ordinary physical spaces can incorporate intelligent sensing capabilities into otherwise passive environments. Such environmental sensing capabilities have great potential to enable preventive and personalized healthcare by unobtrusively collecting longitudinal health data during daily activities. Although microwave radar has been used as a contactless physiological sensor, existing implementations are not suitable for long-term environmental sensing due to challenges associated with sporadic interference in complex real-world environments. In the present disclosure, contactless sensing using near-field electromagnetic interactions with the human body is described. The sensors, systems, and methods of the present disclosure can be used for unobtrusive, long-term health tracking that is not affected by distant interference sources in the background. In an embodiment, this provides multi-point sensing of physiological signals in the surrounding environment.

[0085] Figure 1A and Figure 1B A diagram relating to a system 100 for contactless sensing of a physiological parameter of a body 102 is shown according to an embodiment.

[0086] Figure 1A A schematic diagram of a system 100 for contactless sensing of a physiological parameter of a body 102 is shown. The system 100 includes a sensor 104, a software defined radio (SDR) system 106, and a computer 108.

[0087] The sensor 104 includes a waveguide. The waveguide includes a metamaterial and is configured to propagate a transmit signal 110 from the SDR system 106 along the waveguide in an artificial surface plasmon mode to provide a receive signal 112. The transmit signal 110 propagates along the waveguide in an artificial surface plasmon mode to generate an evanescent electromagnetic field. The evanescent electromagnetic field is used for non-contact sensing of disturbances generated by physiological movements of the body 102, wherein the sensor 104 (or waveguide) is placed at a predetermined distance 114 away from the body 102. The predetermined distance 114 can have a range of 1 to 15 millimeters (mm). The disturbance sensed by the evanescent electromagnetic field produces a phase shift between the transmit signal 110 and the receive signal 112, which can be processed for determining a physiological parameter of the body 102.

[0088] The SDR system 106 is adapted to provide a transmit signal 110 to the sensor 104, and to receive a receive signal 112 from the sensor 104. In the present embodiment, the SDR system 106 is adapted to send the transmit signal 110 at an entry end of a waveguide of the sensor 104, wherein the transmit signal 110 propagates along the waveguide, and is adapted to receive the receive signal 112 at an exit end of the waveguide of the sensor 104. In the present embodiment, the entry end of the waveguide and the exit end of the waveguide are at different ends or ports of the waveguide, but it should be understood that in other embodiments, the entry end and the exit end may also be at the same port of the waveguide.

[0089] Once the SDR system 106 receives the received signal 112, the transmitted signal 110 and the received signal 112 may be processed to determine the phase shift caused by the physiological movement of the body 102. Information or signals related to the determined phase shift may then be used to determine a physiological parameter of the body 102 associated with the physiological movement. Although the SDR system 106 is used in the present embodiment to provide the transmitted signal 110 and to receive the received signal 112, the skilled person will appreciate that this should not be construed as limiting, and other suitable signal systems (e.g., an integrated on-chip radio frequency transceiver) may be used, as long as the phase shift caused by the physiological movement of the body 102 can be determined.

[0090] The computer 108 includes a processor and a data storage storing computer program instructions operable to cause the processor to perform various processing on the time-varying phase signal 116 received from the SDR system 106. For example, in the case where the physiological movement of the body 102 is associated with more than one physiological parameter, the computer 108 may be adapted to process the time-varying phase signal with a bandpass filter to separate the time-varying phase signal into separate components associated with each of the more than one physiological parameter. The computer 108 may be further configured to pre-process the time-varying phase signal 116, calculate the respiratory cycle, and calculate the beat-to-beat interval for obtaining the heart rate and radial artery pulse rate using the received time-varying phase signal 116. In an embodiment, the data storage of the computer 108 stores a trained machine learning model to generate an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal for calculating the pulse transit time (PPT), and converting the calculated PPT into systolic and diastolic pressure values. Other processes that may be performed by the computer 108 are described in conjunction with subsequent figures.

[0091] In an embodiment, the computer 108 may be used to provide instructions 118 to the SDR system 106, for example, regarding controlling parameters associated with the transmit signal 110 and / or controlling processing parameters of the receive signal 112. In another embodiment, the SDR system 106 may operate independently of the computer 108. For example, instructions for waveform generation and signal processing may be programmed into the SDR system 106 as an embedded application. In this case, the SDR system 106 is used to provide the time-varying phase signal 116 to the computer 108 for processing.

[0092] Despite Figure 1A Only one sensor 104 is shown in the system 100, but it should be understood that more than one sensor 104 may be operably connected to the SDR system 106 and the computer 108. For example, as described with respect to Fig. 22 As described, the sensor may be placed on a table for contactless sensing of the radial artery pulse from the wrist of the body 102, and the sensor may be placed on a chair for contactless sensing of the heart pulse and respiratory rate of the body 102. In addition, the SDR system 106 and the computer 108 are arranged in Figure 1A 104, it will be appreciated that in embodiments, the SDR system 106 may be part of the computer 108, or the SDR system and computer 104 may be integrated to form a single processing unit. For example, a dedicated device may be constructed to integrate the RF functionality of the SDR system 106 with the computing functionality required to operate it.

[0093] Figure 1B A diagram 120 is shown to illustrate the Figure 1A The sensor 104 is for non-contact sensing.

[0094] exist Figure 1B The principles of ambient near-field health sensing using sensor 104 are described in . In this embodiment, sensor 104 is a metamaterial fabric sensor that is designed to propagate electromagnetic surface waves 122 that extend evanescently in the normal direction of its surface in an artificial surface plasmon (SSP) mode. Due to the non-radiative nature of the surface waves 122, sensor 104 is not affected by distant interference sources in the environment (such as background clutter or the movement of other people) while facilitating sensitive close-range interaction with one or more parts of the body 102. In particular, by shortening the surface plasmon wavelength (λ) relative to the free space wavelength (λ0), the sensor 104 can sense the surrounding environment and sense the surrounding environment. sp) and confinement of electromagnetic energy on the sensing surface to enhance sensitivity to subtle physiological motions or signals 124, which can strongly couple with heterogeneous tissue layers of the body 102 and respond to cardiopulmonary mediated impedance modulation. In particular, the non-radiative and localized nature of SSP surface waves provides immunity to distant interference sources. Subwavelength energy confinement on the sensor results in enhanced coupling with tissue. The shortened surface plasmon wavelength (λ sp <<λ D ) leads to amplification of Doppler phase changes caused by cardiopulmonary-mediated impedance modulation.

[0095] Figure 2 A flow chart showing the steps of a method 200 for contactless sensing of a physiological parameter of a body 102 using the system 100 of Fig. 1, according to an embodiment. The method uses the sensor 104 as described above.

[0096] In step 202, the sensor 104 is placed at a predetermined distance 114 away from the body 102 for contactless sensing. The predetermined distance may be a range of suitable distances away from the body 102 (eg, 1 mm to 15 mm), so it provides some flexibility. The body 102 is preferably stationary during the measurement.

[0097] In step 204, the transmit signal 110 is provided to the sensor 104. In the present embodiment, the transmit signal 110 is provided by the SDR system 106.

[0098] In step 206, a receive signal 112 is received from the sensor 104 after the transmit signal 110 has propagated through the waveguide of the sensor 104. In the present embodiment, the receive signal 112 is received by the SDR system 106.

[0099] In step 208, the received signal 112 and the transmitted signal 110 are processed to determine a phase shift between the transmitted signal 110 and the received signal 112, the phase shift being caused by a disturbance generated by the physiological movement of the body 102. The determined phase shift can then be used or processed for determining a physiological parameter of the body 102. In particular, in an embodiment, the SDR system 106 is configured to: generate a digital complex baseband signal, convert the digital complex baseband signal using a digital-to-analog converter (DAC) to form an analog baseband signal, modulate the analog baseband signal with a carrier signal to provide a transmit signal, demodulate the received signal to obtain in-phase and quadrature (IQ) components associated with the digital complex baseband signal, and digitize the obtained IQ components. The SDR system 106 is then configured to perform a complex conjugate multiplication of the digital complex baseband signal and the digitized IQ component to determine a phase shift signal associated with the phase shift between the transmitted signal and the received signal. The SDR system 106 is adapted to demodulate and unwrap the phase shift signal inverse tangent to obtain a time-varying phase signal associated with the phase shift between the transmitted signal and the received signal. The time-varying phase signal obtained from the SDR system 106 is then processed by the computer 108 to obtain, for example, the respiratory period, the beat-to-beat interval associated with the heart rate, and / or the beat-to-beat interval associated with the radial artery pulse rate. This will be discussed further in conjunction with subsequent figures.

[0100] In an embodiment of the present disclosure, thin, flexible waveguides that can support artificial surface plasmons (SSPs) at radio frequencies are developed using metamaterials. Waveguides formed using conductive fabric metamaterials can be integrated on clothing, which can significantly improve the transmission efficiency of wireless signals around the body (e.g., human body). These waveguides utilize evanescent fields associated with SSP modes to wirelessly interact with nearby wearable devices or the body for contactless sensing of physiological parameters or vital signs of the body.

[0101] A multiplexed contactless health monitoring system (or vital signs monitoring system) is developed by integrating sensors with artificial surface plasmon (SSP) metamaterial waveguides with software-defined radar technology. In this embodiment, the SSP metamaterial is used to provide a spatially selective sensing modality for cardiopulmonary exercise. The SSP metamaterial waveguide can be integrated with different household furniture or clothing, enabling simultaneous monitoring of physiological parameters (such as breathing rate and heart rate) in the context of daily life. The metamaterial sensor is thin and soft, and is able to monitor health through clothing, accessories or furniture.

[0102] FIG. 3A to FIG. 4B as well as 7A to 8C Concerning a first embodiment of the sensor 104 used in the system 100, the sensor 104 is placed on a chair at the back of the body for non-contact physiological parameter sensing. Figure 5 and Figure 6Respectively, they relate to an SDR system 106 and a method for detecting a heartbeat or a radial artery pulse, and are also applicable to Figures 9 to 36 Subsequent embodiments described.

[0103] Figure 3A and Figure 3B Figures 300, 310 show a portion of the system 100 of Figure 1 according to an embodiment. Figure 1A As depicted, system 100 includes a SSP waveguide sensor 104, and a software defined radio (SDR) system 106 for radar signal conduction and acquisition. Figure 3A An image 300 is shown, which illustrates that the SSP metamaterial waveguide 302 is connected to the SDR system 106 via coaxial cables 304, 306 (note that the SDR system 106 is not in Figure 3A ). The SSP waveguide 302 is designed to support a surface plasmon-like propagation mode at an operating frequency of 2.4 GHz, thereby guiding the transmit signal from the transmit port of the SDR system 106 to the receive port. Figure 3A As shown, a transmit signal Tx is received at one end of the waveguide 302, and a receive signal Rx is provided at the opposite end of the waveguide 302. In contrast to radiating elements, SSP metamaterials exhibit tight spatial wave confinement and high local field strength in the surrounding space. Therefore, a small perturbation near the SPP waveguide 302 will result in a disproportionate change in the electromagnetic field at that location. Such a change or perturbation in the local electromagnetic field (or evanescent electromagnetic field) can be captured as a phase shift (or Doppler phase shift) between the transmit signal Tx and the receive signal Rx, making it highly sensitive to small physiological motions (such as those caused by the cardiopulmonary action of the body). In this way, it is possible to obtain a signal from, for example, Figure 3B Physiological parameters or vital signs, such as respiratory rate and / or heart rate, are obtained from the back of the body in the seated position shown in image 310 .

[0104] In order to implement the system 100 of the present embodiment, a suitable waveguide for use with the sensor 104 is designed and analyzed. The waveguide 402 of the present embodiment includes a metamaterial and has a structure including a planar comb pattern as a top layer 404, an intermediate non-conductive layer (e.g., a fabric layer) 406, and an unpatterned bottom layer 408. Figure 4A. The top layer 404 serves as a sensing layer, and it includes a comb-like rectangular strip having an elongated base 410 and a plurality of teeth 412 extending along and from the elongated base 410, wherein adjacent teeth of the plurality of teeth 412 are separated by gaps. The unpatterned bottom layer 408 serves as a ground layer or ground plane for confining microwave energy only to the sensing side of the waveguide 402 (i.e., one side of the top layer or sensing layer). The parameters of the structure of the waveguide 402 are optimized by calculating the dispersion curve using an eigenmode solver (in this embodiment, CST Microwave Studio). In this embodiment, the geometric parameters of the structure are optimized to have: the height of the teeth h=18mm; the width of the rectangular strip of the sensing layer b=22mm; the width of the teeth a=6mm; and the length of the period (i.e., the distance between adjacent teeth) p=8mm (i.e., the gap (i.e., pa) between two adjacent teeth is 2mm).

