Miniaturized card type human body vital sign monitoring equipment

Abstract

The invention relates to the technical field of intelligent health monitoring, in particular to a miniaturized card type human body vital sign monitoring device which comprises a millimeter wave radar module, a flexible substrate, an ADC conversion unit and a signal processing unit. The millimeter wave radar module comprises an antenna array unit and an antenna substrate. The antenna array unit is etched on the antenna substrate through a folded antenna technology; the flexible substrate is bonded with the antenna substrate; the ADC conversion unit is connected with the millimeter wave radar module; the signal processing unit is connected with the ADC conversion unit and used for extracting human body vital sign signals and calculating medical parameters. The flexible substrate and folded antenna array technology is adopted, the area of a core module of the monitoring equipment is greatly reduced, the size of a credit card can be achieved, the monitoring equipment can adapt to various application scenes, the monitoring equipment is made into a mobile phone shell shape and is coupled with a mobile phone, a monitoring closed loop is achieved, and the monitoring equipment is easy to use and convenient to carry and does not need to be worn on the surface of a human body; and non-contact medical health monitoring can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent health monitoring, and particularly relates to a miniaturized card-type human vital sign monitoring device. Background Art

[0002] In modern life, young people staying up late, Internet cafe users surfing the Internet for overtime, the elderly in need of health protection, and drivers in need of safety guarantee. These scenarios all put forward an urgent need for continuous monitoring of human vital signs. A large amount of data shows that continuous monitoring of vital signs can effectively reduce the incidence risk of heart disease. Currently, there are three core requirements for health monitoring: continuous monitoring, non-intrusive monitoring, and intelligent early warning.

[0003] However, the existing monitoring technologies have shortcomings. Traditional medical devices such as ECG (Electrocardiogram) electrocardiographs are restricted by electrode contact, making it difficult to achieve long-term continuous monitoring, and they have complex operations and isolated data. Most millimeter-wave monitoring devices are fixed, large in size, and difficult to meet the portable requirements. Summary of the Invention

[0004] This application provides a miniaturized card-type human vital sign monitoring device, which adopts a miniaturized card-type form design, combined with a flexible substrate and intelligent interconnection technology, to achieve portable and continuous intelligent health monitoring and meet the needs of modern society for health monitoring.

[0005] Specifically, it is realized through the following technical solutions: This application provides a miniaturized card-type human vital sign monitoring device, including a millimeter-wave radar module, a flexible substrate, an ADC conversion unit, and a signal processing unit; wherein, the millimeter-wave radar module includes an antenna array unit and an antenna substrate; the antenna array unit is etched on the antenna substrate through folded antenna technology; the antenna array unit is used for transmitting a linear frequency modulated continuous wave signal and receiving a heartbeat echo signal based on the linear frequency modulated continuous wave signal; the flexible substrate is bonded to the antenna substrate; the ADC conversion unit is connected to the millimeter-wave radar module and includes a mixer and an analog-to-digital converter; the mixer is used for mixing the linear frequency modulated continuous wave signal and the heartbeat echo signal to generate a heartbeat intermediate frequency signal; the analog-to-digital converter is used for converting the heartbeat intermediate frequency signal into a heartbeat digital signal; the signal processing unit is connected to the ADC conversion unit and is used for extracting human vital sign signals from the heartbeat digital signal and calculating medical parameters according to the human vital sign signals.

[0006] Preferably, the antenna array unit adopts a microstrip patch antenna, and the microstrip patch antenna adopts a serpentine layout conforming to a sine wave curve.

[0007] Preferably, the flexible substrate and the antenna substrate are bonded through oxygen plasma activation and a silane coupling agent.

[0008] Preferably, the antenna array unit is a 4x4 antenna unit, which is embedded between the layers of the flexible substrate through a Z-shaped folding layout. The vertical stacking pitch is 0.3 mm, the equivalent aperture is 4 cm², and the antenna units are connected by a flexible strip line with a bending radius greater than 0.1 mm. The bending loss of the flexible strip line is less than 0.2 dB.

[0009] Preferably, the signal processing unit extracts the human vital sign signals from the heartbeat digital signals, including: the signal processing unit performs range-Doppler analysis on the heartbeat digital signals through the range dimension signal and the Doppler dimension signal generated by the millimeter-wave radar module, and separates the target echo reflection signal, which is used to characterize the human vital signs.

[0010] Preferably, the signal processing unit performs fast Fourier transform analysis on the heartbeat digital signals within a single acquisition period through the range dimension signal to calculate the target distance between the monitoring device and the human body; and performs fast Fourier transform analysis on multiple continuously acquired heartbeat digital signals within a time window through the Doppler dimension signal to calculate the human physiological movement speed; wherein, the number of FFT points for the fast Fourier transform analysis is ; the signal processing unit uses the target distance and the human physiological movement speed to separate the target echo reflection signal.

[0011] Preferably, the signal processing unit is further configured to output the signal difference between two adjacent heartbeat digital signals collected at adjacent times through the differential operation of the MTI filter, suppress the static component in the target echo reflection signal, and obtain the target echo dynamic reflection signal after eliminating the static clutter interference.

[0012] Preferably, the signal processing unit is further configured to filter the target echo dynamic reflection signal through a band-pass filter to obtain the target echo heart rate reflection signal; and use the fast independent component analysis algorithm to separate the target echo heart rate reflection signal to obtain the target echo heartbeat reflection signal, the target echo respiration reflection signal, and the target echo body movement reflection signal.

