Insulin concentration noninvasive detection method and device

Abstract

The invention provides a non-invasive insulin concentration detection method and device, which is a method for cooperatively judging the insulin concentration in blood by combining a dielectric constant and a physiological state of subcutaneous tissue fluid while detecting a surface enhanced Raman scattering spectrum signal capable of reflecting insulin molecule specific vibration characteristics by using near-infrared light. And the problems of inaccurate measurement result and low sensitivity caused by weak and easily interfered Raman spectrum detection signals excited by infrared light are effectively solved. Moreover, by introducing a physiological state as a reference, interference caused by sweating, body temperature fluctuation or motion artifacts and the like can be dynamically eliminated, so that insulin concentration and blood sugar concentration detection values with high accuracy and high robustness can still be provided in a complex physiological environment.

Description

Technical Field

[0001] This disclosure relates to the field of insulin detection, and in particular to a non-invasive method and device for detecting insulin concentration. Background Technology

[0002] Existing non-invasive insulin detection methods often utilize the principle that insulin and other protein molecules have a strong resonance Raman response in the ultraviolet (UV) spectral range, using UV light to excite Raman spectra to obtain insulin concentration. However, UV light not only has extremely shallow penetration depth, making it difficult to reach the interstitial fluid in the subcutaneous tissue, but more seriously, long-term UV radiation poses potential phototoxicity and carcinogenic risks to human skin, making it unsuitable for continuous monitoring with wearable devices. Methods that use infrared light to excite Raman spectra suffer from weak Raman spectral signals and noisy backgrounds. Furthermore, these methods rely on sensors based on a single physical principle, making it difficult to specifically identify insulin molecules, resulting in low measurement accuracy and sensitivity. Moreover, they are prone to significant measurement errors under specific wearing conditions (such as improper wearing or strenuous exercise), exhibiting poor robustness. Summary of the Invention

[0003] In view of the shortcomings of the prior art described above, the purpose of this disclosure is to provide a non-invasive method and device for detecting insulin concentration to solve the aforementioned problems.

[0004] The first aspect of this disclosure provides a non-invasive method for detecting insulin concentration, comprising: acquiring the dielectric constant, physiological state, and surface-enhanced Raman scattering (SERS) spectral signal reflecting the specific vibrational characteristics of insulin molecules in a user's subcutaneous tissue fluid, as measured by a sensing module attached to the surface of the user's skin; wherein the SERS spectral signal is obtained under excitation by a near-infrared light source; and determining the insulin concentration in the user's subcutaneous tissue fluid in a coordinated manner based on the dielectric constant, physiological state, and SERS spectral signal; wherein the physiological state includes at least one of resting, exercise, and stress; and the physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature.

[0005] In one embodiment of the first aspect of this disclosure, the method for collaboratively determining the insulin concentration in a user's subcutaneous tissue fluid based on the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid includes: inputting the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid as input features into a pre-trained multimodal fusion computing model, and inferring the insulin concentration in the user's subcutaneous tissue fluid through the multimodal fusion computing model.

[0006] In one embodiment of the first aspect of this disclosure, the method further includes: acquiring a near-infrared spectral signal measured by a sensing module attached to the surface of the user's skin; determining the blood glucose concentration in the user's subcutaneous tissue fluid based on the near-infrared spectral signal; wherein the method for determining the blood glucose concentration includes: the input features of the multimodal fusion computing model further include the near-infrared spectral signal, and the result obtained by the multimodal fusion computing model inference further includes the blood glucose concentration in the user's subcutaneous tissue fluid; the multimodal fusion computing model is a multi-task learning model; based on the ratio of insulin concentration to blood glucose concentration in the subcutaneous tissue fluid at each time point, fitting the changing trend of the ratio, and determining the user's insulin resistance level based on the changing trend.

[0007] In one embodiment of the first aspect of this disclosure, the multimodal fusion calculation model further includes: real-time compensation of the time difference between the concentrations of insulin and / or blood glucose in tissue fluid and blood based on the blood perfusion index, so as to obtain the concentrations of insulin and / or blood glucose in blood.

