Non-invasive blood glucose wearable monitoring method and system based on reconfigurable intelligent metasurface and multi-modal perception fusion

Abstract

This invention relates to a non-invasive wearable blood glucose monitoring method and system based on a reconfigurable intelligent metasurface and multimodal sensing fusion. Addressing the shortcomings of existing passive metasurface technology, such as its inability to dynamically adapt to individual differences, lack of active enhancement capability for weak dielectric constant perturbations, and susceptibility to multi-source interference, this invention places a reconfigurable electromagnetic metasurface with integrated adjustable components between a radio frequency antenna and human tissue. Its electromagnetic parameters are dynamically adjusted via external control signals to resonate and enhance weak dielectric constant perturbations caused by blood glucose. Simultaneously, multimodal signals such as temperature, humidity, movement posture, and pulse wave are collected and input into a machine learning model for fusion processing, eliminating environmental and physiological interferences and retrieving high-precision blood glucose concentration. This invention achieves non-invasive, consumable-free, highly robust, long-term continuous blood glucose monitoring with low prediction error.

Description

Technical Field

[0001] This invention relates to the fields of biomedical sensing, Internet of Things and machine learning, and in particular to a non-invasive wearable blood glucose monitoring method and system based on reconfigurable smart metasurfaces and multimodal sensing fusion. Background Technology

[0002] Blood glucose monitoring is a core component of diabetes management. Traditional finger-prick blood sampling methods suffer from pain, infection risks, and poor patient compliance. In recent years, non-invasive blood glucose monitoring technology has become a research hotspot, with radio frequency (RF) signal-based monitoring methods attracting significant attention due to their non-invasiveness, continuous monitoring capability, and low cost. The principle behind this is that changes in blood glucose concentration cause minute alterations in the dielectric constant of tissue fluid, thereby affecting the propagation characteristics of the RF signal, such as the phase and amplitude of the reflection coefficient. However, the change in dielectric constant caused by blood glucose is extremely small, typically within 10-1. -3 The signal-to-noise ratio of blood glucose changes detected directly through antenna reflection is extremely low, and the detection results are easily affected by various factors such as human tissue structure, temperature, humidity, and movement, making it difficult to achieve high-precision and high-stability monitoring.

[0003] To improve detection sensitivity, non-invasive blood glucose monitoring technologies based on radio frequency signals and electromagnetic metasurfaces have been extensively studied in recent years. However, most existing solutions employ passively fixed metasurface structures or directly print or attach metasurfaces to the skin surface. These methods suffer from the following three major drawbacks: (1) Passive fixed structures cannot dynamically adjust electromagnetic characteristics according to individual differences, resulting in poor radio frequency impedance matching.

[0004] (2) It lacks the ability to actively enhance the weak dielectric constant perturbation caused by blood glucose, resulting in a weak signal response.

[0005] (3) In practical applications, it is easily affected by changes in environmental temperature and humidity and human physiological rhythms, such as movement deformation and blood flow pulsation. The signal-to-noise ratio of a single radio frequency signal is extremely low, making it difficult to meet the long-term dynamic monitoring requirements of clinical precision.

[0006] Therefore, there is an urgent need for a smart wearable solution that can actively adjust electromagnetic parameters and integrate multimodal physiological characteristics. Summary of the Invention

[0007] The purpose of this invention is to overcome the defects of the prior art by providing a non-invasive wearable blood glucose monitoring method and system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion. This solves or partially solves the problems of existing non-invasive blood glucose monitoring technologies based on passive electromagnetic metasurfaces, which are unable to dynamically adapt to individual differences, lack the ability to actively enhance weak dielectric constant perturbations, and are susceptible to multi-source interference, resulting in low signal-to-noise ratios and difficulty in meeting long-term dynamic clinical accuracy.

[0008] The objective of this invention can be achieved through the following technical solutions: One aspect of the present invention provides a non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion, comprising the following steps: Radio frequency signals are transmitted to human tissue via a radio frequency antenna; A reconfigurable electromagnetic metasurface is disposed between the radio frequency antenna and human tissue, and the reconfigurable electromagnetic metasurface integrates tunable radio frequency active components. The electromagnetic parameters of the reconfigurable electromagnetic metasurface are dynamically adjusted by external control signals to actively enhance the changes in radio frequency reflection signals caused by the disturbance of the dielectric constant of tissue fluid due to changes in blood glucose concentration. The enhanced radio frequency reflection signal was acquired, and multimodal physiological and environmental signals were acquired simultaneously. The enhanced radio frequency reflection signal and the multimodal physiological and environmental signals are input into a trained machine learning model, and the blood glucose concentration is obtained by fusing multi-source data.