[0013] Preferably, the signal processing unit is further configured to: Construct a deep learning model integrating a one-dimensional convolutional neural network and a Transformer, and use the deep learning model to extract heartbeat waveform features from the target echo heartbeat reflection signal, extract respiratory waveform features from the target echo breathing reflection signal, and extract body movement waveform features from the target echo body movement reflection signal; use the heartbeat waveform features, respiratory waveform features, body movement waveform features, and heartbeat echo signal to train a generative adversarial network; wherein, the discriminator of the adversarial network outputs the authenticity probability value of the heartbeat echo signal; input the human vital sign signal into the deep learning model to output the heart rate, respiratory rate, and heart rate variability.

[0014] Preferably, the monitoring device further includes a communication module, and the communication module is used to send the medical parameters to the mobile terminal.

[0015] Preferably, the monitoring device is in the shape of a mobile phone case as a whole, and the mobile terminal includes a smart phone; wherein, the monitoring device in the shape of a mobile phone case is coupled with the smart phone.

[0016] As can be seen from the above, compared with the prior art, the present application has the following beneficial technical effects: The present application adopts a flexible substrate and a folded antenna array technology, which greatly reduces the area of the core module of the monitoring device, can reach the size of a credit card, is simple to use and convenient to carry; and an antenna array is formed by high-frequency millimeter wave technology, which can adapt to a variety of application scenarios, does not need to be worn on the human body surface, can realize non-contact medical-grade health monitoring, and seamlessly integrates into daily life scenarios for all-weather health management. For example, the micro-card type monitoring device of the present application is embedded at the hinge of a laptop, about 0.6m away from the human chest, and the human vital sign indicators can be monitored through millimeter waves penetrating clothes. At the same time, it is not restricted by the electrode contact of traditional ECG electrocardiographs, can realize long-term continuous monitoring, and is simple to operate.

[0017] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0019] Figure 1Schematic structural diagram of a miniaturized card-type human vital sign monitoring device provided by an embodiment of the present application; Figure 2 Schematic flow chart of a miniaturized card-type human vital sign monitoring method provided by an embodiment of the present application; Figure 3 Schematic flow chart of a device behavior state detection and recognition method provided by an embodiment of the present application. Detailed implementation manners

[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0021] Please refer to Figure 1 , Figure 1 which is a schematic structural diagram of a miniaturized card-type human vital sign monitoring device provided by an embodiment of the present application. As Figure 1 shown, the monitoring device includes: a millimeter-wave radar module 100, a flexible substrate 200, an ADC conversion unit 300, and a signal processing unit 400. Among them, the millimeter-wave radar module 100 includes an antenna array unit and an antenna substrate; the antenna array unit is etched on the antenna substrate through a folded antenna technology; the antenna array unit is used to transmit a frequency-modulated continuous wave signal and receive a heartbeat echo signal based on the frequency-modulated continuous wave signal; the flexible substrate 200 is bonded to the antenna substrate; the ADC conversion unit 300 is connected to the millimeter-wave radar module 100 and includes a mixer and an analog-to-digital converter; the mixer is used to mix the frequency-modulated continuous wave signal and the heartbeat echo signal to generate a heartbeat intermediate-frequency signal; the analog-to-digital converter is used to convert the intermediate-frequency signal into a heartbeat digital signal; the signal processing unit 400 is connected to the ADC conversion unit and is used to extract human vital sign signals from the heartbeat digital signals and calculate medical parameters based on the human vital sign signals.

[0022] Specifically, the millimeter-wave radar module 100 uses an integrated millimeter-wave radar SOC (System-on-Chip for Millimeter-Wave Radar) in the 77 GHz band. The chip supports the FMCW (Frequency-Modulated Continuous Wave) frequency-modulated continuous wave mode, with a bandwidth ≥ 4 GHz and a range resolution ≤ 1 mm.

[0023] The antenna array unit is designed as a 4x4 antenna array. By using two transmitters and four receivers to form a virtual 4×4 MIMO array, it is beneficial to improve the spatial resolution. A microstrip patch antenna can be used. The microstrip patch antenna is etched on a polyimide substrate. The polyimide substrate serves as the antenna base, with a thickness <0.5 mm and a dielectric constant ε = 3.5.

[0024] The flexible substrate 200 is selected as a biocompatible flexible material PDMS (Polydimethylsiloxane, a silicone-based polymer flexible material), with a thickness of 0.5 mm and a elongation rate >50%. The PDMS flexible substrate adopts a multi-layer heterogeneous material composite structure, taking into account flexibility, high-frequency signal transmission and mechanical strength. The dielectric constant ε of the PDMS flexible substrate with a thickness of 0.5 mm is approximately 2.7, and it is formed on a glass carrier through a spin coating process and peeled off after curing. A 10-μm-thick polyimide film is laminated on the surface of the PDMS as the base of the above 4x4 antenna array to reduce dielectric loss.

[0025] In some feasible embodiments, PDMS and the polyimide film are bonded through oxygen plasma activation and a silane coupling agent, with an adhesion force >2 N / cm² to avoid interlayer peeling.

[0026] In some feasible embodiments, a copper mesh (mesh density 200 mesh) is embedded in the bottom layer of the PDMS to shield external electromagnetic interference while maintaining the bendability of the substrate.

[0027] In some feasible embodiments, the monitoring device uses a flexible solid-state battery with a capacity of 50 mAh. The operating power consumption of the millimeter-wave radar SOC chip <100 mW, and the standby power consumption <1 mW.

