A non-invasive blood glucose monitoring system, method and wearable device
By combining a miniature terahertz chip with a predictive model, the problem of low accuracy in non-invasive blood glucose monitoring has been solved, achieving high-precision and convenient blood glucose monitoring, which is suitable for wearable devices.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA RUILONG TECH CO LTD
- Filing Date
- 2026-06-03
- Publication Date
- 2026-07-03
AI Technical Summary
Existing non-invasive blood glucose monitoring methods have low accuracy, while invasive and continuous blood glucose monitoring systems are expensive and pose a risk of foreign body reaction.
Employing a miniature terahertz chip, including a terahertz signal source, a dual-mode resonator, and a terahertz detector, the system calculates blood glucose concentration values through differential feature design and a pre-trained prediction model, offsetting the effects of tissue moisture fluctuations and temperature, and is integrated into wearable devices.
It significantly improves the accuracy of blood glucose measurement, avoids the risks of pain and infection, enables continuous and dynamic blood glucose monitoring, and enhances user experience and real-time performance.
Smart Images
Figure CN122320536A_ABST
Abstract
Description
Technical Field
[0001] This specification relates to the field of blood glucose monitoring equipment technology, and in particular to a non-invasive blood glucose monitoring system, method and wearable device. Background Technology
[0002] Blood glucose monitoring is crucial for diabetes management. Current technologies typically employ invasive finger-prick blood sampling for monitoring, which is painful, inconvenient, and carries a risk of infection for patients. Continuous blood glucose monitoring systems provide dynamic data, but their probes still need to be inserted subcutaneously, resulting in high costs and the risk of foreign body reactions. Non-invasive blood glucose monitoring technologies, such as those based on near-infrared spectroscopy and electrical impedance tomography, have long been limited by significant individual skin variability, numerous interfering factors, and low signal-to-noise ratios, making it difficult to achieve clinically acceptable accuracy.
[0003] In view of this, the embodiments of this specification are intended to provide a non-invasive blood glucose monitoring system, method and wearable device. Summary of the Invention
[0004] In view of the problem that the blood glucose monitoring results obtained by the existing non-invasive blood glucose monitoring methods are of low accuracy, the purpose of the embodiments of this specification is to provide a non-invasive blood glucose monitoring system, method and wearable device.
[0005] To solve the above-mentioned technical problems, the specific technical solutions of the embodiments in this specification are as follows:
[0006] Firstly, embodiments of this specification provide a non-invasive blood glucose monitoring system, comprising:
[0007] A miniature terahertz chip includes a terahertz signal source, a dual-mode resonator, and a terahertz detector; the terahertz signal source is used to apply a terahertz signal to a target tissue, the dual-mode resonator is used to generate a dual-mode resonant signal based on the feedback signal from the target tissue, and the terahertz detector is used to extract the dual-mode resonant signal.
[0008] The processing unit is used to extract the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculate the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content.
[0009] The algorithm unit is used to calculate and output the blood glucose concentration value based on the differential feature quantity, temperature and pre-trained prediction model.
[0010] Specifically, the terahertz signal source is used to generate continuous wave or pulsed terahertz signals with frequencies ranging from 0.5 THz to 3.0 THz.
[0011] Preferably, the terahertz signal source is a voltage-controlled oscillator chain or a circuit based on fundamental frequency multiplication.
[0012] Furthermore, the dual-mode resonator is two electromagnetically coupled open-loop resonators, a dual-band dipole antenna, or a photonic crystal defect cavity.
[0013] Furthermore, the two electromagnetically coupled open-loop resonators include a first resonator and a second resonator;
[0014] The geometry of the first resonator is such that its fundamental frequency resonant mode is located within the characteristic absorption band of glucose.
[0015] The geometry of the second resonator is such that its fundamental frequency resonant mode is located in the frequency band where water exhibits characteristic absorption but glucose does not.
[0016] Preferably, the terahertz detector is a square-law detector based on a Schottky diode.