[0008] In one embodiment of the first aspect of this disclosure, the insulin concentration and / or blood glucose concentration in the blood are first calibrated; wherein the method of the first calibration includes: correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model based on reference insulin concentration values ​​and / or reference blood glucose concentration values ​​at several reference times input by the user.

[0009] In one embodiment of the first aspect of this disclosure, a second calibration is performed on the insulin concentration and / or blood glucose concentration in the blood; wherein the method of the second calibration includes: generating a corresponding correction factor based on the systematic characteristic deviation between the multimodal sensing signal at the reference time of the day and the pre-stored multimodal baseline signal, and systematically correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model within the day according to the correction factor.

[0010] In one embodiment of the first aspect of this disclosure, a pre-trained quality control model is used to evaluate the physical quality of each signal measured by each sensing module in real time to obtain a first confidence level, wherein the evaluation criteria for the physical quality include at least one of the signal-to-noise ratio and stability of the signal; the multimodal fusion computing model is used to evaluate the degree of matching between its inference result and the preset physiological logic in real time to obtain a second confidence level; when the first confidence level and the second confidence level simultaneously meet their respective preset thresholds, the inference result of the multimodal fusion computing model is output.

[0011] A second aspect of this disclosure provides a non-invasive insulin concentration detection device, comprising: a flexible housing, a flexible battery, a flexible substrate, and a flexible adhesive layer stacked from top to bottom; wherein the flexible adhesive layer is used to adhere to the surface of a user's skin, and a plurality of first windows are provided at corresponding positions thereon for sensing modules on the flexible substrate to detect the user's skin; the flexible substrate is stacked on top of the flexible adhesive layer, and a signal acquisition module and an insulin concentration generation module comprising a plurality of sensing modules are disposed thereon; wherein the signal acquisition module includes a first sensing module, a second sensing module, and a third sensing module; the first sensing module is used to acquire the dielectric constant of the user's subcutaneous tissue fluid measured by the sensing modules attached to the user's skin surface; the second sensing module is used to acquire the user's physiological state; wherein The physiological state includes at least one of resting, exercise, and stress; the physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature; a third sensing module is used to acquire the user's surface-enhanced Raman scattering (SERS) spectral signal; wherein the SERS spectral signal is obtained under near-infrared light source excitation; an insulin concentration generation module is used to collaboratively determine the insulin concentration in the user's subcutaneous tissue fluid based on the dielectric constant of the subcutaneous tissue fluid, the physiological state, and the SERS spectral signal; the flexible battery is placed above the flexible substrate to power the flexible substrate and shield the circuit signals on the flexible substrate from external environmental interference; the flexible housing is placed above the flexible battery to protect the device.

[0012] In one embodiment of the second aspect of this disclosure, the signal acquisition module further includes a fourth sensing module for acquiring near-infrared spectral signals measured by a sensing module attached to the surface of the user's skin; the third and fourth sensing modules are disposed at the center of the flexible substrate, and the first sensing module is distributed around the center of the flexible substrate.

[0013] In one embodiment of the second aspect of this disclosure, the second sensing module includes a skin conductivity detection unit; the skin conductivity detection unit includes two opposing electrodes.

[0014] In one embodiment of the second aspect of this disclosure, the third sensing module is provided with a near-infrared light source; when the skin conductivity detection unit on the second sensing module or the fifth sensing module included in the signal acquisition module detects that the flexible adhesive layer and the skin are not tightly adhered, the near-infrared light source in the third sensing module is turned off.

[0015] In one embodiment of the second aspect of this disclosure, a second window is provided in the third sensing module region of the flexible substrate. When the flexible substrate and the flexible adhesive are stacked, the second window overlaps with a first window so that the near-infrared light source passes through the second window and the first window to irradiate the skin. A Raman enhancement substrate is provided on the side of the second window facing the skin, and the Raman enhancement substrate is detachably disposed on the flexible substrate.