[0009] As a preferred technical solution, the multimodal physiological and environmental signals include at least temperature signals, humidity signals, motion posture signals, and photoplethysmography (PPG) signals.

[0010] As a preferred technical solution, the external control signal is a voltage control signal. By adjusting the bias voltage applied to the reconfigurable electromagnetic metasurface, its equivalent electromagnetic parameters and resonant frequency are changed in real time to adapt to the skin impedance characteristics of different individuals and achieve impedance matching optimization.

[0011] As a preferred technical solution, the operating frequency band of the reconfigurable electromagnetic metasurface is configured in the ISM band of 2.4 GHz, 5.8 GHz or 24-28 GHz.

[0012] As a preferred technical solution, the following inversion relationship is used as the loss to supervise the training of the machine learning model: in, This represents the change in blood glucose concentration obtained through inversion. Dielectric constant of tissue fluid / blood blood glucose concentration intrinsic sensitivity, For changes in radio frequency reflected signals, The dielectric constant of the RF antenna / metasurface The varying electromagnetic response sensitivity , , These represent the parasitic coupling coefficients of radio frequency signals caused by temperature, humidity, and macroscopic motion, respectively. For higher-order nonlinear errors and system thermal noise, , , These are, respectively, changes in temperature, humidity, and macroscopic motion.

[0013] As a preferred technical solution, the radio frequency reflection signal is the phase and / or amplitude change of the scattering parameter S11; the machine learning model is a long short-term memory network or a temporal convolutional neural network.

[0014] Another aspect of the present invention provides a non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion, for implementing the aforementioned non-invasive wearable blood glucose monitoring method, the system comprising: Radio frequency module, including radio frequency transmitting antenna and receiving antenna; A reconfigurable electromagnetic metasurface is disposed between the contact surface between the radio frequency antenna and human tissue. It integrates tunable radio frequency active components and is driven by external control signals to dynamically reconfigure electromagnetic parameters. A multimodal sensor array, used at least to simultaneously acquire temperature, humidity, motion posture, and pulse wave signals; The signal processing unit is used to receive the enhanced radio frequency reflected signal and the signal collected by the multimodal sensor array, and run the trained machine learning model to output the blood glucose concentration value.

[0015] As a preferred technical solution, the reconfigurable electromagnetic metasurface is composed of arrayed microstructure units, each of which physically integrates at least one of a varactor diode and a MEMS RF switch as the tunable RF active component.

[0016] As a preferred technical solution, the system is integrated into a wearable device, which is a smart bracelet.

[0017] As a preferred technical solution, the signal processing unit is a microcontroller or FPGA, which simultaneously outputs the external control signal to adjust the resonant frequency of the reconfigurable electromagnetic metasurface in real time.

[0018] In another aspect, the present invention provides a computer-readable storage medium including one or more programs executable by one or more processors of an electronic device, the one or more programs including instructions for performing the aforementioned non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion.

[0019] Compared with the prior art, the present invention has at least one of the following beneficial effects: (1) Achieving adaptive adjustment for individuals and wearing status: Existing passive fixed structure metasurfaces cannot be dynamically adjusted according to individual skin impedance differences and wearing status, resulting in poor radio frequency impedance matching effect. In this regard, the present invention introduces a reconfigurable electromagnetic metasurface composed of adjustable radio frequency components such as varactor diodes, and its electromagnetic parameters are controlled in real time by external voltage signals. This realizes dynamic impedance matching and personalized electromagnetic coupling optimization for different individuals and wearing status, and enhances the stability and efficiency of energy transmission between radio frequency signals and human tissue.