[0017] Specifically, the differential feature quantity is calculated using the following method:
[0018] D = Δf_G - k * Δf_W
[0019] Where D is the differential characteristic quantity, Δf_G is the offset of the first resonant mode frequency, Δf_W is the offset of the second resonant mode frequency, and k is the preset compensation coefficient.
[0020] Preferably, the system further includes:
[0021] A control unit is used to control the terahertz signal source to operate in pulse mode;
[0022] The signal conditioning circuit, including an amplifier, a filter, and an analog-to-digital converter, is used to preprocess the dual-mode resonant signal output by the terahertz detector so that the processing unit can calculate differential characteristic quantities based on the preprocessed dual-mode resonant signal.
[0023] Preferably, the prediction model is any one of random forest, XGboost model, support vector machine, long short-term memory network, gated recurrent unit or temporal convolutional network;
[0024] Based on the differential feature quantity, temperature, and pre-trained prediction model, the blood glucose concentration value is calculated and output, further including:
[0025] The differential feature value and temperature corresponding to the current moment, as well as the differential feature value and temperature of at least one historical moment, are input into the prediction model so that the prediction model outputs the blood glucose prediction value for the future moment.
[0026] Furthermore, the micro terahertz chip is fabricated using CMOS, SiGe, or MEMS processes.
[0027] Secondly, embodiments of this specification provide a non-invasive blood glucose monitoring method, which is applied to a non-invasive blood glucose monitoring system as described above, and the method includes:
[0028] A terahertz signal source applies a terahertz signal to the target tissue;
[0029] The dual-mode resonator generates a dual-mode resonant signal based on a feedback signal, wherein the feedback signal is the response signal of the target tissue to the terahertz information;
[0030] The terahertz detector extracts the dual-mode resonant signal;
[0031] The processing unit extracts the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculates the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content.
[0032] The algorithm unit calculates and outputs the blood glucose concentration value based on the differential feature quantity, temperature, and pre-trained prediction model.
[0033] Thirdly, embodiments of this specification provide a wearable device, including a housing and a non-invasive blood glucose monitoring system provided by the above-described technical solution;
[0034] The non-invasive blood glucose monitoring system is housed within the casing.
[0035] The micro terahertz chip has multiple microbumps on one side, which are used for electrical connection with the electrical motherboard inside the housing.
[0036] Preferably, the surface of the micro terahertz chip is further provided with a passivation layer, which is made of silicon nitride.
[0037] By employing the above technical solutions, the non-invasive blood glucose monitoring system, method, and wearable device provided in this specification, through differential feature quantity design, can effectively offset the interference of tissue moisture fluctuations on glucose measurement; simultaneously, through a pre-trained prediction model, it can eliminate the influence of factors such as temperature and individual differences, significantly improving the accuracy of blood glucose measurement; and this system can avoid the pain, inconvenience, and infection risks caused by repeated blood sampling, enhancing the user experience. Furthermore, based on mature semiconductor manufacturing processes, this system features miniaturization and is easily integrated into wearable devices such as watches, rings, and patches, enabling continuous and dynamic blood glucose monitoring. Users can monitor anytime, anywhere, facilitating daily management and blood glucose trend analysis, and offering extremely high real-time performance and convenience.
[0038] To make the above and other objects, features and advantages of the embodiments of this specification more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0039] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0040] Figure 1 This specification shows a schematic diagram of the structure of a non-invasive blood glucose monitoring system provided in an embodiment.
[0041] Figure 2 This specification illustrates a step diagram of a non-invasive blood glucose monitoring method provided in an embodiment.
[0042] Figure 3 A schematic diagram of the structure of a wearable device provided in an embodiment of this specification is shown.
[0043] Explanation of symbols in the attached drawings:
[0044] 100. Non-invasive blood glucose monitoring system;
[0045] 10. Miniature terahertz chip;
[0046] 11. Terahertz signal source; 12. Dual-mode resonator; 13. Terahertz detector;
[0047] 20. Processing unit;
[0048] 30. Algorithm Unit;
[0049] 200. Shell. Detailed Implementation
[0050] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.