[0028] In some feasible embodiments, the millimeter-wave radar SOC chip can also integrate a Bluetooth 5.0 module, which can transmit data to mobile terminals such as mobile phones / tablets in real time, facilitating users to view the human vital sign indicators of the monitored object in real time.

[0029] In some feasible embodiments, the folding antenna technology is used to embed the above 4x4 antenna array in the PDMS interlayer in a Z-shaped folding layout, with a vertical stacking pitch of 0.3 mm, and the equivalent aperture is compressed to 4 cm². A serpentine wire layout is embedded in the PDMS interlayer to avoid circuit breakage during bending. Specifically, the microstrip patch antenna uses a sine wave curve trace (wavelength λ = 0.5 mm, amplitude A = 0.2 mm), and the elongation rate is increased to more than 50%. Inkjet-print silver nanowires (wire width 20 μm, thickness 5 μm), resistivity <5×10⁻ 8Ω·m. The vertical interconnection between PDMS layers is achieved by filling conductive silver paste through laser drilling (aperture 30 μm), and the impedance matching error is <5%. Among them, the operating frequency of the microstrip patch antenna is 77 GHz, and the size is 1.5 mm × 1.5 mm. Each of the antenna units is connected by a flexible strip line with a bending radius greater than 0.1 mm, and the bending loss of the flexible strip line is less than 0.2 dB.

[0030] In some feasible embodiments, the monitoring device further includes a communication module 500, and the communication module is used to send the medical parameters to a mobile terminal. It can be understood that the medical parameters can also be sent to the cloud to realize automatic synchronization of monitoring data to the medical cloud platform, provide periodic health trend reports and professional suggestions, and can act as the role of an AI doctor in the cloud.

[0031] In some feasible embodiments, the monitoring device is in the shape of a mobile phone case as a whole, and the mobile terminal includes a smart phone; among them, the monitoring device in the shape of a mobile phone case is coupled with the smart phone to realize a monitoring closed loop. That is, the miniaturized card-type human vital sign monitoring device described in this application is manufactured in the shape of a mobile phone case and is coupled with the smart phone. The high-frequency millimeter wave of this device can penetrate the mobile phone shell and clothes to extract the heart rate or respiratory rate, realizing non-contact continuous monitoring of human vital signs. For example, when a user plays games / watch short videos for a long time, it is easy to cause hidden health risks such as a sudden increase in heart load, respiratory alkalosis, or missed detection of premonitory signs of sudden death. Through the method described in the embodiments of this application, non-contact continuous monitoring can be realized during the user's use of the smart phone. If it is detected that the user's vital sign index exceeds the preset risk threshold, an early warning prompt will be sent through the smart phone, which is beneficial to reducing the hidden health risks caused by the user's long-term gaming or short video viewing.

[0032] The monitoring device described in this application adopts a PDMS flexible substrate with a thickness of 0.5 mm and a folded antenna array technology, so that the area of the core module can be reduced to the size of a credit card; it is simple to use and convenient to carry; and an antenna array is formed through high-frequency millimeter wave technology, which can adapt to various application scenarios, without the need to be worn on the human body surface, and can realize non-contact medical-grade health monitoring, seamlessly integrating into daily life scenarios for all-weather health management. At the same time, it is not restricted by the electrode contact of traditional ECG electrocardiographs, can realize long-term continuous monitoring, and is easy to operate.

[0033] In summary, the monitoring device described in this application is applicable to a variety of application scenarios, such as: The decline in cardiopulmonary function caused by long-term sitting, the hidden heart rate abnormalities caused by work pressure, and the inability of traditional devices to continuously monitor in the office scenario. Embed the card-type monitoring device described in this application at the rotating shaft of a laptop computer (0.6 m away from the chest), and monitor key indicators through millimeter waves penetrating clothes.

[0034] Baby monitors require wristbands, which can easily cause skin sensitivity; camera monitoring poses a risk of privacy leakage. The card-type monitoring device described in this application can be embedded in the crib railing (at a distance of 0.8 - 1.2 m), and the breathing rhythm can be monitored through millimeter waves penetrating the quilt.

[0035] Heart rate belts constrict the chest and affect athletic performance, and the detection accuracy of smartwatches decreases due to sweat. Therefore, a magnetic attachment method can be used to attach the card-type monitoring device described in this application to the outside of sportswear to dynamically track vital signs during running / swimming.

[0036] Atrial fibrillation patients need continuous monitoring of abnormal heartbeats, and COPD patients need to track changes in breathing patterns. Therefore, the card-type monitoring device described in this application can be embedded in a smart medicine box (at a distance of 0.5 m from the chest), and vital signs fluctuations can be monitored synchronously during the medication period. A curve of vital sign changes is automatically generated for doctors to view remotely. In case of cardiac arrest, the community emergency response can be automatically called.

[0037] Fatigue driving among long-haul truck drivers accounts for 37% of accidents, and existing DMS systems have a high false alarm rate (30% missed detections due to facial occlusion). The card-type monitoring device described in this application can be embedded in the steering wheel (at a distance of 0.3 m from the driver's chest), and millimeter waves penetrate the clothing for continuous monitoring.

[0038] Users are prone to hidden health risks during long hours of gaming / short video browsing: ① A sudden surge in heart load: Competitive games cause abnormal heart rhythms (>120 BPM for 10 minutes).

[0039] ② Respiratory alkalosis: Tension-induced hyperventilation (respiratory rate > 25 / minute).