[0020] (2) Improved sensitivity of blood glucose change detection: The disturbance of the dielectric constant of tissue fluid caused by changes in blood glucose concentration is extremely weak. Traditional passive metasurfaces lack active user calibration capabilities, resulting in extremely weak signal response and difficulty in detection. Therefore, this invention utilizes the resonant characteristics of reconfigurable electromagnetic metasurfaces to actively adjust the resonant frequency within the working frequency band, so that the metasurface can achieve the best resonance effect with the target user's body tissue (such as skin, fat, muscle, etc.), so that it can generate resonance enhancement for the small fluctuations in dielectric constant caused by changes in blood glucose, and can actively amplify the influence of weak dielectric constant disturbances on radio frequency reflection signals, significantly improving the signal sensitivity and signal-to-noise ratio of blood glucose change detection.

[0021] (3) Good dynamic monitoring accuracy: In practical applications, the monitoring results are easily affected by multi-source environmental factors such as temperature, humidity, motion deformation and blood flow pulsation and human physiological rhythms. The signal-to-noise ratio of a single radio frequency signal is extremely low, and the long-term dynamic monitoring accuracy is poor. Therefore, this invention simultaneously collects enhanced radio frequency signals and multi-modal physiological environment signals, and inputs them into a trained machine learning model for multi-source data fusion and feature extraction. Through machine learning, multiple electromagnetic background noise interferences are stripped away and eliminated, which significantly reduces the blood glucose prediction error in complex daily scenarios and achieves long-term dynamic monitoring with high robustness and clinical accuracy.

[0022] (4) Improve user experience: Some existing metasurface solutions require the material to be directly printed or attached to the skin surface, which has problems such as chemical consumable consumption and discomfort after long-term wear, and cannot achieve true daily continuous monitoring. This invention seamlessly integrates the reconfigurable electromagnetic metasurface into the shell of wearable devices, such as smart bracelets, and places it between the antenna and the device contact surface. It does not directly contact the skin, so users do not need to apply or transfer any chemical consumables to the skin surface, realizing truly traceless, consumable-free, long-term comfortable continuous non-invasive blood glucose monitoring, which greatly improves user compliance. Attached Figure Description

[0023] Figure 1 This is a flowchart of the non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion in the embodiment; Figure 2This is a schematic diagram of a non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion in the embodiment. Figure 3 This is a schematic diagram of the reconfigurable electromagnetic metasurface in the embodiment; Figure 4 This is a schematic diagram showing the response curves of the radio frequency reflected signal at different blood glucose concentrations with and without the reconfigurable electromagnetic metasurface of this embodiment. Figure 5 This is a schematic diagram of the multimodal signal fusion process in the embodiment; Figure 6 This is a schematic diagram of the electronic device in the embodiment. Among them, 1. Radio frequency antenna, 2. Reconfigurable electromagnetic metasurface, 201. Tunable radio frequency active component, 3. Multimodal sensor array, 4. Signal processing unit, and 5. Wristband. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0025] Example 1 To address the problems existing in the prior art, this embodiment provides a non-invasive wearable blood glucose monitoring system based on a reconfigurable intelligent metasurface and multimodal sensing fusion. The system aims to enhance the influence of dielectric constant fluctuations caused by changes in blood glucose concentration on the radio frequency signal by introducing a reconfigurable electromagnetic metasurface 2 between the radio frequency antenna and human tissue. Combined with multimodal physiological signal-assisted correction, it achieves high-sensitivity and high-robustness non-invasive blood glucose monitoring.

[0026] See Figure 2 The system is integrated into a wearable device, which in this embodiment is a smart bracelet. The system includes: (1) Radio frequency antenna 1, including radio frequency transmitting antenna and receiving antenna.

[0027] (2) A reconfigurable electromagnetic metasurface 2 is disposed between the radio frequency antenna 1 and the contact surface of human tissue. It integrates an adjustable radio frequency active component 201 and is driven by an external control signal to dynamically reconfigure electromagnetic parameters.

[0028] The reconfigurable electromagnetic metasurface 2 is composed of arrayed microstructure units, each of which physically integrates at least one of varactor diodes and MEMS RF switches as an adjustable RF active component 201.

[0029] In this embodiment, instead of being directly printed onto the skin, a reconfigurable active electromagnetic metasurface 2 is embedded between the RF antenna 1 and the device contact surface. This metasurface is composed of arrayed microstructure units, see [link to documentation]. Figure 3 Each unit physically integrates adjustable RF active components 201 such as variable capacitors, varactor diodes, or MEMS RF switches. Unlike passive metasurfaces with fixed patterns, the active metasurface in this embodiment can reconstruct its equivalent electromagnetic parameters and resonant frequency in real time and dynamically within the operating frequency band, such as 2.4 GHz, 5.8 GHz, or 24-28 GHz, by inputting external voltage or control signals through a control circuit.