[0051] It should be noted that the terms "first," "second," etc., used in this specification, claims, and the foregoing drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0052] This specification provides a non-invasive blood glucose monitoring system, method, and wearable device to address the problem of low accuracy in blood glucose monitoring results obtained by existing non-invasive blood glucose monitoring methods.
[0053] Specifically, such as Figure 1 As shown in the embodiments of this specification, a non-invasive blood glucose monitoring system 100 includes:
[0054] The miniature terahertz chip 10 includes a terahertz signal source 11, a dual-mode resonator 12, and a terahertz detector 13; the terahertz signal source 11 is used to apply a terahertz signal to the target tissue, the dual-mode resonator 12 is used to generate a dual-mode resonant signal based on the feedback signal from the target tissue, and the terahertz detector 13 is used to extract the dual-mode resonant signal.
[0055] Processing unit 20 is used to extract the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculate the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content.
[0056] Algorithm unit 30 is used to calculate and output blood glucose concentration values based on the differential feature quantity, temperature and pre-trained prediction model.
[0057] The non-invasive blood glucose monitoring system provided in this specification, through differential feature design, can effectively offset the interference of tissue moisture fluctuations (sweat, edema, etc.) on glucose measurement. Simultaneously, through a pre-trained prediction model, it can eliminate the influence of factors such as temperature and individual differences, significantly improving the accuracy of blood glucose measurement. Furthermore, it avoids the pain, inconvenience, and infection risks associated with repeated blood sampling, enhancing the user experience. In addition, this non-invasive blood glucose monitoring system is based on a miniature terahertz chip manufactured using mature semiconductor technology. Its miniaturized size makes it easy to integrate into wearable devices such as watches, rings, and patches, demonstrating strong wearability potential. This enables continuous and dynamic blood glucose monitoring; users can monitor anytime, anywhere, facilitating daily management and blood glucose trend analysis, offering extremely high real-time performance and convenience.
[0058] Specifically, the terahertz signal source is used to generate continuous wave or pulsed terahertz signals with frequencies ranging from 0.5 THz to 3.0 THz.
[0059] The embodiments of this specification provide a non-invasive blood glucose monitoring system that, by limiting the frequency range and operating mode of the terahertz signal generated by the terahertz signal source, ensures that the signal can effectively penetrate the skin surface and interact with glucose molecules, thereby achieving blood glucose monitoring.
[0060] The terahertz signal source 11 is a voltage-controlled oscillator chain or a circuit based on fundamental frequency multiplication.
[0061] A voltage-controlled oscillator (VCO) chain can directly generate high-frequency oscillations through multi-stage amplifiers and resonant structures (such as inductor-capacitor LC resonators or transmission line resonators). In the embodiments described in this specification, it may specifically include a core VCO operating at a lower frequency band (e.g., tens of GHz), whose output is locked in frequency and phase by one or more phase-locked loops; then, the signal is gradually boosted to the target terahertz frequency band (0.5-3.0 THz) through an amplifier chain and / or frequency multiplier (e.g., a frequency multiplier); finally, a terahertz power amplifier amplifies the signal to a power level sufficient to penetrate the skin's surface.
[0062] This approach enables continuous wave output, improving the signal-to-noise ratio. It is suitable for continuous wave or narrowband pulse excitation methods, generating pure, low-phase-noise single-frequency signals, which is beneficial for accurate measurement of dual-mode resonators and high-frequency offset detection accuracy. Furthermore, the output frequency can be continuously or incrementally changed within a certain range by adjusting the control voltage, offering high flexibility. In addition, it is easily integrated with CMOS and other processes to design voltage-controlled oscillators, which can then be multiplied by a frequency multiplier chain to reach the terahertz band, meeting the requirements for miniaturization and low cost.
[0063] Circuits based on fundamental frequency multiplication start from a stable, low-phase-noise low-frequency fundamental frequency source (such as a 10GHz or 20GHz crystal oscillator or a silicon-based voltage-controlled oscillator), generate rich harmonics through a nonlinear device (such as a pair of anti-parallel Schottky diodes, a varactor diode, or the nonlinear region of a transistor), and then use a bandpass filter or resonator to filter out the desired Nth harmonic (multiplication) components, thereby obtaining a terahertz signal. For example, a 100 GHz fundamental frequency signal can be tripled to produce a 300 GHz (0.3 THz) output.