[0040] ③ Missed detection of precursors to sudden death: Sudden ventricular arrhythmia when using a mobile phone alone at night.

[0041] The micro millimeter-wave monitoring card can be integrated into the phone case, and millimeter waves penetrate the phone case and clothing to extract the heart rate / respiratory rate.

[0042] The above are typical application scenarios, and users can extend them to more scenarios as needed. The application scenarios of this invention are extremely extensive.

[0043] Its working principle is as follows: Please refer to Figure 2 , Figure 2 which is a schematic flowchart of a miniaturized card-type human vital sign monitoring method provided by an embodiment of this application. As Figure 2 shown, it includes the following: 201. The millimeter-wave radar module generates a linear frequency-modulated continuous wave signal.

[0044] The millimeter-wave radar SOC chip integrated in the above millimeter-wave radar module 100 generates a frequency-modulated continuous wave (FMCW) signal. The frequency starts from 77 GHz and linearly increases to 81 GHz within 1 ms at a frequency modulation slope of 2×10^12 Hz / s, forming a single Chirp signal (i.e., a chirp signal). That is, the duration of the Chirp signal is Tc = 1 ms. The Chirp signal is amplified to 12 dBm by a GaAs (Gallium Arsenide Power Amplifier) power amplifier and drives a 4×4 MIMO antenna array to transmit this millimeter-wave signal, that is, to transmit a frequency-modulated continuous wave signal. This millimeter-wave signal penetrates clothing (attenuation is about 3 dB) and the skin surface layer (penetration depth is 8 cm) and reaches the surface of the heart and chest cavity. The heartbeat causes periodic displacement of the chest cavity (amplitude 0.1 - 0.5 mm), resulting in a change in the phase of the reflected wave (a phase difference of 0.12° corresponds to a displacement of 0.1 mm).

[0045] For the above-mentioned transmitted signal modeling, the mathematical expression of the FMCW signal can be: ; Among them, is the FMCW signal, expressed as a time-domain signal; A is the amplitude of the FMCW signal; is the starting frequency of the FMCW signal, that is, the above-mentioned 77 GHz.

[0046] S is the frequency modulation slope, that is, the above-mentioned 2×10^12 Hz / s; t is the time variable (unit: second).

[0047] 202. The millimeter-wave radar module receives the heartbeat echo signal based on the above-mentioned frequency-modulated continuous wave signal.

[0048] The 4-channel receiving antennas capture the reflected signal based on the above-mentioned frequency-modulated continuous wave signal, and amplify the weak echo through a low-noise amplifier (LNA, gain 25 dB).

[0049] The mixer of the ADC conversion unit mixes the heartbeat echo signal with the transmitted frequency-modulated continuous wave signal to generate a heartbeat intermediate frequency signal.

[0050] The heartbeat echo signal is delayed by τ = , then the mathematical expression of the heartbeat intermediate frequency signal after mixing can be: ; After ignoring the higher-order terms, the frequency of the heartbeat intermediate frequency signal is: ; Among them, S is the frequency modulation slope; is the delay time of the heartbeat echo signal; R is the target distance, where the target distance represents the physical distance between the monitoring device and the human body (heart or chest); c is the speed of light.

[0051] Among them, Sτt represents the linear phase term related to the target distance R and speed v, which determines the frequency of the intermediate frequency signal of the heartbeat; f0τ represents the fixed phase offset caused by the starting frequency of the FMCW signal and the delay of the heartbeat echo signal; represents the high-order phase term (quadratic term), which can be ignored at short distances (τ is extremely small).

[0052] The frequency difference of the intermediate frequency signal of the heartbeat is proportional to the target distance, as shown in the following formula: ; Among them, Δf is the frequency difference of the intermediate frequency signal of the heartbeat.

[0053] 204. The analog-to-digital converter of the ADC conversion unit is used to convert the intermediate frequency signal of the heartbeat into a digital heartbeat signal.

[0054] The intermediate frequency signal of the heartbeat is sampled at a rate of 4 GSPS by a 14-bit ADC and converted into a digital signal. Digital down-conversion (DDC) shifts the signal to the baseband, and the data rate is reduced to 400 kSPS after decimation filtering.

[0055] 205. The signal processing unit performs range-Doppler analysis on the digital heartbeat signal.

[0056] Range and Doppler dimension signals are generated by the millimeter-wave radar SOC chip to separate the reflected signals from the heart area. Including: (1) The signal processing unit performs fast Fourier transform analysis on the digital heartbeat signals within a single acquisition period through the range dimension signal, and calculates the target distance between the monitoring device and the human body.

[0057] Perform 1024-point FFT (Fast Fourier Transform) analysis on each Chirp signal to calculate the target distance. The calculation formula is as follows: ; Among them, bin peak is the bin number where the peak is located in the FFT spectrum; fs is the sampling rate (4 GSPS, that is, 4×109 Samples / s) S is the frequency modulation slope (2×10 12 Hz / s) is the number of points in the first FFT.

[0058] Each chirp is subjected to 1024-point FFT analysis, that is, the 1024 time-domain signal points collected in a single chirp cycle are subjected to fast Fourier transform analysis. By calculating the target distance, a distance profile is generated, and a distance threshold (e.g. 0.3-1.2m) is set to separate the reflected signals in the heart or chest area and filter out non-human targets.