[0030] The reconfigurable electromagnetic metasurface 2 amplifies the dielectric constant disturbance caused by changes in blood glucose levels through its resonant characteristics, resulting in a more significant change in the reflection coefficient S11 of the radio frequency signal. The reconfigurability of the metasurface allows for the active enhancement of blood glucose signals at specific frequencies by adjusting its equivalent electromagnetic parameters through external control signals, such as voltage, temperature, or light control.

[0031] The reconfigurable electromagnetic metasurface 2 uses low-power control circuits, such as microcontrollers, to adjust its electromagnetic properties in real time to adapt to changes in individual skin impedance and environment.

[0032] (3) Multimodal sensor array 3, used at least for synchronous acquisition of temperature, humidity, motion posture and pulse wave signals.

[0033] This system not only relies on changes in the S11 parameter of the radio frequency reflection signal, but also incorporates a multimodal sensor array 3, specifically including an NTC temperature sensor, an SHT31 humidity sensor, an MPU6050 inertial measurement unit (IMU), and a MAX30102 photoplethysmography (PPG) sensor. The acquired enhanced radio frequency signal and multimodal physiological / environmental signals are synchronously input into a built-in machine learning model, such as a temporal convolutional neural network or an LSTM network. Through multi-source data fusion and feature extraction, electromagnetic background noise caused by temperature, sweating, motion artifacts, and blood flow pulsation is actively removed, thereby accurately reproducing the true blood glucose concentration.

[0034] (4) Signal processing unit 4 is used to receive the enhanced radio frequency reflected signal and the signal collected by the multimodal sensor array 3, and run the trained machine learning model to output blood glucose concentration value.

[0035] The signal processing unit 4 is a microcontroller or FPGA, which simultaneously outputs external control signals to adjust the resonant frequency of the reconfigurable electromagnetic metasurface 2 in real time.

[0036] Establish blood glucose concentration ( ) and changes in radio frequency reflected signals ( Mathematical models, such as: in, This represents the change in blood glucose concentration. Dielectric constant of tissue fluid / blood blood glucose concentration Intrinsic sensitivity; The dielectric constant of the RF antenna / metasurface The sensitivity to changing electromagnetic responses; These are the parasitic coupling coefficients caused by temperature, humidity, and macroscopic motion on the radio frequency signal, respectively, i.e., the interference weights; This relates to higher-order nonlinear errors and system thermal noise.

[0037] By training on a dataset including measured blood glucose values ​​and multimodal signals, a regression model is built to learn how to... Parameters, temperature, humidity, and IMU are used to infer changes in blood glucose levels, enabling real-time inversion of blood glucose concentration.

[0038] The physical inversion formula is incorporated as a loss function into the model training. Here, the regression model can use machine learning models with regularization constraints, such as Bayesian ridge regression and symbolic regression. Alternatively, the regression model can use deep learning models, such as 1D convolutional neural networks, autoencoders, and multilayer perceptrons.

[0039] In this embodiment, the input to the model is defined as follows: (1) Main mode data: Radio frequency (RF) characteristics. The data format is a continuous time-series vector or a two-dimensional feature map. Specifically, it contains the amplitude of the reflection coefficient under frequency sweep conditions. .

[0040] (2) Auxiliary mode A: Environmental baseline data.

[0041] The data is in the form of low-frequency scalar values ​​or very short one-dimensional vectors, specifically including skin surface temperature collected by the NTC sensor and local humidity collected by the SHT31.

[0042] (3) Auxiliary mode B: Macroscopic motion data. The data format is a high-frequency time-series vector, specifically including the three-axis acceleration and three-axis angular velocity collected by the MPU6050.

[0043] (4) Auxiliary mode C: Microscopic physiological interference data. The data form is a high-frequency time-series one-dimensional waveform, specifically including the photoplethysmography pulse wave acquired by MAX30102, which reflects the periodic changes in heart rate and subcutaneous blood flow volume. This micro-motion will severely modulate the radio frequency signal.

[0044] The model outputs a single continuous value, representing the estimated change in blood glucose concentration at the current moment. The data is presented as a scalar of type Float, with units of millimoles per liter (mmol / L) or milligrams per deciliter (mg / dL).