[0105] Figure 4A Also shown are simulations without a human torso or human body (i.e., curve 414) and with a human torso or human body (i.e., curve 416). 21 | The curves 414 and 416 are obtained using the straight SSP waveguide design. Figure 4A The simulation S shown 21 The comparison indicates the coupling of electromagnetic energy from the waveguide 402 to the human body, as observed in the transmission coefficient |S at the operating frequency band. 21 In this embodiment, the sensing layer 404 and the grounding layer 408 of the SSP waveguide 402 are manufactured by a commercial printed circuit board (PCB) process using 35 μm thick copper on a flexible polyimide (PI) substrate.

[0106] To illustrate the interaction between the SSP waveguide 402 and the body, a full-wave simulation of the SSP waveguide 402 placed on a voxel model of the human body (Donna, CST Microwave Studio) was performed, which includes the tissue heterogeneity of the body. A radio frequency (RF) signal was launched from the SDR system to the SSP waveguide 402 and propagated along the SSP waveguide 402 as a surface wave.

[0107] Figure 4B A graph 420 is shown illustrating the electric field in a portion of a waveguide 422 in the xz plane when the waveguide 422 is interfaced with a transmit signal from an SDR system to illustrate the interaction between the electric field and a human body 424 for sensing. Figure 4B The white arrow 426 indicates the wave propagation direction of the transmitted signal. Figure 4BAs shown in FIG. 4 , most of the RF energy is coupled into the human body 424 before being provided as a receive signal at the exit end of the waveguide and received at the receiver (Rx) port of the SDR system. The coupling amount can be given by P tissue =P in Description, where P tissue is the power loss in biological tissue, P in is the input power of the SSP waveguide 422. In this embodiment, for a transmission signal frequency of 2.4 GHz, the coupling amount is estimated to be 70.4%.

[0108] To demonstrate the use of the system 100 described above for sensing one or more physiological parameters, the SSP waveguide was interfaced to a USRP B210 SDR system (Ettus Research, National Instruments) via a coaxial cable.

[0109] Figure 5 A schematic diagram showing integration of a fabric sensor with a software defined radio (SDR) system 500 according to an embodiment is shown.

[0110] like Figure 5 As shown, the fabric sensor 502 is configured as a 2-port device, wherein each port or terminal of the sensor 502 is connected to the T of the SDR (Ettus B210 USRP, National Instruments) 500 using a coaxial cable. X Channel 504 and R X Channel 506.

[0111] Used as the transmitting signal f RF The software-defined continuous wave signal is configured by: (i) using the SDR system 500 to generate a digital composite baseband signal, the digital composite baseband signal passes through the built-in digital-to-analog converter (DAC) 508 to form an analog baseband signal f0, and (ii) comparing the analog baseband signal f0 with the 2.4 GHz carrier frequency f LO For mixing or modulation, the carrier frequency f LO is generated by the RF front end 510 of the SDR system 500. Therefore, the RF front end 510 is used to transmit the signal f from the f0=1 MHz baseband signal pair. RF (2.4 GHz) is up-converted and conformally guided through the fabric sensor 502. The transmit (Tx) signal (i.e., f RF ) is guided along the SSP waveguide of the fabric sensor 502, interacts with the body 512, and is collected back at the SDR system 500 as a receive (Rx) signal (i.e., f RF +f D). The interaction with the body 512 causes Doppler phase changes, which are captured as f mixed in the returning RF signal (i.e., the receive (Rx) signal). D At the receive port, the Rx signal is quadrature demodulated to baseband by the SDR's analog RF front end 510, and the resulting in-phase and quadrature (IQ) signals are digitized at a 1 MHz sampling rate. In this embodiment, the SDR system 500 includes a built-in analog-to-digital converter (ADC) 514 after the radio frequency (RF) analog front end 510, which digitizes the IQ signal after demodulation. The phase shift (or Doppler phase shift, i.e., f) in the digital domain caused by physiological movement D Component), obtained as a phase shifted signal by multiplying 516 the baseband signal f0 with the complex conjugate of the Rx signal, followed by low pass filtering 518 and decimation to 400 Hz. The first low pass filtering step and down sampling step are used to remove unwanted high frequency components in the signals received by the SDR system 106, and to record data at a lower sampling rate to reduce the file size associated with these signals. The baseband Rx signal is a low frequency component down converted from a 2.4 GHz carrier frequency signal. The Doppler phase change φ d is obtained by inverse tangent demodulation 520 and then unwrapped 522 to produce the sensor output. This is shown in relation to equations (1) and (2) below.

[0112] The digital composite Doppler phase shift signal is obtained by multiplying the baseband Tx signal by the conjugate of the down-converted Rx signal:

[0113]

[0114] in, represents the captured Doppler shift caused by physiological motion. In this embodiment, the SDR is programmed to transmit the signal at a carrier frequency f RF = 2.4 GHz transmit (Tx) and receive (Rx) continuous wave (CW) signals, where the baseband tone is set to f sampled at 1 MHz BB =10k Hz. At 10dBm Tx power, the simulated whole-body average specific absorption rate (SAR) is 0.019W / kg, which is much lower than the IEEE C.95.1-2019 limit of 2W / kg.

[0115] Then, the time-varying Doppler phase shift φ is obtained by arc tangent demodulation and unwrapping the composite Doppler signal using the following D (t):

[0116]

[0117] In this embodiment, the outlined SDR system 500 is implemented using the open source software GNU Radio on a personal computer. The personal computer is used to power and communicate with the SDR system via a universal serial bus (USB). Control of the SDR system 500 is performed using the GNU Radio software.

[0118] use Figure 3A In the setup shown in , the vital signs or physiological parameters of human subjects were measured to verify the sensing capabilities of the system. The fully clothed subjects were ordered to sit in an office chair configured (see e.g. Figure 3B ) and gently lean against the backrest. The raw Doppler phase shift signal is recorded by the GNU Radio software using the above configuration. Since the Doppler phase information contains the superposition of both breathing and heartbeat, in this embodiment, MATLAB on the computer 108 with a bandpass filter is used to process the recorded time-varying phase shift signal to generate a breathing signal (or a time-varying breathing phase signal) (e.g., having a frequency range of 0.1 to 0.8 Hz) or a heartbeat signal (or a time-varying cardiac phase signal) (e.g., having a frequency range of 0.9 to 5 Hz). In an embodiment where the second sensor is used to detect a radial artery pulse signal (see, for example, below regarding Fig. 22 ), the computer 108 is adapted to filter the time-varying phase shift signal received from the second sensor using a bandpass filter (e.g., having a frequency range of 0.9 to 5 Hz) to obtain a time-varying radial artery pulse phase signal. The accuracy of the present system for heartbeat sensing is verified by using the present system to simultaneously measure heart rate and measure a reference electrocardiogram (ECG, electrocardiogram) signal (MP46, BIOPAC Systems).

[0119] Regarding the detection of the respiratory cycle, the respiratory peak detection algorithm is implemented based on the moving-average curve (MAC) intercept of the sensor data and the BIOPAC reference data. As described above, in this embodiment, the sensor output (the instantaneous phase shift signal φ D (t)) is first bandpass filtered between 0.1 Hz and 0.8 Hz to obtain a time-varying phase shift signal. A Fast Fourier Transform (FFT) is applied to the first 15 seconds of each time-varying respiratory phase signal segment to estimate the respiratory period T est . At 2T estThe MAC at each point is calculated by taking the mean of the time-varying respiratory phase signal over a time window to generate each data point of the MAC. The intercept between the moving average curve and the time-varying respiratory phase signal is calculated and labeled as an upper intercept or a lower intercept based on the slope of the signal at that point. The peak is identified as the maximum value between an intercept with a positive slope and a subsequent intercept with a negative slope, and the respiratory cycle is calculated as the duration between two adjacent peaks.

[0120] Figure 6 A flow chart of a method 600 for processing physiological signals to detect heartbeats and pulse beats according to an embodiment is shown.

[0121] In order to extract the heartbeat or pulse, a zero-crossing detection algorithm with an adaptive threshold is implemented based on the Pan-Tompkins algorithm. As described above, in step 602, the sensor output related to the heart or pulse waveform (i.e., the time-varying phase shift signal φ) is firstly D (t)) is bandpass filtered from 0.9 Hz to 5 Hz to remove direct current (DC) drift, respiration and high frequency noise, and a time-varying phase signal or a time-varying radial artery pulse phase signal is obtained, respectively. A first time derivative waveform is then calculated using a derivative filter for each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in step 604. The first time derivative waveform is then half-wave rectified (i.e., all negative values ​​are set to zero) in step 606, and the resulting waveform is squared in step 608 to minimize small noise peaks. Then, in step 610, the squared signal is passed through a moving average filter with a 150 millisecond (ms) window to produce an integrated signal used in subsequent beat detection. A peak in the integrated signal is detected in step 612, and in the present embodiment, an adaptive thresholding and decision rule according to the original Pan-Tompkins algorithm is used in step 614 to verify the peak in the integrated signal. Similarly, peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal from step 602 are detected in step 616 and, in this embodiment, validated in step 618 using adaptive thresholding and decision rules according to the original Pan-Tompkins algorithm. The detected peaks in the integrated signal are used to validate the detected peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in step 620. Beat positions in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal are then calculated in step 622, where each of the beat positions is the nearest preceding positive zero intercept relative to each validated peak of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal (i.e., the output from step 602).

[0122] A beat-to-beat interval associated with each of the time-varying heart phase signal and the time-varying radial artery pulse phase signal (eg, associated with the heartbeat waveform or the radial artery pulse waveform) may then be calculated as the time interval between consecutive beat positions.

[0123] Fig. 7A and Figure 7B Graphs 700, 710 of measured vital sign waveforms according to an embodiment are shown, wherein Fig. 7A A graph 700 of a breathing signal is shown, and Figure 7B A graph 710 of a heartbeat signal with detected beat positions (shown as circles 712) is shown. A normalized respiratory signal (or a normalized time-varying respiratory phase signal) (e.g. Fig. 7A ) and a normalized heartbeat signal (or a normalized time-varying cardiac phase signal) (as shown Figure 7B ) are obtained simultaneously from the same recording or measurement using waveguide 302. The heartbeat positions detected using the aforementioned method 600 are shown on the filtered signal as Figure 7B Circle 712 in.

[0124] Fig. 8A A graph 800 is shown of the measured beat-to-beat interval (BBI) obtained from the SSP waveguide 302 of the sensor 104. This is compared in graph 800 to a reference electrocardiogram (ECG) R-peak (RR) interval in milliseconds for 125 heart beats recorded using 120 seconds of data. Fig. 8A Good agreement between the BBI obtained using the present system 100 and the reference data is shown.

[0125] Figure 8B Graph 810 shows the linear correlation between the BBI measured using the SSP sensor 104 and the interval measured by the ECG. Figure 8B As shown, BBI measurements showed a strong positive linear correlation with the reference ECG RR intervals, with a Pearson correlation coefficient of r = 0.8929 (P < 0.001).

[0126] Figure 8C A Bland-Altman plot 820 is shown showing a pairwise comparison between the ECG RR intervals and the SSP BBI, showing a mean μ = -0.0403 ms and a standard deviation σ = 1.8893 ms of the differences. Figure 8CThe Bland-Altman analysis in FIG. 820 shows good agreement between the present system 100 and the ECG sensor, as evidenced by a small deviation (μ = -0.0403 ms) and more than 95% of the data points being within the limits of agreement (LoA = [μ - 1.96σ; μ + 1.96σ] = [-37.0634; 36.9828] ms). The Bland-Altman plot 820 shown relates to data obtained from 1 of the 10 subjects. Fig. 24B Bland-Altman plots for 10 subjects are shown.

[0127] The following Figures 9 to 32B Another example of using the system 100 is provided in which two sensors 104 are used to track physiological signals from different parts of the body 102 .

[0128] Fig. 9 An example of using the embodiment is shown. Figure 1A FIG900 illustrates multi-point tracking of physiological signals from different body parts by system 100 .