[0059] (2) The Doppler signal is used to calculate the time window The fast Fourier transform analysis of multiple heartbeat digital signals collected continuously in the system is performed to calculate the physiological movement speed of the human body; the number of FFT points of the fast Fourier transform analysis is .

[0060] For example, for 128 consecutive chirp signals (time window ) to perform 256-point FFT analysis to calculate the physiological movement speed of the human body (for example, including the heart beat speed: the periodic displacement speed of the chest cavity caused by the heart contraction / diastole, and the respiratory movement speed: the chest cavity rise and fall speed caused by the expansion and contraction of the lungs). The calculation formula is as follows: ; in, ; in, λ is the wavelength of millimeter wave (3.9 mm); fd is the Doppler frequency shift; PRF is the pulse repetition frequency, which is the number of pulses emitted per second; ; bin peak is the bin number where the peak value in the FFT spectrum is located; is the number of points of the second FFT.

[0061] By performing 256-point FFT analysis on 128 consecutive chirps, speed information is extracted (resolution 0.05m / s) and information other than human physiological movement speed is filtered out.

[0062] 206. The signal processing unit uses the target distance and the physiological movement speed of the human body to separate and obtain the target echo reflection signal, and the target echo reflection signal is used to characterize the vital signs of the human body.

[0063] 207. The signal processing unit outputs the signal difference between two heartbeat digital signals collected at adjacent moments through the differential operation of the MTI filter, suppresses the static component in the target echo reflection signal, and obtains the target echo dynamic reflection signal after eliminating the static clutter interference.

[0064] Use an MTI (Moving Target Indication) filter to eliminate the interference of stationary objects in the environment, such as static background reflections from walls, furniture, etc. Its mathematical expression can be represented by a transfer function: ; Among them, H(z) describes the frequency-domain characteristics of the filter, and suppresses the static component by taking the difference between the current signal and the signal at the previous sampling point; represents the unit delay operation (i.e., the signal at the previous sampling point).

[0065] Differential operation: y[n]=x[n]−x[n−1] Among them, x[n] is the input signal at the current moment (the nth sampling point); x[n−1] is the input signal at the previous moment (the (n−1)th sampling point); y[n] is the filtered output signal, representing the signal difference between two adjacent sampling points.

[0066] Frequency response: ; Among them, f is the frequency of the input signal (unit: Hz).

[0067] T is the pulse repetition period (unit: second), T = 1 / PRF.

[0068] The cut-off frequency is 0.1 Hz, which suppresses the static component with f < 0.1 Hz, eliminates static background reflections from walls, furniture, etc., and retains the heartbeat / respiration dynamic signal to obtain the target echo dynamic reflection signal.

[0069] 208. After the signal processing unit filters the target echo dynamic reflection signal through a band-pass filter, a target echo heart rate reflection signal is obtained; For example, by setting the parameters of the band-pass filter, the frequency band of 0.5~5 Hz is retained, corresponding to the heart rate signal of 30~300 BPM.

[0070] 209. The signal processing unit uses the fast independent component analysis algorithm to separate the target echo heart rate reflection signal, obtaining the target echo heartbeat reflection signal, the target echo respiration reflection signal, and the target echo body movement reflection signal.

[0071] (1) Preprocess the target echo heart rate reflection signal: Center the target echo heart rate reflection signal: ; Among them, is the original observation signal matrix, with dimensions N×T, where N is the number of sensor channels and T is the number of time samples; is the mean vector of each channel signal (dimension N×1), used to eliminate the DC component in the signal.

[0072] Whiten the target echo heart rate reflection signal: ; Among them, Wwhiten is the whitening matrix, obtained through the eigenvalue decomposition of the covariance matrix; Wwhiten = Λ−1 / 2·UT; U and Λ are the eigenvector matrix and eigenvalue diagonal matrix of Xcentered· The Z is the signal matrix after whitening processing, satisfying Z·ZT = I, eliminating the second-order correlation between channels.

[0073] (2) Iterative optimization: Use the maximum non-Gaussianity (approximate negative entropy) function as the objective function: ; Among them, ω is the separation weight vector, with dimensions m×1; G(u) is a non-linear function (logcosh(u)), G(u) = logcosh(u); v is a standard Gaussian variable (mean 0, variance 1), used to compare non-Gaussianity.

[0074] E[⋅] is the mathematical expectation (statistical average).

[0075] Weight update: ; g(ωTZ) = tanh(ωTZ): the derivative of the non-linear function.

[0076] g′(ωTZ) = 1−tanh2(ωTZ): the second derivative (implied in the formula).

[0077] Normalization: ω = w+ / ||w+|| to ensure the unitization of the weight vector.

[0078] Until convergence.

[0079] (3)Separate the signals: The column vectors of the demixing matrix W are independent components, and the signal S = W·Z.

[0080] W is an m×m matrix, and each column corresponds to a separation weight vector ω.

[0081] S is the separated independent signal component matrix, with dimensions m×n.

[0082] Output the target echo heartbeat reflection signal, target echo respiration reflection signal, and target echo body movement reflection signal components.

[0083] 210. The signal processing unit calculates medical parameters.

[0084] The signal processing unit constructs a deep learning model that fuses a one-dimensional convolutional neural network and a Transformer, and uses the deep learning model to extract heartbeat waveform features from the target echo heartbeat reflection signal, extract respiration waveform features from the target echo respiration reflection signal, and extract body movement waveform features from the target echo body movement reflection signal; uses the heartbeat waveform features, respiration waveform features, body movement waveform features, and heartbeat echo signal to train a generative adversarial network; wherein, the discriminator of the adversarial network outputs the authenticity probability value of the heartbeat echo signal; inputs the human vital sign signal into the deep learning model, and outputs the heart rate, respiration rate, and heart rate variability.