[0016] In one embodiment of the second aspect of this disclosure, the third sensing module includes a Raman signal acquisition unit, the Raman signal acquisition unit being covered with a flexible metal shielding foil.

[0017] A third aspect of this disclosure provides an electronic terminal comprising: a memory; and a processor coupled to the memory, the processor being configured to execute the noninvasive insulin concentration detection method described in any of the preceding claims based on instructions stored in the memory.

[0018] As stated above, this disclosure has the following beneficial effects:

[0019] This disclosure utilizes near-infrared light to detect surface-enhanced Raman scattering (SERS) spectral signals that reflect the specific vibrational characteristics of insulin molecules. It combines this with the dielectric constant of subcutaneous tissue fluid and physiological state to collaboratively determine blood insulin concentration. This effectively overcomes the problems of inaccurate measurement results and low sensitivity caused by the weak and easily interfered signals of infrared-excited Raman spectroscopy. Furthermore, by introducing physiological state as a reference, this disclosure can dynamically eliminate interference caused by sweating, body temperature fluctuations, or motion artifacts. Therefore, even under complex physiological environments, it can still provide highly accurate and robust insulin and blood glucose concentration detection values. Attached Figure Description

[0020] Figure 1 The diagram shown is a flowchart of a non-invasive insulin concentration detection method according to an embodiment of this disclosure. Figure 1 .

[0021] Figure 2 The diagram shown is a flowchart of a non-invasive insulin concentration detection method according to an embodiment of this disclosure. Figure 2 .

[0022] Figure 3 The diagram shown is a flowchart of a non-invasive insulin concentration detection method according to an embodiment of this disclosure. Figure 3 .

[0023] Figure 4 The diagram shown is a structural schematic of a non-invasive insulin concentration detection device according to an embodiment of this disclosure.

[0024] Figure 5The diagram shown is a structural schematic of an electronic terminal according to an embodiment of this disclosure. Detailed Implementation

[0025] It should be noted first that:

[0026] 1. The following specific embodiments are merely illustrative of the principles and effects of this disclosure and are not intended to limit this disclosure. Any person skilled in the art can make various modifications, additions, changes, or equivalent substitutions to the above embodiments without departing from the spirit and scope of this disclosure. All technical solutions equivalent to those defined in the claims of this disclosure, or changes that a person skilled in the art can conceive of after reading this disclosure without creative effort, should be covered within the protection scope of this disclosure. The protection scope of this disclosure should be determined by the scope of the claims.

[0027] 2. Where there is no conflict, the various embodiments and features in the embodiments of this disclosure can be combined with each other, and the technical solutions formed by the combination are all considered to be the content of this disclosure. For the sake of brevity, this specification will not exhaustively list all possible combinations, but these combinations are also within the protection scope of this disclosure.

[0028] In this disclosure, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0029] 3. The division of modules (or units) in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of this disclosure can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0030] Furthermore, those skilled in the art will recognize that the various illustrative logic blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0031] 4. In the embodiments of this disclosure, terms such as "zeroth" and "first" are used to distinguish identical or similar items with essentially the same function and effect. For example, "first XX" and "second XX" are merely used to distinguish different XXs and do not limit their order, quantity, or execution sequence. Furthermore, terms such as "zeroth" and "first" do not necessarily imply that they are different.

[0032] 5. In the embodiments disclosed herein, the terms "exemplary" or "for example" indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this disclosure should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0033] The following specific examples illustrate the implementation of this disclosure, and those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed herein:

[0034] like Figure 1 As shown, the first aspect of this disclosure provides a non-invasive method for detecting insulin concentration, comprising:

[0035] Step S1: Acquire the dielectric constant, physiological state, and surface-enhanced Raman scattering (SERS) spectral signal reflecting the specific vibrational characteristics of insulin molecules in the user's subcutaneous tissue fluid, measured by a sensing module attached to the user's skin surface; wherein the SERS spectral signal is obtained under near-infrared light source excitation; the physiological state includes at least one of resting, exercise, and stress; the physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature.