[0129] Multi-point tracking of physiological signals from different body parts is achieved through the local sensing characteristics of the SSP sensor. In this embodiment, the first SSP sensor 902 is provided at or near the back of the body, and the second SSP sensor 904 is provided at or near the wrist of the body. Being thin and soft, the SSP sensors 902, 904 can be easily integrated into everyday objects for simultaneously monitoring multiple local vital signs on the body, such as breathing and heartbeat signals from the back and radial artery pulse signals from the wrist, such as Fig. 9 shown.

[0130] Similar to the previous embodiments, the present embodiment involving multi-point tracking of physiological signals includes SSP sensors 902, 904 interfaced with an SDR system 906 programmed to generate a signal at a frequency f RF =2.4 GHz transmits a continuous wave (CW) transmit (Tx) signal that propagates conformally along each of the SSP sensors 902, 904 as an SSP surface wave, allowing it to interact strongly with biological tissue at close range. Therefore, a small perturbation near the sensing surface of the SSP sensors 902, 904 due to cardiopulmonary activity will result in a large corresponding change in the electromagnetic field at that location, which is captured as a receive (Rx) signal (i.e., f) at the SDR system 906. RF +f D). This provides local sensing capabilities for these SSP sensors 902, 904, which allows the SSP sensors 902, 904 to obtain physiological information from smaller body parts (such as the radial artery), in contrast to conventional radar sensors that detect free space radiation and are therefore subject to background noise over a large space. Fig. 10A and 10B Shown in.

[0131] Fig. 10A and Fig. 10B 1000, 1010 show full wave simulations of electric fields at 2.4 GHz for sensing the radial artery, according to an embodiment. Fig. 10A As shown, conventional radiation sensing involves detecting free-space radiation over a large area or space 1002. In contrast, Fig. 10B As shown, disturbance sensing using SSP surface waves of the present invention involves non-radiative sensing over a local area / space 1012 .

[0132] Return to reference Fig. 9 , processing of the received (Rx) signal is performed at the SDR system 906 (the details are discussed above and therefore will not be repeated here for the sake of brevity) to obtain the time-varying Doppler phase shift φ at each of channel 1 and channel 2 D The time-varying Doppler phase shift φ is calculated using MATLAB (R2021a, The MathWorks) on computer 108. D The (t) signal is band-pass filtered to obtain a respiratory signal (also referred to as a time-varying respiratory phase signal in the present disclosure) (e.g., having a frequency range of 0.1 to 0.8 Hz) and a heartbeat signal (also referred to as a time-varying cardiac phase signal in the present disclosure) (e.g., having a frequency range of 0.9 to 5 Hz) at channel 1 and a radial artery pulse signal (also referred to as a time-varying radial artery pulse phase signal in the present disclosure) (e.g., having a frequency range of 0.9 to 5 Hz) at channel 2. Fig. 9 908. The reference heart beat information and pulse beat information are directly obtained from the BIOPAC Student Lab software.

[0133] The heartbeat signal and radial artery pulse signal can be further processed at computer 108 using a machine learning model to obtain a blood pressure value. Fig. 9 910 and will be described below with respect to Figure 25 to Figure 28 Further detailed discussion.

[0134] Fig.11An illustration 1100 of an arm 1102 placed over a metamaterial fabric waveguide 1104 is shown in accordance with an embodiment. The metamaterial fabric waveguide 1104 includes a patterned conductive top layer 1106, an intermediate non-conductive layer 1108, and a conductive bottom isolation layer 1110. Similar to the previous embodiments, the patterned conductive top layer 1106 serves as a sensing layer and includes a comb-like rectangular strip having an elongated base and a plurality of teeth extending along and from the elongated base, wherein adjacent teeth of the plurality of teeth are separated by gaps. The conductive bottom isolation layer 1110 acts as a ground layer or plane for limiting microwave energy to only the sensing side of the waveguide 1104 (i.e., one side of the top layer or sensing layer).

[0135] In this embodiment, to make the fabric waveguide 1104, a commercially available conductive fabric (Holland Shielding Systems) was patterned by laser cutting (Universal Laser Systems, VLS2.30) and attached to a polyester fabric sheet. An SMA connector was attached by adhesive coating using a conductive epoxy (CW 2460, Chemtronics) at room temperature. Compared to the waveguide in the previous embodiment as shown with respect to FIG. 4 , the fabric waveguide 1104 in this embodiment does not require a substrate because the conductive fabric / fabric used in the patterned conductive top layer 1106 and the conductive bottom isolation layer 1110 can be directly attached to the polyester fabric sheet forming the middle non-conductive layer 1108.

[0136] CST Microwave Studio (Dassault Systems) is used for FIG. 12A to FIG. 20B Results of electromagnetic simulations are shown.

[0137] Fig. 12A and Fig. 12B Shows about Fig.11 Graphs 1200 , 1210 of a geometric parameter h of a metamaterial fabric waveguide 1104 , where the parameter h is related to the height of the teeth of the patterned conductive top layer 1106 .

[0138] Fig. 12A A graph 1200 of the plasma wave number βp as a function of the geometric parameter h for different fabric substrates used as the middle non-conductive layer 1108 is shown. The three different fabric substrates used in this embodiment include cotton fabric, polyester fabric, and canvas fabric, each of which has ε 棉 , ε 聚酯纤维 and ε 帆布 The dielectric constant of the base fabric is designated as ε 棉 =1.6,ε 聚酯纤维 =1.9,ε 帆布= 2.1. Curve 1202 shows cotton fabric, curve 1204 shows polyester fabric, and curve 1206 shows canvas fabric.

[0139] SSP modes with different degrees of wavelength confinement (from 0.25πrad cm -1 to 0.85πrad cm -1 The quantization of the plasma wave number βp in the range can be achieved by tuning the geometric parameter h of the structure and using different fabric materials as the intermediate layer, such as Fig. 12A To enhance sensitivity, a high degree of wavelength confinement (i.e., a large βp value) is desired, which can be achieved with a larger h.

[0140] To illustrate this in relation to the geometric parameter h, Fig.13 1 shows a dispersion graph 1300 of a metamaterial fabric waveguide having different geometric parameters h according to one embodiment. The dispersion graph 1300 includes a dispersion curve 1302 of a metamaterial fabric waveguide having a parameter h=16 mm, a dispersion curve 1304 of a metamaterial fabric waveguide having a parameter h=18 mm, and a dispersion curve 1306 of a metamaterial fabric waveguide having a parameter h=20 mm. It is clear from the graph 1300 that for larger h, there is less dispersion (e.g., comparing the dispersion curve 1302 of h=16 mm to the dispersion curve 1306 of h=20 mm).

[0141] Return to reference Fig. 12A A high degree of wavelength confinement (ie, a large βp value) can also be achieved by selecting a suitable fabric material for the intermediate non-conductive layer 1108.

[0142] However, there is a design trade-off between wavelength limitation and transmission loss, as increasing h results in higher transmission loss. Fig. 12B Shown in.

[0143] Fig. 12B A graph 1210 of transmission loss at 2.4 GHz as a function of fabric conductivity for different geometric parameters h is shown. The transmission loss for h = 16 mm is shown with respect to curve 1212, the transmission loss for h = 18 mm is shown with respect to curve 1214, and the transmission loss for h = 20 mm is shown with respect to curve 1216. As shown in graph 1210, the larger the value of the geometric parameter h, the higher the transmission loss for a given conductivity. Since high losses are detrimental to the function of the sensor, a conductive fabric (σ≈5×10 5 Sm -1 ) When manufacturing the sensor, h = 18 cm was selected to keep the -1 transmission loss.

[0144] To improve the performance of the metamaterial fabric waveguide, a planar matching section using gradient corrugated strips with a linear taper was designed to achieve negligible surface-to-propagation conversion losses from a 50Ω port of the RF front end 510 of the SDR system 106 .

[0145] Fig.14A and Fig. 14B A schematic diagram 1400 and a graph 1410 are shown to illustrate the design of a planar matching section of a metamaterial fabric waveguide according to an embodiment.

[0146] Fig.14A A diagram 1400 is shown to illustrate the structure of a planar matching section 1402 at the end of a metamaterial waveguide. In this embodiment, three matching units are used (ie, N=3).

[0147] Fig. 14B A graph 1410 of the transmission coefficient as a function of the geometric parameter h is shown for different numbers of matching units N. The transmission coefficient as a function of h for N=0 is shown with respect to curve 1412, the transmission coefficient as a function of h for N=1 is shown with respect to curve 1414, the transmission coefficient as a function of h for N=2 is shown with respect to curve 1416, the transmission coefficient as a function of h for N=3 is shown with respect to curve 1418, and the transmission coefficient as a function of h for N=4 is shown with respect to curve 1420. Fig. 14B As shown, N=3 provides the highest transmission and is therefore selected as the number of matching units used.

[0148] Fig.15A and Fig. 15B Diagrams 1500 , 1510 showing full wave simulation during sensing according to an embodiment. Fig.15A A diagram 1500 showing a full wave simulation of a cross section of an arm 1502 during radial artery pulse sensing, and Fig. 15B A graph 1510 showing a full wave simulation of a cross section of a human torso 1512 during heartbeat sensing or respiration sensing is shown.

[0149] Computational arm and body voxel models are used to simulate the electromagnetic field distribution. The resolution of the voxel is 2mm×2mm×2mm. The radial artery pulse and the cardiac cycle of the heartbeat are simulated by periodic geometric changes on the voxel models of the radial artery and the heart. The simulated tissue motion leads to a change in the effective dielectric constant, which is demodulated as a phase change Δφ of the scattering parameter S21.

[0150] As about Fig.15A and 15BAs shown in the full-wave simulation of , the metamaterial fabric sensor of the present embodiment allows electromagnetic energy to be confined on its surface to interact with nearby computing arm 1502 and torso 1512 models.

[0151] Fig.16 Schematic diagram 1600 showing the electric field distribution in the yz plane in the top plate 1602 and the xz plane in the bottom plate 1604 of a conventional microstrip and a metamaterial fabric sensor of the present disclosure according to an embodiment. The white dashed line 1606 shows the attenuation length 3α -1 .

[0152] SSP modes can propagate conformally on the surfaces of everyday objects and interact strongly with the radial artery and heart via the evanescent field, enabling unobtrusive wrist pulse and heartbeat detection. In contrast, conventional waveguides (such as microstrips) lack this evanescent field, resulting in less efficient coupling to tissue. This is consistent with the Fig.16 It is shown that the attenuation length of the microstrip is 3α -1 In addition, the high energy confinement and shortened wavelength of the SSP mode of the metamaterial waveguide significantly enhances the conversion of subtle tissue impedance modulation into larger phase changes of the propagating electromagnetic wave.

[0153] Fig.17A and Fig. 17B shows simulation |S without interaction with the body and due to interaction with different body parts according to an embodiment 21 | Graphs of transmission spectra 1700, 1710. This also provides evidence of lower coupling efficiency to tissue between conventional microstrip fabric sensors compared to the SSP metamaterial sensors of the present disclosure.

[0154] Fig.17A shows a simulation of an SSP metamaterial fabric sensor including a waveguide 1104 |S 21 |Transmission spectrum graph 1700. Graph 1702 shows the simulated |S without interaction with the body (ie, free space). 21 | Transmission spectrum, graph 1704 shows the simulation of the wrist interaction with the body |S 21 | transmission spectrum, and curve 1706 shows the simulated |S in the case of interaction with the back of the body 21 | Transmission spectrum. Simulation of graph 1700 obtained using a U-shaped SSP metamaterial fabric sensor |S 21 | transmission spectrum, which is similar to Figure 4A Different results were obtained using the straight SSP metamaterial fabric sensor.

[0155] Fig. 17Bshows the simulation of microstrip line (MSL) fabric sensor |S 21 |Transmission spectrum graph 1710. Graph 1712 shows the simulated |S without interaction with the body (ie, free space). 21 | Transmission spectrum, graph 1714 shows the simulation of the wrist interaction with the body |S 21 | transmission spectrum, and curve 1716 shows the simulated |S in the case of interaction with the back of the wrist of the body 21 |Transmission spectrum.

[0156] from Fig.17A and Fig. 17B It can be seen that the coupling efficiency between the SSP metamaterial fabric sensor and body tissue is higher than that between the MSL fabric sensor. In the presence of body tissue, the SSP metamaterial fabric sensor is more sensitive to the body tissue than the MSL fabric sensor. 21 | Larger changes in the transmission spectrum.