[0064] In this way, the baseband source can operate in a lower frequency band with better performance and greater stability, and the baseband source itself can provide higher power, achieving higher terahertz output power than voltage-controlled oscillator chains in certain frequency bands.
[0065] In the embodiments described in this specification, the dual-mode resonator 12 is two electromagnetically coupled open-loop resonators, a dual-band dipole antenna, or a photonic crystal defect cavity.
[0066] Through electromagnetically coupled open-loop resonators, dual-band dipole antennas, or photonic crystal defect cavities, the dual-mode resonator in the embodiments of this specification can simultaneously and independently excite two resonant modes in a compact space to meet the needs of blood glucose monitoring.
[0067] In one specific embodiment, the two electromagnetically coupled open-loop resonators include a first resonator and a second resonator;
[0068] The geometry of the first resonator is such that its fundamental frequency resonant mode is located within the characteristic absorption band of glucose.
[0069] The geometry of the second resonator is such that its fundamental frequency resonant mode is located in the frequency band where water exhibits characteristic absorption but glucose does not.
[0070] The open-loop resonator generates high-quality resonance, which is evident in the transmission or reflection spectrum, thus improving the sensitivity of frequency detection. Even a small change in its dielectric constant can cause a significant frequency shift, forming the basis for the high-precision measurement of the non-invasive blood glucose monitoring system provided in this specification. Furthermore, its structure is easy to manufacture and integrate, and it can be easily integrated with terahertz signal sources and terahertz detectors onto the same chip.
[0071] The non-invasive blood glucose monitoring system provided in this specification, by setting the first resonator in the open-loop resonator to be sensitive to glucose (e.g., the resonant mode is located at 1.4THz±0.1THz, in which water is also relatively sensitive) and the second resonator to be sensitive to the characteristic absorption of water (e.g., its resonant mode is located at 1.6THz±0.1THz, in which only water is sensitive), can effectively eliminate the interference of water fluctuations on blood glucose monitoring through real-time differential calculation.
[0072] Specifically, in the embodiments of this specification, the terahertz detector 13 is a square-law detector based on a Schottky diode. This facilitates chip miniaturization and low-cost manufacturing.
[0073] Specifically, the differential characteristic quantities described in the embodiments of this specification are calculated using the following method:
[0074] D = Δf_G - k * Δf_W;
[0075] Where D is the differential characteristic quantity, Δf_G is the offset of the first resonant mode frequency, Δf_W is the offset of the second resonant mode frequency, and k is the preset compensation coefficient.
[0076] By calculating the differential feature quantity, the influence of water on blood glucose monitoring can be deducted, and the compensation ratio for water in the differential feature quantity can be dynamically adjusted by the compensation coefficient k, making the algorithm more adjustable and adaptable.
[0077] Preferably, in the embodiments of this specification, the system further includes:
[0078] A control unit is used to control the terahertz signal source to operate in pulse mode;
[0079] The signal conditioning circuit, including an amplifier, a filter, and an analog-to-digital converter, is used to preprocess the dual-mode resonant signal output by the terahertz detector so that the processing unit can calculate differential characteristic quantities based on the preprocessed dual-mode resonant signal.
[0080] By adding a control unit and signal conditioning circuit, the non-invasive blood glucose monitoring system provided in the embodiments of this specification can be made more controllable and have higher signal quality, providing more accurate data for subsequent blood glucose prediction.
[0081] In some specific embodiments, the prediction model is one of a traditional machine learning model or a neural network model. The traditional machine learning module may include at least random forest, XGboost model, support vector machine, etc., and the neural network model may include at least long short-term memory network, gated recurrent unit or temporal convolutional network.
[0082] In the embodiments of this specification, the algorithm module calculates and outputs the blood glucose concentration value based on the differential feature quantity, temperature, and pre-trained prediction model as follows:
[0083] The differential feature quantity and temperature corresponding to the current moment, as well as the differential feature quantity and temperature of at least one historical moment, are input into the prediction model so that the prediction model outputs the user's blood glucose prediction value for the future moment.