[0085] (1)1D-CNN (One-Dimensional Convolutional Neural Network) and Transformer fusion model: Input layer: The input size is 512×4 (512 time steps, 4 antenna channels).

[0086] 1D-CNN convolutional layer: 64 filters, kernel size 5, stride 1, and the activation function ReLU is: ; where i is the output time step index (range: 0 ≤ i < 508); j is the filter index (output channel, range: 0 ≤ j < 64); k is the time position of the convolutional kernel (0 ≤ k < 5); d is the input channel index (range: 0 ≤ d < 4); is the weight of the convolutional kernel at time position k and input channel d; is the value of the input data at time step (i + k) and channel (j - d); b is the bias term.

[0087] Output layer: The output size is 508×64.

[0088] Transformer encoder: Multi-head self-attention (4 heads): ; Among them, Q (Query), K (Key), and V (Value) represent three matrices obtained by linearly transforming the input signal, representing query, key, and value respectively; the input feature dimension is 64 (corresponding to 64 in the input size of 508×64 of the Transformer encoder). The dimension of each attention head is dk (here dk = 64 / 4 = 16), indicating that the 64-dimensional input is split into 4 independent self-attention heads (16 dimensions for each head). dk represents the scaling factor, which is used to prevent the dot product result from being too large and causing the gradient to disappear.

[0089] The feed-forward network (FFN) is: ; Among them, W1 and W2 represent learnable weight matrices; W1 maps the input from 64 dimensions to a higher dimension, and after being activated by ReLU, it is mapped back to 64 dimensions through W2. b1 and b2 represent bias terms.

[0090] The output size of the feed-forward network (FFN) is 508×64.

[0091] Global average pooling: Pooling along the time dimension 508×64 → 64 Fully connected layer: Output three medical parameters of heart rate (HR), respiratory rate (BR), and heart rate variability (HRV).

[0092] Training configuration: Loss function: Mean squared error (MSE) and cross-entropy loss of the heart rate peak; Optimizer: Adam (learning rate 3×10−4).

[0093] (2) Generative adversarial network (GAN): Network structure: Input of the generator: Millimeter wave signal x ∈ R512×4; Contains 5 layers of 1D transposed convolution, upsampling to the ECG waveform y ∈ R512; Final activation function: Tanh (output normalized to [-1, 1]); Discriminator: Input: real ECG or generated signal y; Structure: 4 layers of 1D convolution (kernel size 5, stride 2) + fully connected layer; Output: probability value p ∈ [0, 1]; The probability value p ∈ [0, 1] represents the discriminator's judgment result on the authenticity of the input signal. If p = 1, the discriminator believes that the input signal is a real medical-grade physiological signal; if p = 0, the discriminator believes that the input signal is an analog signal forged by the generator.

[0094] Loss function: Wasserstein GAN loss (stable training): ; where E represents the mathematical expectation, which means taking the average of the losses for a batch of samples; D represents the discriminator network, which outputs a score for the authenticity of the input signal (the larger the value, the more real); G(x) represents the generator network, which converts the input millimeter-wave signal x into an analog ECG waveform; x represents the input of the generator, with dimensions x ∈ R512×4, representing the signal of 512 time steps and 4 antenna channels obtained from the millimeter-wave radar.

[0095] LDisc = E[D(y)] − E[D(G(x))] + λ · gradient penalty; where E[D(y)] represents the expected value of the discriminator's score for the real ECG signal y; y represents the real ECG waveform, with dimensions y ∈ R512; E[D(G(x))] represents the expected value of the discriminator's score for the generated signal G(x); λ represents the gradient penalty coefficient, which controls the weight of the gradient penalty term; the gradient penalty represents a regularization term that constrains the discriminator's gradient to prevent training collapse.

[0096] In addition, the signal processing unit of the monitoring device described in this application is also used to: dynamically adjust the filter parameters of the 1D-CNN convolutional layer using the Recursive Least Squares (RLS) algorithm to adapt to different user body fat rates and posture changes to complete adaptive filtering. When the monitoring device described in this application is used for the first time, a user's heartbeat waveform template is established through a 30-second benchmark test, and user personalized modeling is completed through Dynamic Time Warping (DTW) matching during subsequent monitoring.

[0097] The specific implementation is as follows: (1) Recursive Least Squares (RLS) filtering: First, initialize the filter: (0) = 0; ; Among them, represents the initial value of the filter weight; represents the initial value of the inverse covariance matrix; δ is the regularization parameter; I is the identity matrix.

[0098] Then, perform iterative updates: Verify the error through the following formula to measure the deviation between the prediction result of the current weight estimate ω(n−1) for the input signal vector x(n) and the true value d(n): e(n) = d(n) − (n−1)Tx(n); Among them, x(n) represents the input signal vector at the nth moment; d(n) represents the expected output at the nth moment.

[0099] Dynamically adjust the gain vector for weight update through the following formula: ; Among them, represents the product of the inverse covariance matrix P(n - 1) and the input x(n), reflecting the spatial distribution of the input signal; contains the forgetting factor λ and the projection of the input signal, used to normalize the gain.

[0100] Finally, update the weight through the following formula: ; Among them, represents the updated filter weight.

[0101] Adjust the weight through the Kalman gain and the error to make the prediction closer to the true value.