[0157] To evaluate the enhancement of phase modulation, the responses of the SSP metamaterial fabric sensor and the microstrip sensor to pulsatile cardiac motion were compared, which was simulated as geometric changes of the radial artery and heart in a computational voxel model.

[0158] Fig.18A and Fig.18B Graphs 1800 , 1810 of simulated phase changes ΔΦ obtained from a conventional microstrip fabric sensor and a metamaterial fabric sensor of the present disclosure are shown, according to an embodiment.

[0159] Fig.18A Graph 1800 is shown of simulated phase changes ΔΦ as a result of the radial artery pulse during a cardiac cycle. Fig.18A Also shown is an inset 1802 of an arm-torso simulation model. Curve 1804 shows simulated phase changes ΔΦ as a result of the radial artery pulse during the cardiac cycle for the SSP metamaterial fabric sensor, while curve 1806 shows simulated phase changes ΔΦ as a result of the radial artery pulse during the cardiac cycle for the MSL fabric sensor.

[0160] Fig.18B A graph 1810 of simulated phase changes ΔΦ as a result of heart beats within a cardiac cycle is shown. Fig.18B Also shown is an inset 1812 of a simulated model of a human torso. Graph 1814 shows simulated phase changes ΔΦ as a result of a heartbeat during a cardiac cycle for the SSP metamaterial fabric sensor, while graph 1816 shows simulated phase changes ΔΦ as a result of a heartbeat during a cardiac cycle for the MSL fabric sensor.

[0161] As about Fig.18A and Fig.18B As shown, the simulated phase changes ΔΦ due to radial artery pulse and heartbeat were observed to be enhanced by 2 and 15 times for the SSP metamaterial fabric sensor, respectively, compared to the conventional microstrip fabric sensor. The sensitivity can be tuned by changing the amount of electromagnetic energy coupled into the body and is controlled by the distance of the body from the SSP metamaterial fabric sensor. Fig.19 Shown in.

[0162] Fig.19 A graph 1900 of simulated phase change ΔΦ as a function of separation distance d at the back of the body and wrist is shown, according to an embodiment, where distance d varies from 1 mm to 10 mm. The simulated phase change ΔΦ as a function of separation distance d is shown for a conventional microstrip (MSL) fabric sensor and the SSP metamaterial fabric sensor of the present disclosure. Fig.19 , where curve 1902 shows the simulated phase change ΔΦ for the SSP metamaterial fabric sensor as a function of the separation distance d at the back of the body, curve 1904 shows the simulated phase change ΔΦ for the SSP metamaterial fabric sensor as a function of the separation distance d at the wrist of the body, curve 1906 shows the simulated phase change ΔΦ for the MSL fabric sensor as a function of the separation distance d at the back of the body, and curve 1908 shows the simulated phase change ΔΦ for the MSL fabric sensor as a function of the separation distance d at the wrist of the body. Inset 1910 shows a cross section of the MSL fabric sensor with a width w=6.8 mm of the MSL. This parameter is used for simulations related to the MSL fabric sensor. Fig.19 It is shown that the SSP metamaterial fabric sensor exhibits a 10-fold larger phase change compared to the MSL fabric sensor, indicating that the SSP metamaterial fabric sensor has higher achievable sensitivity and wider tunability.

[0163] Due to the near-field nature of the present sensing approach utilizing SSP metamaterial fabric sensors, localized multi-point sensing of vital signs can be achieved through strategic design and placement of sensors on ordinary everyday objects.

[0164] The SSP metamaterial fabric sensor is robust in folding and bending, so the bending of the SSP metamaterial fabric sensor in two different planes is studied.

[0165] Fig. 20A and Fig. 20B Graphs 2000, 2010 respectively show transmission loss of an SSP metamaterial fabric sensor as a function of bending radius in the xz plane and the xy plane according to an embodiment. Fig. 20A and Fig. 20B As shown, the transmission loss of the SSP metamaterial fabric sensor decreases with increasing bending radius. Graphs 2000, 2010 show that the transmission loss of the SSP metamaterial fabric sensor remains low for a wide range of bending radii in both the xz plane and the yz plane, respectively (except when the bending is too extreme, such as Fig. 20B The SSP metamaterial fabric waveguide is therefore robust to bending and this makes it a suitable candidate for conformal integration into many different types of surfaces / accessories for contactless sensing without any significant degradation in performance.

[0166] In the present embodiment for multi-point sensing, two sensor designs - a straight sensor design and a U-shaped sensor design - are used to facilitate conformal and universal incorporation into tables and chairs of various shapes and sizes. Fig.21A and Fig. 21B Two sensor designs are shown. Fig.21A and Fig. 21B Figures 2100, 2110 show two SSP metamaterial fabric sensor designs according to embodiments, wherein Fig.21A shows a U-shaped sensor design for a chair, and Fig. 21B A straight sensor design for use on a table is shown.

[0167] Fig. 22 The use of the embodiment is shown Figure 1A 2200 of a vital sign data collection setup of the system 100. In particular, respiratory and cardiac signals are measured from the back of a body 2202 via channel 1 (CH1) 2204 using a U-shaped sensor provided on the backrest of an office chair, and a radial artery pulse signal from the wrist of the body 2202 is measured via channel 2 (CH2) 2206 using a straight sensor provided on a table. The Doppler phase changes caused by the corresponding physiological movements are demodulated at the SDR 2208 and bandpass filtered in a computer to separate the respiratory signal (0.1 to 0.8 Hz) and cardiac signal (0.9 to 5.0 Hz) in CH1 2204, and the radial artery pulse signal (0.9 to 5.0 Hz) in CH2 2206. A beat detection algorithm 2210 is developed for the respiratory signal, the cardiac signal, and the pulse signal to extract the corresponding beat-to-beat rates. This is shown in FIG. Figure 6 described and in Fig. 222204, and the radial artery pulse rate 2216 can be extracted from the signal received from CH2 2206. The collected heartbeat and radial artery pulse data are aligned by a machine learning model LSTM aligner 2218 for calculating the pulse transit time (PTT), which is used to calculate the blood pressure (BP) value.

[0168] To collect respiration and heartbeat data, a U-shaped SSP metamaterial fabric sensor was attached to the backrest of an office chair. The participant was ordered to sit upright in the chair and lean gently against the backrest. The sensor was connected to CH1 2204 of the SDR system 2208, and the Doppler phase signal was recorded using a computer. To collect data, the participant was asked to breathe normally while minimizing limb movement. For each of the respiration and heartbeat sensing assessments, a reference signal was collected using a respiratory effort transducer (SS5LB, BIOPAC Systems) fastened around the participant's chest or a 3-lead ECG transducer (SS2LB, BIOPAC Systems) attached to the participant's legs and right arm, respectively. Synchronization of the recorded physiological signals from the sensor measurements and reference measurements was performed using the system timestamp obtained from the screen video recording.

[0169] For radial artery pulse data collection, a straight SSP metamaterial fabric sensor was placed on an office table and connected to CH2 2206 of an SDR system 2208. Participants were instructed to place either their right or left wrist on the straight SSP metamaterial fabric sensor with the radial artery directly on top of the conductive fabric, and were asked to breathe normally while minimizing movement. Doppler signals from radial artery pulse motion were recorded in the same manner as for the breathing or heartbeat assessment experiments. A reference pulse signal was simultaneously collected using a finger PPG sensor (SS4LA, BIOPAC Systems, Inc.) fastened to the index finger on the same side as the wrist under measurement.

[0170] Fig.23 The use of the embodiment is shown Fig. 22 Plot 2300 of sample data collected from two sensor channels CH1 2204 and CH2 2206 of the vital signs data collection setup.

[0171] The raw data 2302 from CH1 shows the superposition of breathing and heartbeat information, while the raw data 2304 from CH2 shows the clear radial artery pulse. D(t)) is processed by bandpass filtering to obtain a respiration signal 2306 (also referred to as a time-varying respiration phase signal in the present disclosure), a cardiac signal 2308 (also referred to as a time-varying cardiac phase signal in the present disclosure), and a pulse signal 2310 (also referred to as a time-varying radial artery pulse phase signal in the present disclosure). Beat-by-beat respiration, heartbeat, and pulse rates can be obtained using a beat detection algorithm as previously described.

[0172] Fig.23 Qualitatively, detected respiration 2312, heartbeat 2316, and pulse 2320 on a recorded signal segment are shown, superimposed with true reference respiration 2314, reference ECG R peak 2318, and reference PPG peak 2322 obtained from their respective references (i.e., respiration belt, electrocardiogram (ECG), and finger photoplethysmography (PPG)). For respiration signal 2306, it can be observed that the respiration peak in the measured signal closely follows the reference signal, indicating good temporal consistency. However, for heart signal 2308 and pulse signal 2310, consistent temporal shifts between the detected beats and the reference beats can be observed due to the different selection of beat markers in each waveform. Specifically, while ECG R peak 2318 and PPG peak 2322 are conventionally used as beat locations, the beat-by-beat algorithm of the present disclosure detects the positive zero intercept as a beat.

[0173] To validate the accuracy of the SSP metamaterial fabric sensor, three independent experiments were performed on human volunteers to benchmark respiration rate, heart rate, and pulse beat-to-beat interval (BBI) against their respective reference measurements. FIG. 24A to FIG. 24C As shown, the degree of agreement between the sensor-measured beats and the reference beats was assessed by performing a Bland-Altman analysis on all recorded BBIs for respiration, heartbeat, and radial artery pulse.

[0174] For each experiment, healthy participants (20-35 years old) with no known pre-existing cardiopulmonary conditions were recruited. The number of participants for each evaluation experiment is provided in Table 1 below:

[0175]

[0176] Table 1: Number of participants in each evaluation experiment

[0177] Participants wore normal office clothing, such as T-shirts and shirts. All experiments were conducted in an open dry laboratory space with other laboratory members not involved in the experiment. Each trial consisted of recording data for each participant for 10 minutes.

[0178] Reference data were measured using a BIOPAC MP462 channel data acquisition (DAQ) unit connected to an associated compatible respiratory, ECG or PPG transducer, powered by a computer and controlled by the BIOPAC Student Lab software.

[0179] Fig.24A , Fig. 24B and Fig.24C Bland-Altman plots 2400 , 2410 , 2420 of hop-by-hop intervals obtained from reference measurements and using the metamaterial fabric sensor of the present disclosure are shown, according to an embodiment. Fig.24A A Bland-Altman plot 2400 for respiration rate is shown. Fig. 24B A Bland-Altman plot 2410 for heart rate is shown, and Fig.24C A Bland-Altman plot 2420 is shown for radial artery pulse rate. As previously described, Fig. 24B The Bland-Altman plot 2410 uses data from 10 subjects and includes Figure 8C Bland-Altman plot of 820 data.

[0180] Each of the Bland-Altman graphs 2400, 2410, 2420 shows the deviation (or mean) 2402, 2412, 2422 and the limits of agreement (LoA) 2404, 2414, 2424 for respiratory rate, heart rate, and radial pulse rate, respectively. Insets 2406, 2416, 2426 show the number of male and female participants of the graphs 2400, 2410, 2420, respectively.

[0181] like Fig.24A , Fig. 24B and Fig.24C As shown in the Bland-Altman graphs 2400, 2410, 2420 of FIG. 2404, good agreement was observed for all sensed targets with low deviations of 0.627ms, -0.143ms, and 0.128ms for respiration rate, heart rate, and radial artery pulse rate, respectively. Narrow limits of agreement (LoA) represented by dashed lines 2404, 2414, 2424 were also observed.

[0182] Along with respiratory rate and heart rate, blood pressure (BP) is another vital parameter of a person's long-term health and is a particularly insightful indicator of cardiovascular health that is routinely monitored by healthcare providers. In current BP measurement practices, cuff-based methods (such as auscultation or oscillometric methods) are the most popular, especially among which oscillometric methods are the "clinical standard" because they are non-invasive and well-validated. However, the use of cuffs in BP measurements is inconvenient, time-consuming, and non-continuous, resulting in unsatisfactory detection and management of hypertension and related cardiovascular diseases. Therefore, there is a strong need for a convenient, unobtrusive, and low-cost ubiquitous continuous BP monitoring technology that can be used during daily activities with minimal interruption.

[0183] In this disclosure, as for Fig. 22 As shown in the diagram 2200 of FIG. 1 , the system 100 is adapted to obtain highly accurate vital signs from multiple locations on the body, and this can be used to provide ubiquitous, cuffless, and continuous BP monitoring during daily life. This is discussed below with respect to Figures 25 to 36 Give a description.