[0084] The current moment and historical moments are continuous in time, so the prediction model can capture the dynamic process and trend of the user's physiological parameters evolving over time, and can avoid instantaneous noise and random disturbances, thus improving the accuracy of blood glucose prediction.
[0085] The prediction model is trained through the following steps:
[0086] Training samples are collected to form a training dataset. These training samples include the differential feature quantity and temperature at a first time point, the differential feature quantity and temperature at a second time point, and the blood glucose value at a third time point. The second time point is a moment that precedes the first time point, meaning it is a historical moment relative to the first time point; and the second and first times points can be temporally continuous. The third time point is a moment that follows the first time point, meaning it is a future moment relative to the first time point. There is at least one second time point; when there are multiple second times points, they are temporally continuous.
[0087] An initial prediction model is constructed. Using the differential feature values and temperatures at the first and second time points from the training samples as input data, the initial prediction model is iteratively optimized until the loss function between the blood glucose prediction value output by the initial model and the blood glucose value at the third time point converges to a preset threshold. This yields a fully trained prediction model. Finally, the trained prediction model is deployed and integrated into the memory of the algorithm unit.
[0088] During model training, the blood glucose level at a third time point is used as the monitoring target. This third time point blood glucose level can be obtained through invasive or minimally invasive blood glucose meters, which helps to train the prediction model in the embodiments of this specification to output accuracy synchronized with invasive and non-invasive blood glucose meters. The temperature collected at each time point can include ambient temperature and user body temperature.
[0089] Before inputting the corresponding data from the first and second time points into the initial prediction model, preprocessing such as data cleaning and outlier removal can be performed on the input data.
[0090] The XGboost model efficiently captures the complex nonlinear relationship between blood glucose levels and factors such as moisture, temperature, and other characteristics that may affect blood glucose prediction. Through extensive training with clinical data, it automatically finds the optimal conversion rule, achieving high-precision prediction results. Furthermore, the XGboost model is computationally efficient and has a small processing scale, making it easy to deploy on resource-constrained devices such as wearable devices, meeting product practicality requirements.
[0091] Temporal convolutional networks can learn dynamic patterns and physiological inertia of blood glucose changes, enabling them to predict and warn about blood glucose levels. For example, they can predict earlier the trend of blood glucose entering a high-risk zone (such as hypoglycemia) and issue an early warning, providing users with a valuable window for blood glucose intervention. Furthermore, by combining historical blood glucose data, they can accurately distinguish between real physiological changes and random noise, making the predicted blood glucose values closer to the actual values and significantly reducing prediction errors during periods of rapid blood glucose changes (such as after meals). Moreover, with increased usage time, temporal convolutional networks can continuously adapt to individual user differences (such as unique glucose metabolism kinetics, such as digestion rate and insulin sensitivity), resulting in continuously improving blood glucose prediction results for individual users over time.
[0092] Preferably, in the embodiments of this specification, the micro terahertz chip is fabricated using CMOS, SiGe, or MEMS processes.
[0093] By employing mature CMOS, SiGe, MEMS and other processes to chipify the terahertz signal source, dual-mode resonator and terahertz detector, the system size is reduced to the cubic millimeter level and the average power consumption is less than 10 milliwatts, thus making it perfectly compatible with various consumer wearable devices.
[0094] Furthermore, this specification also provides a non-invasive blood glucose monitoring method, which can be applied to a non-invasive blood glucose monitoring system as described above. Figure 2 The diagram shown is a step-by-step illustration of a non-invasive blood glucose monitoring method provided in an embodiment of this specification. This specification provides the operational steps of the method described in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operational steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only possible execution order. In actual system or device products, the methods shown in the embodiments or drawings can be executed sequentially or in parallel. Specifically, as shown... Figure 2 As shown, the method may include the following steps:
[0095] S210: A terahertz signal source applies a terahertz signal to the target tissue;
[0096] S220: The dual-mode resonator generates a dual-mode resonant signal based on the feedback signal, wherein the feedback signal is the response signal of the target tissue to the terahertz information;
[0097] S230: The terahertz detector extracts the dual-mode resonant signal;
[0098] S240: The processing unit extracts the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculates the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content.