[0036] It should be understood that although existing methods for non-invasive detection of insulin using ultraviolet light can provide good Raman scattering spectral signals, ultraviolet light has a shallow penetration depth and is difficult to reach subcutaneous tissue fluid. Furthermore, long-term ultraviolet radiation poses phototoxicity and carcinogenic risks to human skin. Therefore, it is not suitable for long-term continuous monitoring of insulin concentration in wearable devices.

[0037] While near-infrared light can also excite insulin molecules, its long wavelength results in extremely weak Raman scattering. In addition, the concentration of insulin in subcutaneous tissue fluid is extremely low and easily interfered with by autofluorescence signals from skin tissue. This leads to a low signal-to-noise ratio in the directly detected signal, making it difficult to extract effective insulin characteristic spectral information from complex background noise. Consequently, it is difficult to output insulin concentration detection results with high accuracy and sensitivity, and the measurement results have poor robustness.

[0038] This disclosure creatively proposes a method for determining blood insulin concentration by simultaneously using near-infrared light to detect surface-enhanced Raman scattering (SERS) spectral signals that reflect the specific vibrational characteristics of insulin molecules, and combining this with the dielectric constant of subcutaneous tissue fluid and physiological state. This effectively overcomes the problems of inaccurate measurement results and low sensitivity caused by the weak and easily interfered signals of infrared-excited Raman spectroscopy. Furthermore, by introducing physiological state as a reference, this disclosure can dynamically eliminate interference caused by sweating, body temperature fluctuations, or motion artifacts, thus providing highly accurate and robust insulin and blood glucose concentration values ​​even under complex physiological environments.

[0039] The main components of human subcutaneous tissue fluid are water, blood glucose, insulin, and electrolytes. Because blood glucose and insulin molecules have specific polarities, changes in their concentration alter the overall polarizability of the tissue fluid, interfering with Raman scattering spectral signal measurements and causing signal distortion. Therefore, this disclosure decouples the change in dielectric constant from the insulin concentration results obtained from Raman scattering, thereby eliminating background interference caused by changes in the hydration state of the tissue fluid and decoupling the true signal component related only to insulin. This solves the problem of low quantitative accuracy in complex physiological environments with existing single Raman detection methods.

[0040] Furthermore, changes in physiological state can lead to errors in the signals collected by the sensing module, resulting in incorrect insulin or blood glucose outputs. For example, when a user is under stress or engaged in strenuous exercise, increased sweating occurs. Since sweat is rich in highly conductive electrolytes, it significantly alters the dielectric environment of the skin surface, causing a falsely high dielectric constant reading, which the algorithm may misinterpret as a significant increase in insulin or blood glucose concentration. Similarly, when a user is in a low-temperature or resting state, peripheral blood vessels constrict, causing errors in the measurement of surface-enhanced Raman scattering (SERS) signals (e.g., weakened characteristic peak intensity), leading the algorithm to misinterpret as a decrease in insulin or blood glucose concentration. Therefore, this disclosure eliminates these interferences by dynamically compensating for and correcting signal deviations caused by changes in physiological state, thereby ensuring that the final output insulin concentration value more accurately reflects changes in the user's biochemical indicators.

[0041] Furthermore, heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature can be specifically decoupled and compensated for physical and physiological disturbances in different dimensions. Specifically, the perfusion index and body surface temperature can be used to correct optical transmission errors and metabolic time differences (see below for details); skin conductivity can be used to correct dielectric environment and electrochemical noise (e.g., to remove the artificially high component caused by the high conductivity of sweat from the dielectric constant reading, preventing it from being misjudged as a change in polarization rate caused by hyperglycemia / hyperinsulin); heart rate and heart rate variability can be used to assess the level of systemic metabolic stress (e.g., to identify whether the user is in an energy-consuming exercise state or a resting state, thereby dynamically adjusting the insulin metabolic consumption rate model and assisting in correcting the final insulin concentration calculation results).