[0184] Specifically, the present system 100 uses pulse transit time (PTT) to measure BP without a cuff, where PTT is defined as the time it takes for a cardiac pulse wave to travel between two arterial sections on the body. Conventional PTT measurements rely primarily on ECG and PPG sensors, which require mechanically stable contact with the skin. In addition, other non-contact vital sign monitoring technologies, such as those using radiated RF waves, do not have the necessary spatial selectivity for PTT measurements and are therefore not suitable for BP measurements. In contrast, the present system 100 includes an SSP metamaterial fabric sensor that is suitable for non-radiative sensing over a local area / space. This provides a path for accurate cuffless BP monitoring.

[0185] Using the multi-point sensing capability of SSP metamaterial fabric sensors, Fig. 22 The sensor placement shown allows for simultaneous measurement of the cardiac signal and the pulse signal, which can provide additional time information associated with each heartbeat. Conventionally, PTT is measured as the time delay between the ECG R peak and a specific feature of the PPG waveform (such as a foot or peak). Since the measured cardiac signal and radial artery pulse signal are correlated to their corresponding references (i.e., ECG and PPF, respectively) at a physiological level (i.e., originating from the same physiological process), the timing of the ECG R peak and the PPG maximum first derivative (MFD) point used to obtain PTT can be inferred from the measured cardiac signal and the measured radial artery pulse signal, respectively. In order to perform PTT measurements that follow conventional standards, a machine learning model has been developed to align the prominent features (e.g., peaks) of the measured cardiac signal and radial artery pulse signal with their corresponding reference features.

[0186] For the machine learning model, signal alignment is formulated as a sequence-to-sequence learning task to convert the measured Doppler phase signals (or time-varying phase shift signals) (which represent cardiac or radial artery pulse information depending on the original sensor channel (i.e., CH1 or CH2)) into correlated time series with temporal features that closely match their respective references. The algorithm consists of two long short-term memory (LSTM) networks followed by fully connected (FC) layers working independently on each sensor channel, as Fig.25 shown.

[0187] Fig.25 A diagram 2500 illustrating a machine learning model according to an embodiment is shown, which includes a long short-term memory (LSTM) network and a fully connected (FC) layer for performing alignment of a heart signal (CH1) and a pulse signal (CH2) to obtain a pulse transit time (PTT) for blood pressure measurement.

[0188] First, the measured Doppler phase signal (instantaneous phase shift signal φ) is received from the sensor channel D (t)). These sensor outputs from CH1 and CH2 (instantaneous phase shift signal φ D (t)) is band-pass filtered from 0.9 Hz to 5 Hz to obtain the input heart signal 2502 of CH1 and the input pulse signal 2504 of CH2, as shown in FIG. Fig.25 As shown. A reference ECG signal 2506 and a reference PPG signal 2508 are also shown as references. Before being processed by the machine learning model, as shown Fig.25 As shown, the peak of the input cardiac signal 2502 is not aligned with the ECG R peak of the reference ECG signal 2506.

[0189] Then, in step 2510, the input cardiac signal 2502 is shifted left by 0.2 seconds (i.e., time advance) relative to the reference ECG signal 2506 to form a processed time-varying cardiac phase signal, and in step 2512, the first-order time derivative of the input pulse signal 2504 is taken to form a processed time-varying radial artery pulse phase signal. Then, these processed time-varying cardiac phase signals and processed time-varying radial artery pulse phase signals are input into their respective sequence input layers 2514 to transform them into sequences of the same length, and then their sequence outputs are provided to their respective long short-term memory (LSTM) networks 2516 and subsequently to their respective fully connected (FC) layers 2518, as shown. Fig.25As shown. The processing layers (i.e., the processing layers associated with the processing step 2510 for shifting the input cardiac signal 2502 to the left and the processing step 2512 for applying the time derivative to the input pulse signal 2504), the sequence input layer 2514, the LSTM network 2516, and the FC layer 2518 are collectively referred to as the "LSTM aligner" in the present disclosure and form a machine learning model. The LSTM aligner outputs an aligned time-varying cardiac phase signal 2520 and an aligned time-varying radial artery pulse phase signal 2522, whose peaks are aligned with the reference ECG R peak of the reference ECG signal 2504 and the PPG MFD point of the time derivative of the reference PPG signal 2524, respectively, as shown in FIG. Fig.25 The beat-by-beat PTT 2526 can then be calculated as the time delay between the detected peaks of these aligned time-varying cardiac phase signals 2520 and the aligned time-varying radial artery pulse phase signal 2522, as Fig.25 shown.

[0190] The computer 108 can be configured to perform a PTT calculation on the aligned time-varying cardiac phase signal 2520 and the aligned time-varying radial artery pulse phase signal 2522. In the present embodiment, for each located peak of the aligned time-varying cardiac phase signal 2520, the computer 108 stores an algorithm adapted to search for a corresponding local maximum of the corresponding aligned time-varying radial artery pulse phase signal 2522 within a subsequent time window of 0.15s to 0.4s thereof. The algorithm can also be configured to search for and remove erroneously detected peaks by detecting outliers in the BBI in an intermediate step. For example, an outlier can be defined as a value of more than five scaled absolute median differences of the median of all BBIs in a period or record, but other definitions of outliers can be used in other embodiments.

[0191] Fig.26 The use of the embodiment is shown Fig.25 Bland-Altman plot 2600 of estimated PTT from test data using a machine learning model. A Bland-Altman plot 2600 of estimated PTT is formed using the LSTM aligner output and reference ECG and PPG measurements obtained from the test data, which was collected from 2 subjects for 10 minutes each.

[0192] Fig.26 The Bland-Altman plot 2600 in shows a small mean deviation 2602 compared to the reference at -0.192 ms, with the data points being within the limits of agreement (LoA) 2604 of approximately 50 ms.

[0193] Fig. 27is a flow chart showing the steps of a computer-implemented method 2700 for generating an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal to calculate pulse transit time (PPT) for conversion to systolic and diastolic pressure values, according to an embodiment. Fig.25 An overview of the aforementioned process is described.

[0194] In step 2702, a time-varying cardiac phase signal and a time-varying radial artery pulse phase signal are received. This corresponds to the input cardiac signal 2502 and the input pulse signal 2504, which are obtained by bandpass filtering their respective time-varying phase shifted signals from 0.9 Hz to 5 Hz.

[0195] In step 2704, the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal are processed to form a processed time-varying cardiac phase signal and a processed time-varying radial artery pulse phase signal, respectively. This may include left-shifting the time-varying cardiac phase signal by a predetermined amount of time and differentiating the time-varying radial artery pulse phase signal with respect to time to generate a time derivative of the time-varying radial artery pulse phase signal.

[0196] In step 2706, an aligned time-varying cardiac phase signal 2520 and an aligned time-varying radial artery pulse phase signal 2522 are generated for each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal using a machine learning model including a long short-term memory (LSTM) network followed by a fully connected (FC) layer. In particular, the peak of the aligned time-varying cardiac phase signal 2520 corresponds to the electrocardiogram (ECG) R wave peak, and the peak of the aligned time-varying radial artery pulse phase signal 2522 corresponds to the photoplethysmography (PPG) maximum first derivative (MFD) point, which can be used for subsequent PPT calculations.

[0197] In step 2708, the pulse transit time (PPT) is calculated as the time delay between one of the peaks of the aligned time-varying cardiac phase signal 2520 and a corresponding one of the peaks of the aligned time-varying radial artery pulse phase signal 2522. In an embodiment, for each located peak of the aligned time-varying cardiac phase signal 2520, the computer 108 stores an algorithm adapted to search for a corresponding local maximum of the corresponding aligned time-varying radial artery pulse phase signal 2522 within a subsequent time window of 0.15 s to 0.4 s thereof.

[0198] In step 2710, the PPT calculated from step 2708 is converted to systolic and diastolic blood pressure values. This may involve using a linear PPT-blood pressure relationship. This is discussed below with respect to Fig.32A and Fig.32B Further discussion.

[0199] In order for the machine learning model to provide acceptable output with respect to aligning the input cardiac signal 2502 and the input pulse signal 2504 for PPT calculation, the machine learning model must be trained to form a trained machine learning model.

[0200] Fig.28 A diagram 2800 is shown illustrating a Fig.25 Architecture and training of one of the LSTM aligner channels of the machine learning model. Fig.25 The machine model shown performs heart signal (CH1) and pulse signal (CH2) alignment through two separate LSTM-based networks, but for simplicity, only one LSTM-based network is shown here.

[0201] Physiological data (e.g., training heart data and training pulse data) were collected from all subjects during the accuracy evaluation experiment and used as training data for the LSTM aligner. Two separate data sets were constructed to train the corresponding networks for heartbeat alignment and pulse alignment. The training time-varying heart signal and the training time-varying radial artery pulse signal were obtained by bandpass filtering their corresponding training time-varying phase signals from 0.9 Hz to 5 Hz. Input features 2802 were constructed by first generating incremental left-shifted copies of the training time-varying heart signal and the training time-varying radial artery pulse signal with a step size of 5 samples until the desired number of features were obtained. In the present embodiment, the number of features N is 5 and 15 for heart signal alignment (CH1) and pulse signal alignment (CH2), respectively.

[0202] Each feature is divided into a non-overlapping window 2804 of 5000 samples, corresponding to 12.5 seconds. This produces a total training sample size of N×5000, which is the size of a training sample. As shown in Table 3 below, multiple training samples of this size are obtained from each subject ID. Before forming a non-overlapping window 2804 of 5000 samples, the cardiac signal feature is additionally subjected to a left shift of 0.2s. For the pulse signal feature, the time derivative of the pulse signal feature is obtained before forming a non-overlapping window of 5000 samples. A test data set for evaluating the LSTM aligner was constructed separately using additional dual-channel data collected from 2 healthy volunteers for a total of 10 minutes. All data processing, network training, and evaluation are performed in MATLAB (R2021a, The MathWorks).

[0203] The training input features are provided to an LSTM-based network 2806 (including a sequence input layer, an LSTM layer, and a fully connected (FC) layer), and the output signal generated by the LSTM-based network 2806 is a transformed sequence of the same length. In the LSTM-based network 2806, the sequence layer is used to feed the training samples into the LSTM layer (80 hidden units), followed by a 20% random dropout and an FC layer. The training target 2808 is the time derivative of the corresponding 5000 sample window ECG R peak mask or PPG. The LSTM-based network 2806 is trained or optimized by minimizing the mean square error (MSE) loss between the output signal and the training target using a regression layer 2810. Examples of these signals are shown in Fig.28 Each of the LSTM-based networks for cardiac signal (CH1) and pulse signal (CH2) alignment are trained separately using their respective training and target datasets.

[0204] The ability of the LSTM aligner to generalize in unseen subjects was evaluated using a subject-based cross-validation procedure. For each subject, each LSTM-based network was trained using data from the other 9 subjects and then evaluated on that subject, resulting in a 10-fold cross-validation procedure. Finally, the corresponding network was trained on all training data and then evaluated on the test dataset.

[0205] The LSTM-based network parameters for each of the LSTM-based networks for CH1 and CH2 are shown in Table 2.

[0206]

[0207] Table 2: LSTM-based network parameters Table 3 showing the number of training samples by subject ID is provided below:

[0208]

[0209] Table 3: Training samples by subject ID

[0210] Fig.29A and Fig.29B Plots 2900, 2910 show sample data for validating pulse signal alignment and cardiac signal alignment according to an embodiment. Fig.25 Machine learning models.

[0211] Fig.29A A plot 2900 of average detection error versus subject ID for pulse signal alignment is shown, and Fig.29BA plot 2900 of the mean detection error for cardiac signal alignment versus subject ID is shown. Fig.29A and Fig.29B As shown, the average detection error for all subject IDs for pulse signal alignment is below 40 ms, and the average detection error for all subject IDs for cardiac signal alignment is below 80 ms.

[0212] Use Fig. 22 The sensor arrangement shown and utilized as described in Fig.25 The described machine learning model can enable seamless integration of SSP sensors in common physical spaces and can implement environmental sensing methods for unobtrusive, ubiquitous, and convenient health monitoring without requiring any active participation or effort from the user. Therefore, environmental sensors have become increasingly relevant in healthcare as they allow for the long-term collection of continuous health data to provide a more comprehensive understanding of an individual's health trends and patterns, making them well suited for preventive health monitoring and early disease detection.