[0102] Update the inverse covariance matrix through the following formula: ; Among them, The forgetting factor λ = 0.99, and λ - 1 is used to exponentially decay the influence of old data; represents the updated inverse covariance matrix.

[0103] The outer product term k(n)x(n)ᵀP(n−1) can reduce the estimation uncertainty.

[0104] Dynamically adjust the filter parameters to adapt to the signal attenuation differences caused by different users' body fat percentages and posture changes, and complete adaptive filtering.

[0105] (2)Dynamic Time Warping (DTW) When the user uses it for the first time, a personalized heartbeat waveform template, that is, the template signal, is established through a 30 - second benchmark test, as follows: T = { t 1 ,t 2 ,...,t M}; where, t i represents the amplitude value of the i th sampling point of the template signal (such as the heartbeat waveform voltage); M is the total number of sampling points of the template signal (i.e., the time window length).

[0106] The currently detected heartbeat waveform segment, that is, the real - time signal, is: S = { s 1 ,s 2 ,...,s N } ; where, s j represents the amplitude value of the j th sampling point of the real - time signal (such as the heartbeat waveform voltage); N is the total number of sampling points of the real - time signal (i.e., the time window length).

[0107] Construct a cost matrix D based on the template signal and the real - time signal, as follows: D ∈ RM×N; where, ; It can be seen that the cost matrix D is a local distance matrix between each point of the template signal and the real - time signal. Among them, represents the absolute difference between the i th point of the template signal and the j th point of the real - time signal, reflecting local similarity.

[0108] Then, solve the optimal path through the recurrence formula for dynamic programming, as shown in the following formula: ; represents the minimum cumulative cost from the starting point (1,1) to the current point (i,j); Dacc(i−1,j) represents the cumulative cost of moving from the left to the current point; Dacc(i,j−1) represents the cumulative cost of moving from above to the current point; Dacc(i−1,j−1) represents the cumulative cost of moving from the diagonal to the current point.

[0109] Path constraints: monotonicity, continuity, and boundary alignment.

[0110] Finally, the similarity score is calculated through the minimum cumulative cost, as shown in the following formula: ; where represents the total difference after the template signal is aligned with the real-time signal. The smaller its value, the higher the similarity.

[0111] Finally, it is determined through the threshold θ. If score < θ, it is determined as a normal heartbeat waveform.

[0112] In some feasible embodiments, to achieve all-weather battery life, the monitoring device described in this application further includes an energy consumption management module for adaptively optimizing the device power consumption through mode division and state migration.

[0113] In the embodiments of this application, the intelligent hierarchical energy consumption management of the device is mainly described in a three-level sensing mode.

[0114] The first level, continuous monitoring mode (100mW): full-function medical-grade monitoring, suitable for first aid or abnormal tracking. The triggering conditions include active enabling, or when events such as HRV≥50ms or apnea≥10 seconds are detected, it is automatically enabled.

[0115] The second level, intermittent monitoring mode (25mW): daily health monitoring, automatically switches after the user is determined to be stationary by a 6-axis sensor. The radar sampling rate (initial set value is 80Hz) drops to 10Hz, and redundant RF channels are turned off (retaining a 2T4R array).

[0116] The third level, triggered monitoring mode (1mW): only basic signal detection is maintained when the device is idle, the millimeter-wave radar is in deep sleep, and the 6-axis motion sensor and MCU perform low-power duty.

[0117] Through the built-in 6-axis motion sensor, the device state (handheld / desktop / worn) can be automatically recognized to switch to the optimal working mode.

[0118] In some feasible embodiments, to achieve multi-scenario adaptive monitoring, the monitoring device described in this application further includes a behavior state recognition module for detecting and identifying the behavior state of the device. A multi-modal perception network is realized by integrating a 6-axis motion sensor (a three-axis accelerometer and a three-axis gyroscope). The hardware configuration is specifically implemented using an Inertial Measurement Unit (Micro-Electro-Mechanical Systems Inertial Measurement Unit, MEMS IMU). The range of this IMU is dynamically adjustable, with an accelerometer of ±8g / ±16g and a gyroscope of ±250dps / ±2000dps. The behavior state recognition module is connected to the main control MCU (Microcontroller Unit) of the signal processing unit through the SPI bus (Serial Peripheral Interface), supporting a sampling rate of 100Hz and DMA (Direct Memory Access) transmission to ensure that the timing synchronization error < 1ms. The acceleration signal is filtered by an IIR low-pass filter (cutoff frequency 5Hz), and the gyroscope data is applied with a sliding average window (window length 10 points) to achieve dynamic noise reduction. When the device is started for the first time, 9-axis calibration (combined with a geomagnetic sensor) is performed to eliminate the installation attitude deviation.

[0119] Please refer to Figure 3 , Figure 3 which is a schematic flow chart of the device behavior state detection and identification method provided by the embodiment of this application. As Figure 3 shown, the steps for the behavior state recognition module to detect and identify the behavior state of the device will be described in detail below: 301. Calculate the variance of the device acceleration modulus.

[0120] The calculation formula is as follows: ; Where is the variance of the device acceleration modulus; N is the number of sampling points within the time window; is the combined modulus of the three-axis acceleration at the i-th sampling point; is the mean value of the acceleration modulus within the time window.

[0121] 302. Determine whether is less than a preset variance threshold (which can be set to in the embodiment of this application); if so, go to step 303, otherwise go to step 305.