[0099] S250: The algorithm unit calculates and outputs the blood glucose concentration value based on the differential feature quantity, temperature and pre-trained prediction model.
[0100] The non-invasive blood glucose monitoring method provided in this specification effectively eliminates the main interference of moisture fluctuations through the core step of dual-mode resonant differential extraction, significantly improving the accuracy of glucose-specific detection. Furthermore, by inputting the differential feature quantity and temperature data into a pre-trained machine learning model, it can intelligently compensate for complex individual differences and physiological lag effects, thereby outputting clinically accurate blood glucose concentration results. Moreover, the entire method described in this specification can be implemented on a microchip, supporting rapid and continuous non-invasive blood glucose measurement, and further enabling blood glucose trend prediction and abnormal warning, providing a convenient tool for blood glucose monitoring and management.
[0101] Furthermore, such as Figure 3 As shown in the figure, this specification also provides a wearable device, including a housing 200 and a non-invasive blood glucose monitoring system 100 as provided in the above technical solution;
[0102] The non-invasive blood glucose monitoring system 100 is disposed within the housing 200;
[0103] The micro terahertz chip has multiple microbumps on one side, which are used for electrical connection with the electrical motherboard inside the housing.
[0104] By using bumps to achieve the shortest electrical connection path between the micro terahertz chip and the device's electrical motherboard, parasitic losses and interference in the transmission of high-frequency terahertz signals in long leads can be greatly reduced, ensuring the integrity of the core sensing signals. At the same time, it helps to improve the mechanical strength of the structure and reduce the device's footprint, providing key support for the overall compact and lightweight design of wearable devices.
[0105] In this embodiment of the specification, the wearable device is a smart ring. However, it should be noted that the wearable device can also be other devices besides a smart ring, including but not limited to smart bracelets and watches, head-mounted devices, ear-loop devices, patch / skin-attached devices, smart fabrics, smart necklaces / pendants / armbands, etc. The wearable device in this embodiment of the specification is preferably a device that can be closely attached to the target monitoring area (such as the arm, wrist, neck, etc.) and has a high wearing comfort.
[0106] Preferably, in the embodiments of this specification, the surface of the micro terahertz chip is further provided with a passivation layer; more preferably, the passivation layer is made of silicon nitride. Furthermore, to optimize coupling, a matching medium with a known dielectric constant (such as a silicon dioxide layer of a specific thickness) can be filled between the resonator and the skin of the target detection site.
[0107] The passivation layer provides robust protection for the delicate terahertz chip, effectively blocking the intrusion of moisture, sweat, and ionic contaminants from the external environment, preventing chip corrosion and electrical performance degradation. Silicon nitride is a biocompatible inert material that ensures safety and reliability even in direct skin contact. Furthermore, silicon nitride exhibits low absorption and low dispersion in the terahertz frequency band, preventing significant attenuation or interference with the terahertz signals emitted and received by the chip. This helps ensure the originality and accuracy of the sensing signal, and contributes to the long-term accuracy of blood glucose monitoring.
[0108] It should be noted that, in the embodiments of this specification, the use of the terms "comprising" or "including" to describe combinations of elements, components, parts, or steps herein also contemplates embodiments essentially composed of these elements, components, parts, or steps. The use of the term "may" herein is intended to indicate that any described attribute "may" include is optional. Multiple elements, components, parts, or steps can be provided by a single integrated element, component, part, or step. Alternatively, a single integrated element, component, part, or step can be divided into multiple separate elements, components, parts, or steps. The use of "a" or "an" to describe an element, component, part, or step does not imply exclusion of other elements, components, parts, or steps.
[0109] The various embodiments described in this specification are presented in a progressive manner, with each embodiment focusing on its differences from the others. Similar or identical parts between embodiments can be referred to interchangeably. The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made according to the spirit and essence of the present invention should be included within the scope of protection of the present invention.