[0045] Specifically, in this embodiment, a radio frequency antenna 1 with an operating frequency of 2.45 GHz is designed and integrated into the smart bracelet. A reconfigurable electromagnetic metasurface 2 is placed between the antenna and the skin. Its unit is a metal S-shaped structure, integrating a varactor diode, such as AVX 0805-200pF, to adjust the resonant frequency in the 3-4 GHz band via a voltage signal. A temperature sensor (NTC), a humidity sensor (SHT31), an IMU sensor (MPU6050), and a PPG sensor (MAX30102) are also integrated.

[0046] Multimodal data from 100 subjects were collected to establish a blood glucose-radio frequency signal model, which was trained using an LSTM network to achieve a blood glucose prediction error of <15 mg / dL (MARD).

[0047] See Figure 4 This is a schematic diagram showing the response curves of the radio frequency reflection signal at different blood glucose concentrations with and without the reconfigurable electromagnetic metasurface 2 of this embodiment. Figure 4 The study demonstrates the importance of metasurfaces for blood glucose concentration sensing. The left figure shows that without a metasurface, changes in blood glucose concentration cannot be reflected in the radio frequency reflectance spectrum; the right figure shows that with a calibrated metasurface, blood glucose concentration can be clearly reflected in the radio frequency reflectance spectrum.

[0048] Example 2 Based on Example 1, this example provides a non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion. See [link to example]. Figure 1 It includes the following steps: Step S1: Transmit radio frequency signals to human tissue through radio frequency antenna 1.

[0049] Step S2: A reconfigurable electromagnetic metasurface 2 is disposed between the radio frequency antenna 1 and the human tissue. The reconfigurable electromagnetic metasurface 2 integrates a tunable radio frequency active component 201.

[0050] Step S3: The electromagnetic parameters of the reconfigurable electromagnetic metasurface 2 are dynamically adjusted by an external control signal to actively enhance the changes in radio frequency reflection signals caused by the disturbance of the dielectric constant of tissue fluid due to changes in blood glucose concentration.

[0051] Step S4: Acquire the enhanced radio frequency reflection signal and simultaneously acquire multimodal physiological and environmental signals.

[0052] Step S5: The enhanced radio frequency reflection signal and multimodal physiological and environmental signals are input into the trained machine learning model. By fusing multi-source data to eliminate environmental and human interference, the blood glucose concentration is obtained.

[0053] See Figure 5 The process of multimodal signal fusion involves extracting features from RF reflection signals, temperature signals, humidity signals, IMU motion signals, and PPG signals, fusing the data based on machine learning models such as LSTM, and achieving blood glucose concentration inversion through interference suppression.

[0054] This invention has the following characteristics: (1) Active environmental adaptation and individual calibration. Unlike traditional fixed or passive printed metasurfaces, this invention uses a reconfigurable active metasurface. The system can dynamically fine-tune the resonant frequency point through control signals such as voltage according to the skin impedance characteristics and wearing status of different users, so as to achieve personalized optimal electromagnetic coupling and radio frequency signal enhancement.

[0055] (2) Multimodal anti-interference capability. Abandoning the traditional physical modeling method that relies solely on a single frequency offset, this invention innovatively integrates PPG pulse wave, IMU motion posture and temperature and humidity environmental data, and uses deep learning networks to eliminate multiple human and environmental interferences, so that the prediction error MARD is significantly reduced to less than 15 mg / dL, greatly improving the monitoring robustness in complex daily scenarios.

[0056] (3) High integration and reusability. This invention seamlessly integrates the intelligent metasurface into the wearable device, eliminating the need for users to apply, print, or transfer any chemical consumables to the skin surface, thus achieving truly long-term, traceless, and consumable-free daily continuous monitoring.

[0057] Example 3 Based on the foregoing embodiments, this embodiment provides a computer-readable storage medium including one or more programs executable by one or more processors of an electronic device, the one or more programs including instructions for executing the aforementioned non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion.