[0122] 303. Calculate the pitch angle of the device.

[0123] The pitch angle of the device is calculated in real time by the quaternion method, and the calculation formula is as follows: ; where, is the pitch angle of the device; is the x-axis component of the triaxial accelerometer; is the y-axis component of the triaxial accelerometer; is the z-axis component of the triaxial accelerometer.

[0124] 304. Determine whether the pitch angle of the device is equal to 0. If so, identify the behavior state of the device as the desktop state; otherwise, go to step 305.

[0125] 305. Perform FFT analysis on the acceleration Z-axis signal, and extract the energy ratio (i.e., the periodicity index) of the preset target frequency band (which can be set to 1 - 4 Hz in the embodiments of the present application).

[0126] 306. Determine whether the energy ratio is less than the preset ratio threshold (which can be set to 0.7 in the embodiments of the present application). If so, identify the behavior state of the device as the wearable state; otherwise, identify the behavior state of the device as the handheld state.

[0127] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0128] The above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) having similar functions disclosed in the present invention.

Claims

1. A miniaturized card-type human vital sign monitoring device, characterized in that, The monitoring device includes a millimeter-wave radar module, a flexible substrate, an ADC conversion unit, and a signal processing unit; wherein, The millimeter-wave radar module includes an antenna array unit and an antenna substrate; the antenna array unit is etched on the antenna substrate by a folded antenna technique; the antenna array unit is used to transmit a linear frequency modulation continuous wave signal and receive a heartbeat echo signal based on the linear frequency modulation continuous wave signal; The flexible substrate is bonded to the antenna substrate; The ADC conversion unit is connected to the millimeter-wave radar module and includes a mixer and an analog-to-digital converter; the mixer is used to mix the linear frequency modulation continuous wave signal and the heartbeat echo signal to generate a heartbeat intermediate frequency signal; the analog-to-digital converter is used to convert the heartbeat intermediate frequency signal into a heartbeat digital signal; The signal processing unit is connected to the ADC conversion unit and is used to extract human vital sign signals from the heartbeat digital signal and calculate medical parameters based on the human vital sign signals.

2. The miniaturized card-type human vital sign monitoring device according to claim 1, characterized in that, The antenna array unit uses a microstrip patch antenna, and the microstrip patch antenna adopts a serpentine layout conforming to a sine wave curve.

3. The miniaturized card-type human vital sign monitoring device according to claim 1, characterized in that, The flexible substrate and the antenna substrate are bonded through oxygen plasma activation and a silane coupling agent.

4. A miniaturized card-type human vital sign monitoring device according to claim 1, characterized in that, The signal processing unit extracts human vital sign signals from the heartbeat digital signal, including: The signal processing unit performs range-Doppler analysis on the heartbeat digital signal through the range dimension signal and Doppler dimension signal generated by the millimeter-wave radar module, and separates the target echo reflection signal, and the target echo reflection signal is used to characterize human vital signs.

5. The miniaturized card-type human vital sign monitoring device according to claim 4, characterized in that, The signal processing unit performs fast Fourier transform analysis on the heartbeat digital signals within a single acquisition period through the range dimension signal, and calculates the target distance between the monitoring device and the human body; and, performs fast Fourier transform analysis on multiple heartbeat digital signals continuously acquired within the time window through the Doppler dimension signal, and calculates the physiological movement speed of the human body; wherein, the number of FFT points for the fast Fourier transform analysis is ; the signal processing unit uses the target distance and the physiological movement speed of the human body to separate and obtain the target echo reflection signal.​​​​​​ 6. The miniaturized card-type human vital sign monitoring device according to claim 5, characterized in that, The signal processing unit is also used to output the signal difference between two heartbeat digital signals collected at adjacent times through the differential operation of the MTI filter, suppress the static component in the target echo reflection signal, and obtain the target echo dynamic reflection signal after eliminating static clutter interference.

7. The miniaturized card-type human vital sign monitoring device according to claim 6, characterized in that, The signal processing unit is also used to filter the target echo dynamic reflection signal through a band-pass filter to obtain a target echo heart rate reflection signal; and, use the fast independent component analysis algorithm to separate the target echo heart rate reflection signal to obtain a target echo heartbeat reflection signal, a target echo respiration reflection signal, and a target echo body movement reflection signal.

8. The miniaturized card-type human vital sign monitoring device according to claim 7, characterized in that, The signal processing unit is also used for: Construct a deep learning model that fuses a one-dimensional convolutional neural network and a Transformer, and use the deep learning model to extract heartbeat waveform features from the target echo heartbeat reflection signal, extract respiration waveform features from the target echo respiration reflection signal, and extract body movement waveform features from the target echo body movement reflection signal; Use the heartbeat waveform features, respiration waveform features, body movement waveform features, and heartbeat echo signal to train and generate an adversarial network; wherein, the discriminator of the adversarial network outputs the authenticity probability value of the heartbeat echo signal; Input the human vital sign signal into the deep learning model, and output the heart rate, respiration rate, and heart rate variability.

9. A miniaturized card-type human vital sign monitoring device according to any one of claims 1 to 8, characterized in that, It further includes a communication module, and the communication module is used to send the medical parameters to a mobile terminal.

10. A miniaturized card-type human vital sign monitoring device according to claim 9, characterized in that, The monitoring device as a whole is in the shape of a mobile phone case, and the mobile terminal includes a smart phone; wherein, the monitoring device in the shape of a mobile phone case is coupled to the smart phone.

Images