[0213] This is particularly important in the modern workplace environment, where sedentary lifestyles have become prevalent among adults, such as sitting at a desk for a significant number of hours. As higher amounts of sitting time are associated with an increased risk of cardiovascular disease and mortality, long-term physiological monitoring in the office can serve as an effective management strategy to improve overall public health standards.

[0214] Fig.30 Photos 3000, 3002, 3004 of multiplexed health monitoring using a dual-channel metamaterial fabric sensor system in an office environment according to an embodiment are shown. As shown in photos 3002 and 3004, a U-shaped metamaterial fabric sensor connected to CH1 is provided on the back of a chair, and a straight metamaterial fabric sensor connected to CH2 is provided on a table in front of a keyboard for simultaneously continuously recording data for 100 minutes. Fig.30 The illustrated experimental setup was used to perform ambient contactless sensing of multiple health metrics on healthy male volunteers during a typical 100-minute work session.

[0215] By definition, environmental sensors are ubiquitous and responsive to the user while imposing no restrictions on the user's movement or behavior. Therefore, the experiments performed adhered to the expected use cases by simulating real-world working conditions, where human subjects and other workers can move freely in the office space. In particular, the subjects were ordered to sit on a chair and perform routine tasks on the computer, such as reading, using a mouse and typing on a keyboard as needed during the normal course of work. Movement of the arms and legs and other activities (such as drinking, talking to others, and getting up from a sitting position) were not restricted. Other workers in the same office space were also free to move during this period. In addition, the subject's body and wrist position were not limited to a specific position for continuous measurement. In contrast, the dual-channel metamaterial fabric sensor system is implemented to perform its sensing function whenever the corresponding channel meets the correct conditions to detect the presence of the subject's vital signs.

[0216] Throughout the entire period of the experiment, data is continuously collected from two channels. A digital sphygmomanometer (OMRON HEM-7156-A) is used to provide reference BP measurements from the left arm at intervals of approximately 20 minutes, for a total of 5 measurements. 2 minutes before each reference measurement, the subject is ordered to place his right wrist on CH2 for pulse sensing until the reference BP value has been successfully recorded. In order to calculate BP values ​​from PTT, the linear relationship between PTT and BP is used in this embodiment, which is obtained from the separate calibration by the least squares regression of the measured PTT and BP pairs.

[0217] Fig.32A and Fig.32B Graphs 3200 , 3210 are shown of linear regression using calculated PTT values ​​for estimating blood pressure, according to an embodiment. Fig.32A A graph 3200 showing a linear regression for estimating systolic blood pressure, and Fig.32B A graph 3210 of linear regression for estimating diastolic pressure is shown.

[0218] For each data point, systolic and diastolic BP values ​​were measured by a commercial digital BP monitor (OMRON) and analyzed using the same protocol as described for Fig.25 The PTT value is calculated based on the sensor output described. The BP of the subject is changed by moderate exercise followed by a period of rest. The equations 3202, 3204 of the best fit line for the linear regression of systolic and diastolic pressures, respectively, are used for subsequent BP estimation based on the calculated PTT value.

[0219] Referring back to the multiplexed health monitoring in an office environment, periods of large movement are excluded from rate calculations because the subject is allowed to move freely and random movement of the subject may result in spikes in the sensor output that mask the desired signal. In this embodiment, an algorithm is implemented to automatically detect "absent" periods, during which the subject is not sitting in the chair, and "sustained BP" periods, during which high-quality data is available from both sensor channels for BP calculations.

[0220] In this embodiment, presence detection and continuous BP period detection are performed in a 30s period. For absence detection, a threshold is implemented on the spectral power stored in the 99% occupied bandwidth (OBW) of the filtered bandpass sensor output from CH1. In this embodiment, the threshold is selected to be -45dB, and periods below this threshold are marked as "absent". It should be understood that different thresholds can be selected in other embodiments.

[0221] The automatic detection of the "sustained BP" period is based on the fact that if the heart signal and the pulse signal are present in one period at the same time, the heart rate (HR) and the pulse rate (PR) calculated from the corresponding signals should be equal. In the present embodiment, in order to implement the automatic detection of the "sustained BP" period, the absence detection method as described above is first used to identify whether the signal is present in both channels. Then the HR mean and the PR mean are calculated for the identified period (i.e., the signal is present in both channels), and their absolute difference is used as a similarity measure. If the absolute difference is less than 2 beats per minute (bpm), the period is selected as a candidate period for sustained BP detection. In order to further improve the robustness, based on the basic principle that in high-quality signals, the HR mean or PR mean calculated by BBI should be consistent with the number of beats detected per minute, a signal quality measure is defined for the two signals in each candidate period, as follows:

[0222]

[0223] Where N is the number of beats detected in one period, t e is the length of an epoch in minutes, and I is the BBI calculated for each consecutive pair of detected beats in seconds. Finally, if for both the heart signal and the pulse signal in the candidate epoch, Q e >0.5, a candidate period is selected (the selected candidate period is referred to as a detection period in this disclosure) for continuous BP detection.

[0224] Fig.33A and Fig.33BGraphs 3300 , 3310 are shown regarding selection of candidate epochs for continuous blood pressure detection according to an embodiment.

[0225] Fig.33A A graph 3300 is shown for identifying candidate periods for blood pressure sensing using the absolute difference between the mean heart rate (HR) and the mean pulse rate (PR) for each period. The graph 3300 includes the absolute difference on the y-axis and the time on the x-axis. The graph 3300 shows data collected over a 100 minute period. Periods 3302 identified as "absent" of a heart signal or pulse signal are assigned a value of 0. The threshold is set to 2 bpm, as shown by line 3304. Periods with an absolute difference below the threshold are considered "sustained BP" candidates (or candidate periods).

[0226] Fig.33B The use of signal quality metrics (Q e )choose Fig.33A Graph 3310 of candidate epochs for continuous blood pressure detection identified in Graph 3300. Signal quality metrics are calculated for all "continuous BP" candidate epochs identified in Graph 3300. Non-candidate epochs are assigned Q e The value is 0. The threshold is set to 0.5, as shown by line 3312. Periods where both the heart signal 3316 and the pulse signal 3318 have a quality above this threshold are identified as "sustained BP" periods (or detection periods), as indicated by the shaded box 3314. PTT and BP are then calculated for these "sustained BP" periods.

[0227] Fig.31 Shown are normalized outputs from both channels CH1 and CH2, as well as graphs 3100, 3102, 3104, 3106 of detected vital signs including heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP), and diastolic blood pressure (DBP), according to an embodiment.

[0228] Graphs 3100, 3102 show two 30 second periods of normalized output from both channels CH1 and CH2, while graphs 3104, 3106 show detected vital signs including heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP) and diastolic blood pressure (DBP). A reference BP measurement is shown for the second period of graph 3106. Fig.31 The graph 3100 shown relates to a normalized bandpass filtered signal from 0.1 Hz to 5 Hz and includes both a time-varying cardiac phase signal component (0.1 to 0.8 Hz) and a time-varying respiratory phase signal component (0.9 to 5 Hz). Fig.31The graph 3102 shown relates to the normalized time-varying radial artery pulse phase signal (i.e., the time-varying phase signal from CH2 that has been bandpass filtered at 0.9 to 5 Hz). In particular, Fig.31 3106 ). An enlarged view of two exemplary periods at the 61 minute mark of the experiment is shown in FIG. 3107 . To verify the accuracy of the BP measurements, the subject was asked to place his wrist on the CH2 sensor for 2 minutes approximately every 20 minutes, and the subject's reference BP was measured during these periods using a conventional digital BP monitor. Since only a single pair of SBP and DBP values ​​(i.e., SBP ref. and DBP ref. as shown in graph 3106 ) was available for each reference measurement, they were considered to be averaged for the entire period, as shown in FIG. Fig.31 The curve graph 3106 and about Fig.34 3406. In contrast, during each "continuous BP" period, the present system can be used to provide a continuous record of BP values ​​without having to actively initiate measurements, making them more insightful and convenient than conventional BP measurements using an arm cuff.

[0229] Fig.34 Long-term health monitoring data of normalized sensor outputs from both channels CH1 and CH2 according to an embodiment are shown, along with graphs 3400 , 3402 , 3404 , 3406 of detected vital signs including heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP) and diastolic blood pressure (DBP).

[0230] Fig.34 Filtered and normalized outputs from two sensor channels and all extracted physiological signals for the entire 100 minute work period are shown. Graphs 3400, 3402 show the normalized outputs from CH1 and CH2, respectively, graph 3404 shows detected vital signs including heart rate (HR), respiratory rate (RR), and graph 3406 shows measured and reference values ​​for systolic blood pressure (SBP) and diastolic blood pressure (DBP) for the 100 minute work period. The output is preprocessed by dividing the output into non-overlapping 30 second periods, each period containing data from 2 channels (i.e., CH1 and CH2). As shown Fig.34 The graph 3400 shown relates to a normalized bandpass filtered signal from 0.1 Hz to 5 Hz and includes both a time-varying cardiac phase signal (0.1 to 0.8 Hz) component and a time-varying respiratory phase signal (0.9 to 5 Hz) component. Fig.34The graph 3402 shown relates to the normalized time-varying radial artery pulse phase signal (i.e., the time-varying phase signal from CH2 that has been bandpass filtered at 0.9 to 5 Hz). The beat-by-beat HR and RR for each epoch are calculated from the CH1 output. Because random movement of the subject causes spikes in the sensor output that mask the desired signal, epochs in which there is large movement, such as in the case of Fig.34 3408. As discussed above, "absent" periods, during which the subject is not sitting in the chair, and "sustained BP" periods, during which high-quality data is available from both sensor channels for BP calculations, can be automatically detected by the system. The "absent" periods and "sustained BP periods" are shown as boxes 3410 and 3412, respectively. Using the same techniques as previously described with respect to FIG. Fig.25 The described LSTM aligner pipeline computes PTT for each “persistent BP” epoch and then uses the PTT from Fig.32A and Fig.32B The linear PTT-BP relationship obtained by the calibration procedure prior to the described experiments converts PTT into systolic BP (SBP) values ​​3414 and diastolic BP (DBP) values ​​3416 as shown in the curve graph 3406. Fig.34 Also shown is a dashed line 3418, which relates to Fig.31 Two periods are shown.

[0231] In another embodiment, the fabric sensor system of the present disclosure is implemented in an aircraft cabin simulator to demonstrate the ability of the SSP metamaterial fabric sensor to implement in-flight environmental monitoring applications. During air travel, especially on long-haul flights, passenger health monitoring is not only important for early detection of in-flight medical emergencies, but also a great tool for collecting passenger health and comfort information.

[0232] Fig.35 Photos 3500, 3510 of a passenger health monitoring system including a metamaterial fabric sensor 3502 according to an embodiment are shown. The sensor 3502 is implemented as an in-flight health tracker by integration with an aircraft seat. To evaluate the sensing capability, a 50-minute continuous HR monitoring test was conducted on healthy male volunteers.

[0233] In particular, healthy male volunteers were asked to sit, as shown in photo 3510. Volunteers were also asked to wear an Apple Watch (version 7, Apple) on their left wrist for reference heart rate recording. The assessment trial lasted a total of 50 minutes, including 3 uninterrupted time periods. In the first and third 20-minute periods, volunteers were ordered to sit still in an upright posture. In the middle 10-minute period, volunteers were ordered to get up from the chair to perform a static running exercise. Sensor signals were continuously recorded throughout the period. The continuous HR using the SSP metamaterial fabric sensor 3502 was extracted according to the method previously outlined. The reference HR comes from the Apple watch, obtained from the iOS health app on the paired iPhone (version 13Pro, Apple), which is provided as a minimum to maximum range every 2 minutes.

[0234] Fig.36 Showing the use according to the embodiment Fig.35 Graphs 3600, 3610 of normalized output and detected heart rate measured by a passenger health monitoring system, and using Graph 3620 of the reference heart rate measured by the watch.

[0235] Graph 3600 shows the normalized output, with the shaded box 3602 indicating the time period during which the volunteer was commanded to get up from the chair and perform a static running exercise. Graph 3600 relates to the normalized time-varying cardiac phase signal (i.e., the time-varying phase signal from CH1, which has been bandpass filtered at 0.9 to 5 Hz). Graph 3610 shows the detected continuous heart rate measured by sensor 3502. Illustrations 3612, 3614 show enlarged views of the sensor output heart rate waveform before and after performing a static exercise, respectively. For each of these illustrations 3612, 3614, a scale of 0.5 s is also shown. Graph 3620 shows the reference HR values ​​measured by the Apple Watch, which are represented as bars showing the minimum to maximum range for each 2-minute period.