[0058] Specifically, Figure 6 This is a schematic diagram of an electronic device in the embodiment. At the hardware level, the electronic device includes a processor, an internal bus, a network interface, memory, and non-volatile memory, and may also include other hardware required for the services. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to achieve the above-mentioned functions. Figure 1The method described herein. Of course, in addition to software implementation, this invention does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0059] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0060] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0061] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A non-invasive wearable blood glucose monitoring method based on reconfigurable intelligent metasurfaces and multimodal sensing fusion, characterized in that, Includes the following steps: Radio frequency signals are transmitted to human tissue via a radio frequency antenna (1); A reconfigurable electromagnetic metasurface (2) is disposed between the radio frequency antenna (1) and human tissue, and the reconfigurable electromagnetic metasurface (2) integrates an adjustable radio frequency active component (201). The electromagnetic parameters of the reconfigurable electromagnetic metasurface (2) are dynamically adjusted by external control signals to actively enhance the changes in radio frequency reflection signals caused by the disturbance of the dielectric constant of tissue fluid due to changes in blood glucose concentration. The enhanced radio frequency reflection signal was acquired, and multimodal physiological and environmental signals were acquired simultaneously. The enhanced radio frequency reflection signal and the multimodal physiological and environmental signals are input into a trained machine learning model, and the blood glucose concentration is obtained by fusing multi-source data.

2. The non-invasive wearable blood glucose monitoring method based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 1, characterized in that, The multimodal physiological and environmental signals include at least temperature signals, humidity signals, motion posture signals, and photoplethysmography (PPG) signals.

3. The non-invasive wearable blood glucose monitoring method based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 1, characterized in that, The external control signal is a voltage control signal. By adjusting the bias voltage applied to the reconfigurable electromagnetic metasurface (2), its equivalent electromagnetic parameters and resonant frequency are changed in real time to adapt to the skin impedance characteristics of different individuals and achieve impedance matching optimization.

4. The non-invasive wearable blood glucose monitoring method based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 1, characterized in that, The machine learning model is trained under supervision using the following inversion relationship as the loss: in, This represents the change in blood glucose concentration obtained through inversion. Dielectric constant of tissue fluid / blood blood glucose concentration intrinsic sensitivity, For changes in radio frequency reflected signals, The dielectric constant of the RF antenna / metasurface The varying electromagnetic response sensitivity , , These represent the parasitic coupling coefficients of radio frequency signals caused by temperature, humidity, and macroscopic motion, respectively. For higher-order nonlinear errors and system thermal noise, , , These are, respectively, changes in temperature, humidity, and macroscopic motion.

5. The non-invasive wearable blood glucose monitoring method based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 1, characterized in that, The reconfigurable electromagnetic metasurface (2) operates in the ISM band of 2.4 GHz, 5.8 GHz or 24-28 GHz, and the radio frequency reflected signal is the phase and / or amplitude change of the scattering parameter S11; the machine learning model is a long short-term memory network or a temporal convolutional neural network.

6. A non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion, characterized in that, For implementing the non-invasive wearable blood glucose monitoring method as described in any one of claims 1-5, the system comprises: The radio frequency antenna (1) includes a radio frequency transmitting antenna and a receiving antenna; A reconfigurable electromagnetic metasurface (2) is disposed between the contact surface between the radio frequency antenna and the human tissue. It integrates an adjustable radio frequency active component (201) and is driven by an external control signal to dynamically reconfigure electromagnetic parameters. A multimodal sensor array (3) is used to simultaneously acquire temperature, humidity, motion posture and pulse wave signals; The signal processing unit (4) is used to receive the enhanced radio frequency reflection signal and the signal collected by the multimodal sensor array (3), and run the trained machine learning model to output the blood glucose concentration value.

7. A non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 6, characterized in that, The reconfigurable electromagnetic metasurface (2) is composed of arrayed microstructure units, each of which physically integrates at least one of a varactor diode and a MEMS RF switch as the tunable RF active component (201).

8. A non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 6, characterized in that, The system is integrated into a wearable device, which is a smart bracelet.

9. A non-invasive wearable blood glucose monitoring system based on reconfigurable intelligent metasurfaces and multimodal sensing fusion according to claim 6, characterized in that, The signal processing unit (4) is a microcontroller or FPGA, which simultaneously outputs the external control signal to adjust the resonant frequency of the reconfigurable electromagnetic metasurface (2) in real time.

10. A computer-readable storage medium, characterized in that, Includes one or more programs that are executed by one or more processors of an electronic device, the one or more programs including instructions for executing the non-invasive wearable blood glucose monitoring method based on reconfigurable smart metasurfaces and multimodal sensing fusion as described in any one of claims 1-5.

Images