Claims
1. A non-invasive blood glucose monitoring system, characterized in that, The system includes: A miniature terahertz chip includes a terahertz signal source, a dual-mode resonator, and a terahertz detector; the terahertz signal source is used to apply a terahertz signal to a target tissue, the dual-mode resonator is used to generate a dual-mode resonant signal based on the feedback signal from the target tissue, and the terahertz detector is used to extract the dual-mode resonant signal. The processing unit is used to extract the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculate the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content. The algorithm unit is used to calculate and output the blood glucose concentration value based on the differential feature quantity, temperature and pre-trained prediction model.
2. The system according to claim 1, characterized in that, The terahertz signal source is used to generate continuous wave or pulsed terahertz signals with frequencies ranging from 0.5 THz to 3.0 THz.
3. The system according to claim 2, characterized in that, The terahertz signal source is a voltage-controlled oscillator chain or a circuit based on fundamental frequency multiplication.
4. The system according to claim 1, characterized in that, The dual-mode resonator is two electromagnetically coupled open-loop resonators, a dual-band dipole antenna, or a photonic crystal defect cavity.
5. The system according to claim 4, characterized in that, The two electromagnetically coupled open-loop resonators include a first resonator and a second resonator. The geometry of the first resonator is such that its fundamental frequency resonant mode is located within the characteristic absorption band of glucose. The geometry of the second resonator is such that its fundamental frequency resonant mode is located in the frequency band where water exhibits characteristic absorption but glucose does not.
6. The system according to claim 1, characterized in that, The terahertz detector is a square-law detector based on a Schottky diode.
7. The system according to claim 1, characterized in that, The difference feature quantity is calculated using the following method: D = Δf_G - k * Δf_W Where D is the differential characteristic quantity, Δf_G is the offset of the first resonant mode frequency, Δf_W is the offset of the second resonant mode frequency, and k is the preset compensation coefficient.
8. The system according to claim 1, characterized in that, The system also includes: A control unit is used to control the terahertz signal source to operate in pulse mode; The signal conditioning circuit, including an amplifier, a filter, and an analog-to-digital converter, is used to preprocess the dual-mode resonant signal output by the terahertz detector so that the processing unit can calculate differential characteristic quantities based on the preprocessed dual-mode resonant signal.
9. The system according to claim 8, characterized in that, The prediction model can be any one of random forest, XGboost model, support vector machine, long short-term memory network, gated recurrent unit or temporal convolutional network; Based on the differential feature quantity, temperature, and pre-trained prediction model, the blood glucose concentration value is calculated and output, further including: The differential feature value and temperature corresponding to the current moment, as well as the differential feature value and temperature of at least one historical moment, are input into the prediction model so that the prediction model outputs the blood glucose prediction value for the future moment.
10. The system according to claim 1, characterized in that, The micro terahertz chip is fabricated using CMOS, SiGe, or MEMS processes.
11. A non-invasive blood glucose monitoring method, characterized in that, The method, applied to the non-invasive blood glucose monitoring system as described in any one of claims 1 to 10, comprises: A terahertz signal source applies a terahertz signal to the target tissue; The dual-mode resonator generates a dual-mode resonant signal based on a feedback signal, wherein the feedback signal is the response signal of the target tissue to the terahertz information; The terahertz detector extracts the dual-mode resonant signal; The processing unit extracts the first resonant mode frequency and the second resonant mode frequency from the dual-mode resonant signal and calculates the differential characteristic quantity, wherein the first resonant mode frequency is sensitive to changes in glucose concentration and the second resonant mode frequency is sensitive to changes in moisture content. The algorithm unit calculates and outputs the blood glucose concentration value based on the differential feature quantity, temperature, and pre-trained prediction model.
12. A wearable device, characterized in that, Includes a housing and a non-invasive blood glucose monitoring system as described in any one of claims 1 to 10; The non-invasive blood glucose monitoring system is housed within the casing. The micro terahertz chip has multiple microbumps on one side, which are used for electrical connection with the electrical motherboard inside the housing.
13. The wearable device according to claim 12, characterized in that, The surface of the micro terahertz chip is also provided with a passivation layer; The passivation layer is made of silicon nitride.