[0236] Fig.36 The ability of the SSP sensor to capture a wide range of HR was shown by increasing the HR of the subjects with exercise. During the seated period (first and last 20 minutes), the sensor could collect the subjects' continuous beat-by-beat HR, which was in good agreement with the reference, without the need for active participation of the volunteers. Therefore, these results demonstrate the potential of the SSP metamaterial fabric sensor for ambient contactless physiological monitoring in a wide range of complex environments.

[0237] Fig.37is a block diagram illustrating a technical architecture 3700 of a computer 108. The computer 108 includes a memory storing computer program modules implementing methods and / or processes for obtaining vital signs from a body, such as breathing cycle, heart rate, radial pulse rate, and blood pressure.

[0238] The technical architecture 3700 of the computer 108 includes a processor 3702 (which may be referred to as a central processing unit or CPU) in communication with memory devices including secondary storage 3704 (such as a disk drive), read-only memory (ROM) 3706, and random access memory (RAM) 3708. The processor 3702 may be implemented as one or more CPU chips. The technical architecture 3700 may also include input / output (I / O) devices 3710 and network connection devices 3712.

[0239] Secondary storage 3704 typically includes one or more disk drives or tape drives and is used for non-volatile storage of data and as an overflow data storage device if RAM 3708 is not large enough to hold all working data. Secondary storage 3704 can be used to store programs that are loaded into RAM 3708 when they are selected for execution.

[0240] In this embodiment, the secondary storage 3704 has a processing component 3704a that includes non-transitory instructions that are operated by the processor 3702 to perform various operations of the methods and processes of the present disclosure. These instructions include various processes described for processing the time-varying phase signal received from the SDR system 106 to obtain respiratory cycles and beat-to-beat intervals related to the heart rate and radial artery pulse rate, and / or using the methods described above. Fig.25 The described machine learning model is used to calculate the PPT and obtain the blood pressure value of the body 102. In this embodiment, the auxiliary storage 3704 also stores the machine learning model including the LSTM layer and the fully connected layer as described. ROM 3706 is used to store instructions and possible data read during program execution. Auxiliary storage 3704, RAM 3708 and / or ROM 3706 may be referred to as computer-readable storage media and / or non-transitory computer-readable media in some contexts.

[0241] I / O devices 3710 may include a printer, video monitor, liquid crystal display (LCD), plasma display, touch screen display, keyboard, keypad, switches, dials, mouse, trackball, voice recognizer, card reader, paper tape reader, or other known input devices.

[0242] The network connection device 3712 can take the form of a modem, a modem bank, an Ethernet card, a universal serial bus (USB) interface card, a serial interface, a token ring card, a fiber distributed data interface (FDDI) card, a wireless local area network (WLAN) card, a radio transceiver card that facilitates radio communication using protocols such as code division multiple access (CDMA), global system for mobile communications (GSM), long term evolution (LTE), world wide interoperability for microwave access (WiMAX), near field communication (NFC), radio frequency identification (RFID) and / or other air interface protocol radio transceiver cards and other known network devices. These network connection devices 3712 can enable the processor 3702 to communicate with the Internet or one or more intranets. With such a network connection, it is expected that the processor 3702 can receive information from the network during the execution of the above-mentioned method operations, or can output information to the network. Such information, usually represented as a sequence of instructions to be executed using the processor 3702, can be received from the network and output to the network, for example, in the form of a computer data signal embodied in a carrier wave. In particular, in the described embodiment, the network connection device 3712 of the computer 108 is suitable for connecting to the SDR system 106 for providing instructions to the SDR system 106 and / or receiving outputs from the SDR system 106 (e.g., time-varying phase signals associated with channel 1 and / or channel 2 of the SDR system).

[0243] The processor 3702 executes instructions, codes, computer programs, scripts that it accesses from a hard disk, floppy disk, optical disk (these various disk-based systems can all be considered secondary storage 3704), flash drive, ROM 3706, RAM 3708, or network connection device 3712. Although only one processor 3702 is shown, multiple processors may be present. Thus, although instructions may be discussed as being executed by a processor, the instructions may be executed by one or more processors simultaneously, serially, or otherwise.

[0244] Although the technical architecture 3700 is described with reference to a computer, it should be understood that the technical architecture can be formed by two or more computers that communicate with each other to collaborate to perform tasks. For example, but not by way of limitation, an application can be partitioned in a manner that allows concurrent and / or parallel processing of instructions of the application. Alternatively, the data processed by the application can be partitioned in a manner that allows different parts of the data set to be processed concurrently and / or in parallel by two or more computers. In an embodiment, the technical architecture 3700 can use virtualization software to provide the functionality of multiple servers that are not directly bound to multiple computers in the technical architecture 3700. In an embodiment, the functionality disclosed above can be provided by running an application and / or multiple applications in a cloud computing environment. Cloud computing can include providing computing services via a network connection using dynamically scalable computing resources. A cloud computing environment can be established by an enterprise and / or can be rented from a third-party provider as needed.

[0245] It should be understood that by programming and / or loading executable instructions onto the technical architecture 3700, at least one of the CPU 3702, RAM 3708, and ROM 3706 is changed, thereby partially transforming the technical architecture into a special-purpose machine or device having novel functions taught by the present disclosure. Functions that can be implemented by loading executable software into a computer can be converted into hardware implementations by known design rules, which are basic to the fields of electrical engineering and software engineering.

[0246] Thus, embodiments of the present disclosure provide one or more thin and conformal metamaterial sensors using conductive fabrics that can support efficient artificial surface plasmon propagation at microwave frequencies as localized non-radiative surface waves for sensing vital signs. A multi-channel sensor system is implemented that is easily integrated with common general-purpose furniture for continuous tracking of multiple vital signs and benchmarking the measured vital signs against a gold standard reference through evaluation experiments with healthy volunteers. The general utility of the SSP metamaterial fabric sensor in a real physical environment is also shown through continuous monitoring demonstrations of respiratory rate, heart rate, and cuff-free blood pressure. Thus, the present disclosure highlights the potential of metamaterial fabric sensors for relevant healthcare applications.

[0247] Exemplary embodiments of the above-described systems and methods provide an effective way for contactless sensing of one or more physiological parameters of the body. Additionally, some advantages associated with the above-described systems and methods are provided in Table 4 as shown below.

[0248]

[0249] Table 4: Advantages of the system or method

[0250] Other advantages may include: (i) the waveguide can be customized by modifying the shape of the metamaterial structure to change the localization of the artificial surface plasmons and thus modify the sensitivity of the waveguide. For example, the sensitivity of the waveguide can be reduced to increase resilience to environmental impacts, or further improved for applications in gesture sensing, proximity detection, and physiological monitoring; and (ii) improved signal security due to the localized sensing range / proximity of the waveguide.

[0251] The exemplary embodiments of the systems and methods described above should not be construed as limiting, and variations are possible. For example, other systems, methods, or software may be used to extract the detected phase shift between the transmitted signal and the received signal, and / or other systems or methods may be used to determine physiological parameters or vital signs, and / or to separate or distinguish between physiological parameters in the measured signal if more than one physiological parameter is captured in the measurement. In addition, it should be understood that other machine learning models may be trained and used to align the cardiac signal and the radial artery pulse signal for PTT calculations. Further, it should be understood that other algorithms other than the Pan-Tompkins algorithm may be used to extract the heartbeat or pulse beat.

[0252] Other alternative embodiments include: (1) the transmit signal and the receive signal can be provided / received from the same end or the same port of the waveguide (i.e., not limited to sending the transmit signal at one end of the waveguide and receiving the receive signal at the opposite end of the waveguide), (2) for the waveguide that propagates at least one artificial surface plasmon mode, other suitable materials are used (e.g., metals such as silver or gold, or composite materials) in addition to copper or conductive fabric on a polyimide substrate, (3) different forms of waveguides are used in addition to comb patterns (e.g., having a repeating C-shaped pattern, repeating rectangular blocks, or repeating cylindrical blocks), (4) waveguide structures do not include a ground layer or an unpatterned bottom layer, (5) the use of an integrated on-chip RF transceiver instead of an SDR system, or the use of Other SDR systems include those embedded with computing functions so that they can operate without a computer, (6) using a carrier signal with a frequency of 400 MHz to 200 GHz, (7) detecting or sensing other physiological parameters (e.g., walking pace or stepping speed or posture), (8) digitizing the obtained IQ components or signals using a sampling rate of 0.5 MHz to 10 MHz, (9) multiplying the complex conjugate of the baseband Tx signal by the Rx signal, low-pass filtering, and decimating to a frequency in the frequency range of 100 Hz to 500 Hz, (10) sensing physiological parameters of the body of a human, animal or organism, (11) detecting or sensing blood pressure, as exemplified above (this can be achieved in the following manner), for example, (8) using 0.5 MHz to 10 MHz sampling rate to digitize the obtained IQ components or signals, (9) the complex conjugate multiplication of the baseband Tx and Rx signals is low pass filtered and extracted to a frequency in the frequency range of 100 Hz to 500 Hz, (10) sensing physiological parameters of the body of a human, animal or organism, (10) sensing physiological parameters of the body of a human, animal or organism, (11) being able to detect or sense blood pressure through a local sensing modality such as the exemplary above content (this can be achieved by, for example, concurrently measuring the distal heart rate and the proximal heart rate, for example by using separate metamaterial sensors and a synchronized SDR system, and using the pulse transit time associated with the detected distal heart rate and the detected proximal heart rate to infer the blood pressure), the local sensing modality allows real-time concurrent sensing of physiological parameters from multiple locations of the body, (12) using other machine learning models instead of the LSTM network and / or the FC layer, (13) PTT can be converted into blood pressure values ​​(i.e., systolic BP values ​​and / or diastolic BP values) by using other forms of regression and not limited to linear regression (or linear PTT-BP relationship), for example, people with chronic heart disease can exhibit nonlinear PTT-BP relationship, (14) different numbers of matching units N for the metamaterial fabric waveguide can be used, for example, 2, 4, 5 or 6, (15) other shapes of metamaterial fabric sensors can be used (i.e., not limited to U-shaped or straight designs), (16) other placement locations of the metamaterial fabric sensor (e.g., placed on the seat of a chair to detect / measure distal pulse, etc.), (17) a dedicated device that is constructed to integrate SDR hardware and computing functions so that it can operate on its own, and (18) providing wireless communication capabilities to the dedicated device of (17) so that data can be wirelessly transmitted from the dedicated device to a computer / tablet / display device for display and analysis.

[0253] Although only certain embodiments of the present invention have been described in detail, many variations are possible according to the appended claims. For example, features described with respect to one embodiment may be incorporated into one or more other embodiments, and vice versa.

Claims

1. A sensor for non-contact sensing of physiological parameters of a body, the sensor comprising: a waveguide comprising a metamaterial and configured to receive a transmit signal and to propagate the transmit signal along the waveguide in an artificial surface plasmon mode to generate an evanescent electromagnetic field and provide a receive signal, The waveguide is placed at a predetermined distance away from the body and is used to perform non-contact sensing of disturbances caused by physiological movements of the body using the evanescent electromagnetic field, wherein the disturbance produces a phase shift between the transmitted signal and the received signal, which is used to determine the physiological parameters of the body.

2. The sensor according to claim 1, wherein: The waveguide includes a sensing layer on a sensing side of the waveguide suitable for detecting the disturbances generated by the physiological movements of the body, a ground layer on an opposite side of the sensing side, and a non-conductive layer sandwiched between the sensing layer and the ground layer, wherein the ground layer is configured to confine the evanescent electromagnetic field to the sensing side of the waveguide.

3. The sensor according to claim 1 or claim 2, wherein: The sensing layer includes a comb-like rectangular strip having an elongated base and a plurality of teeth extending along and from the elongated base, wherein adjacent teeth of the plurality of teeth are separated by a gap.

4. The sensor according to claim 3, wherein: The height of the plurality of teeth measured from the elongated base can be adapted to vary the degree of wavelength confinement of the artificial surface plasmon mode.

5. A system for contactless sensing of physiological parameters of a body, the system comprising one or more sensors according to any one of claims 1 to 4, and a software defined radio (SDR) system configured to provide the transmit signal and receive the receive signal.

6. The system according to claim 5, wherein: The SDR system comprises a digital-to-analog converter (DAC), and the SDR system is configured to: generating a digital complex baseband signal; converting the digital complex baseband signal using the DAC to form an analog baseband signal; modulating the analog baseband signal with a carrier signal to provide the transmit signal; demodulating the received signal to obtain in-phase and quadrature (IQ) components associated with the digital complex baseband signal; and The obtained IQ components are digitized.

7. The system according to claim 6, wherein: The SDR system is configured to perform a complex conjugate multiplication of the digital complex baseband signal and the digitized IQ component to determine a phase shift signal associated with a phase shift between the transmit signal and the receive signal.

8. The system according to claim 7, wherein: The SDR system is configured to filter the phase-shifted signal with a low-pass filter and downsample the filtered phase-shifted signal to form a decimated phase-shifted signal.

9. The system according to claim 8, wherein: The SDR system is adapted to perform arctangent demodulation and unwrapping on the extracted phase-shifted signal to obtain a time-varying phase signal associated with the phase shift between the transmit signal and the receive signal.

10. The system according to claim 9, wherein: The physiological movement is associated with more than one physiological parameter, the system further comprising a processor and a data storage storing computer program instructions operable to cause the processor to: The time-varying phase signal is processed with a bandpass filter to separate the time-varying phase signal into separate components associated with each of the more than one physiological parameters.

11. The system according to claim 10, wherein: The one or more sensors include a first sensor and a second sensor, wherein the first sensor is provided at the back of the body and is suitable for detecting a respiratory signal and a cardiac signal associated with the body, and the second sensor is provided at the wrist of the body and is suitable for detecting a radial artery pulse signal associated with the body, and the system further includes a processor and a data storage, wherein the data storage stores computer program instructions, and the computer program instructions are operable to cause the processor to: processing a first time-varying phase signal associated with the first sensor with a bandpass filter to separate the first time-varying phase signal into a time-varying respiratory phase signal and a time-varying cardiac phase signal; and A second time-varying phase signal associated with the second sensor is processed with a bandpass filter to obtain a time-varying radial artery pulse phase signal.

12. The system according to claim 11, wherein: The data storage further stores computer program instructions operable to cause the processor to: performing a fast Fourier transform on the first 15 seconds of each signal segment of the time-varying respiratory phase signal to estimate the respiratory period; Calculating a moving average curve by taking the mean of the time-varying respiratory phase signal over a time window equal to twice the respiratory period, so as to generate each data point of the moving average curve; Calculating an intercept between the moving average curve and the time-varying respiratory phase signal; identifying peaks on the time-varying respiratory phase signal using the calculated intercepts, wherein each of the peaks is identified as a maximum between an intercept having a positive slope and a subsequent intercept having a negative slope; and The respiratory cycle was calculated as the duration between two adjacent peaks.

13. A system according to claim 11 or claim 12, wherein: The data storage further stores computer program instructions operable to cause the processor to: calculating a first-order time derivative waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; setting all negative values ​​of the first-order time derivative waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to zero to form a composite waveform of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; squaring the composite waveform associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a squared signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; filtering the squared signal using a moving average filter having a predetermined time window to produce an integrated signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; detecting a peak in the integrated signal associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; detecting peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; using the detected peaks in the integrated signal to validate the detected peaks in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; calculating beat positions in the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal, each of the beat positions being the nearest preceding positive zero intercept relative to each verified peak of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; and A beat-to-beat interval associated with each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal is calculated, the beat-to-beat interval being a time interval between consecutive beat positions, wherein the beat-to-beat interval associated with the time-varying cardiac phase signal is related to heart rate and the beat-to-beat interval associated with the time-varying radial artery pulse phase signal is related to radial artery pulse rate.

14. A system according to any one of claims 11 to 13, wherein: The data storage further stores computer program instructions operable to cause the processor to: receiving the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; processing the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a processed time-varying cardiac phase signal and a processed time-varying radial artery pulse phase signal; generating an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal based on the processed time-varying cardiac phase signal and the processed time-varying radial artery pulse phase signal using a trained machine learning model, wherein a peak of the aligned time-varying cardiac phase signal corresponds to an electrocardiogram (ECG) R wave peak, and a peak of the aligned time-varying radial artery pulse phase signal corresponds to a photoplethysmography (PPG) maximum first derivative (MFD) point; calculating a pulse transit time (PPT) as a time delay between one of the peaks of the aligned time-varying cardiac phase signal and a corresponding one of the peaks of the aligned time-varying radial artery pulse phase signal; and The calculated PPT is converted into systolic and diastolic blood pressure values.

15. The system of claim 14, wherein: The data storage further stores computer program instructions operable to cause the processor to: In a subsequent time window of 0.15s to 0.4s of the one of the peaks of the aligned time-varying cardiac phase signal, a local maximum of the aligned time-varying radial artery pulse phase signal is searched, the local maximum being the corresponding one of the peaks used to calculate the PPT.

16. A system according to claim 14 or claim 15, wherein: The data storage further stores computer program instructions operable to cause the processor to: receiving training data, wherein the training data includes a training time-varying cardiac phase signal and a training time-varying radial artery pulse phase signal; processing the training data to form a training processed time-varying cardiac phase signal and a training processed time-varying radial artery pulse phase signal; and training a machine learning model to form a trained machine learning model, wherein the data storage stores computer program instructions operable to cause the processor to train the machine learning model, and the data storage further stores computer program instructions operable to cause the processor to: generating, using the machine learning model, a training time-varying cardiac phase signal output and a training time-varying radial artery pulse phase signal output based on the training processed time-varying cardiac phase signal and the training processed time-varying radial artery pulse phase signal; and Using a regression layer, the mean square error (MSE) between each of the training time-varying cardiac phase signal output and the training time-varying radial artery pulse phase signal output and the corresponding target time-varying cardiac phase signal and time-varying radial artery pulse phase signal is minimized to form the trained machine learning model.

17. A system according to any one of claims 14 to 16, wherein: The trained machine learning model includes a long short-term memory (LSTM) network followed by a fully connected (FC) layer for each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal.

18. A system according to any one of claims 14 to 17, wherein: The data storage stores computer program instructions operable to cause the processor to process the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal, and the data storage also stores computer program instructions operable to cause the processor to: shifting the time-varying cardiac phase signal left by a predetermined amount of time to form the processed time-varying cardiac phase signal; and The time-varying radial artery pulse phase signal is differentiated with respect to time to generate a time derivative of the time-varying radial artery pulse phase signal to form the processed time-varying radial artery pulse phase signal.

19. A system according to any one of claims 14 to 18, wherein: The data storage further stores computer program instructions operable to cause the processor to: identifying epochs for blood pressure sensing, the identified epochs each being a time window having a predetermined time period during which both the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal are present; calculating a mean heart rate and a mean pulse rate for the identified period; selecting one or more candidate epochs from the identified epochs, wherein an absolute difference between the mean heart rate and the mean pulse rate for each of the one or more candidate epochs is less than two beats per minute; A signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in each of the one or more candidate epochs is calculated. e ) is calculated as: Wherein N is the number of detected beats of the time-varying cardiac phase signal or the time-varying radial artery pulse phase signal in each of the one or more candidate epochs, t e is the length of each of the corresponding one or more candidate epochs, in minutes, and I is the hop-to-hop interval of each consecutive pair from detected hops, in seconds; and One or more detection periods are selected from the one or more candidate periods for continuous blood pressure detection, wherein the signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in the one or more detection periods is e ) is greater than 0.

5.

20. A method for non-contact sensing of a physiological parameter of a body using one or more sensors, wherein: Each of the one or more sensors includes a waveguide, the waveguide including a metamaterial and configured to propagate a transmit signal along the waveguide in an artificial surface plasmon mode to generate an evanescent electromagnetic field and provide a receive signal, the evanescent electromagnetic field being used for contactless sensing of disturbances generated by physiological movement of the body, the method comprising: (i) placing the waveguide at a predetermined distance away from the body for non-contact sensing of the disturbance; (ii) providing the transmit signal to the waveguide; (iii) receiving the receive signal from the waveguide; and (iv) processing the received signal and the transmitted signal to determine a phase shift between the transmitted signal and the received signal caused by the disturbance for use in determining the physiological parameter of the body.

21. The method according to claim 20, wherein: The waveguide includes a sensing layer on a sensing side of the waveguide suitable for detecting disturbances generated by the physiological movement of the body, a ground layer on an opposite side of the sensing side, and a non-conductive layer sandwiched between the sensing layer and the ground layer, wherein the ground layer is configured to confine the evanescent electromagnetic field to the sensing side of the waveguide.

22. The method of claim 20 or claim 21, wherein: The sensing layer includes a comb-like rectangular strip having an elongated base and a plurality of teeth extending along and from the elongated base, wherein adjacent teeth of the plurality of teeth are separated by a gap.

23. The method according to any one of claims 20 to 22, further comprising: generating a digital complex baseband signal; converting the digital complex baseband signal using a digital-to-analog converter (DAC) to form an analog baseband signal; modulating the analog baseband signal with a carrier signal to provide the transmit signal; demodulating the received signal to obtain in-phase and quadrature (IQ) components associated with the digital complex baseband signal; and The obtained IQ components are digitized.

24. The method according to claim 23, further comprising: performing a complex conjugate multiplication of the digital complex baseband signal and the digitized IQ component; and A phase shift signal associated with a phase shift between the transmit signal and the receive signal is determined.

25. The method according to claim 24, further comprising: Filtering the phase-shifted signal using a low-pass filter; and The filtered phase-shifted signal is downsampled to form a decimated phase-shifted signal.

26. The method according to claim 25, further comprising: The extracted phase-shifted signal is subjected to arctan demodulation and unwrapping to obtain a time-varying phase signal associated with the phase shift between the transmit signal and the receive signal.

27. The method according to claim 26, wherein: The one or more sensors include a first sensor and a second sensor, wherein the first sensor is provided on the back of the body and is suitable for detecting a respiratory signal and a cardiac signal associated with the body, and the second sensor is provided on the wrist of the body and is suitable for detecting a radial artery pulse signal associated with the body, and the method further includes: processing a first time-varying phase signal associated with the first sensor with a bandpass filter to separate the first time-varying phase signal into a time-varying respiratory phase signal and a time-varying cardiac phase signal, and A second time-varying phase signal associated with the second sensor is processed with a bandpass filter to obtain a time-varying radial artery pulse phase signal.

28. The method according to claim 27, further comprising: receiving the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal; processing the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal to form a processed time-varying cardiac phase signal and a processed time-varying radial artery pulse phase signal; generating an aligned time-varying cardiac phase signal and an aligned time-varying radial artery pulse phase signal based on the processed time-varying cardiac phase signal and the processed time-varying radial artery pulse phase signal using a trained machine learning model, wherein a peak of the aligned time-varying cardiac phase signal corresponds to an electrocardiogram (ECG) R wave peak, and a peak of the aligned time-varying radial artery pulse phase signal corresponds to a photoplethysmography (PPG) maximum first derivative (MFD) point; calculating a pulse transit time (PPT) as a time delay between one of the peaks of the aligned time-varying cardiac phase signal and a corresponding one of the peaks of the aligned time-varying radial artery pulse phase signal; and The calculated PPT is converted into systolic and diastolic blood pressure values.

29. The method according to claim 28, further comprising: In a subsequent time window of 0.15s to 0.4s of the one of the peaks of the aligned time-varying cardiac phase signal, a local maximum of the aligned time-varying radial artery pulse phase signal is searched, the local maximum being the corresponding one of the peaks used to calculate the PPT.

30. The method of claim 28 or claim 29, further comprising: identifying epochs for blood pressure sensing, the identified epochs each being a time window having a predetermined time period during which both the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal are present; calculating a mean heart rate and a mean pulse rate for the identified period; selecting one or more candidate epochs from the identified epochs, wherein an absolute difference between the mean heart rate and the mean pulse rate for each of the one or more candidate epochs is less than two beats per minute; A signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in each of the one or more candidate epochs is calculated. e ) is calculated as: Wherein N is the number of detected beats of the time-varying cardiac phase signal or the time-varying radial artery pulse phase signal in each of the one or more candidate epochs, t e is the length of each of the corresponding one or more candidate epochs, in minutes, and I is the hop-to-hop interval of each consecutive pair from detected hops, in seconds; and One or more detection periods are selected from the one or more candidate periods for continuous blood pressure detection, wherein the signal quality metric (Q) of each of the time-varying cardiac phase signal and the time-varying radial artery pulse phase signal in the one or more detection periods is e ) is greater than 0.5.

Images