[0042] It should be understood that the types of physiological states mentioned include, but are not limited to: stress states such as tension, anxiety, and fright; exercise states such as running and walking; and resting states such as sitting or lying down. It should be understood that the aforementioned physiological states do not constitute a limitation on the scope of protection of this disclosure. Any type of physiological state required to detect blood glucose and insulin based on surface-enhanced Raman scattering spectral signals and the dielectric constant of subcutaneous tissue fluid falls within the scope of protection of the physiological states in this disclosure.

[0043] Step S2: Based on the dielectric constant of the subcutaneous tissue fluid, physiological state, and surface-enhanced Raman scattering spectral signal, the insulin concentration in the user's subcutaneous tissue fluid is determined collaboratively.

[0044] It should be understood that step S2 aims to achieve accurate and highly sensitive measurement of insulin concentration by decoupling the effects of changes in the dielectric constant of tissue fluid and changes in physiological state on the Raman scattering spectral signal measurement from the measured insulin concentration value, and extracting the true characteristic signal that is only related to the insulin molecule.

[0045] In one embodiment of the first aspect of this disclosure, the method for collaboratively determining the insulin concentration in a user's subcutaneous tissue fluid based on the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid includes: inputting the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid as input features into a pre-trained multimodal fusion computing model, and inferring the insulin concentration in the user's subcutaneous tissue fluid through the multimodal fusion computing model.

[0046] Due to the extreme complexity of the subcutaneous physiological environment in the human body, the relationship between surface-enhanced Raman scattering (SERS) spectral signal intensity and insulin concentration is not a simple linear one. Instead, it is influenced by the nonlinear coupling of dielectric constant and physiological state, essentially representing a high-dimensional mapping relationship involving multiple variables. Therefore, this embodiment employs a multimodal fusion computing model based on machine learning or deep learning architecture to achieve high-precision feature extraction and concentration inversion.

[0047] Specifically, since Raman spectroscopy data represents high-dimensional vector signals, while dielectric constant and physiological state represent low-dimensional scalar signals, the multimodal fusion computational model maps the aforementioned multi-source heterogeneous data to a unified hidden feature space through a feature embedding layer for splicing and fusion. During pre-training, the model automatically learns the interference weights of different physiological states and dielectric constants on the intensity of Raman spectral characteristic peaks, thereby establishing a functional relationship between Raman spectra, physiological states, dielectric constants, and insulin concentrations. Subsequently, in the inference phase, after inputting the newly measured dielectric constant of subcutaneous tissue fluid, physiological states, and surface-enhanced Raman scattering spectral signals into the pre-trained multimodal fusion computational model, the corresponding insulin concentration can be inferred.

[0048] Preferably, the multimodal fusion computing model is a computing model pre-trained in the cloud using TensorFlow or PyTorch. More preferably, the multimodal fusion computing model is optimized through quantization (preferably 8-bit integer quantization), pruning, etc., and then converted to TensorFlow Lite for Microcontrollers or an equivalent format for execution.

[0049] like Figure 2 As shown, in one embodiment of the first aspect of this disclosure, the non-invasive method for detecting insulin concentration further includes:

[0050] Step S31: Acquire the near-infrared spectral signal measured by the sensing module attached to the user's skin surface.

[0051] Step S32: Determine the blood glucose concentration in the user's subcutaneous tissue fluid based on the near-infrared spectral signal; wherein, the method for determining the blood glucose concentration includes: the input features of the multimodal fusion calculation model also include the near-infrared spectral signal, and the result obtained by the multimodal fusion calculation model inference also includes the blood glucose concentration in the user's subcutaneous tissue fluid; the multimodal fusion calculation model is a multi-task learning model; based on the ratio of insulin concentration to blood glucose concentration in the subcutaneous tissue fluid at each time point, the changing trend of the ratio is fitted, and the user's insulin resistance level is determined based on the changing trend.

[0052] Preferably, the architecture of the multi-task learning model includes a shared feature encoder and two independent task decoders. Specifically, the shared feature encoder simultaneously receives Raman spectroscopy, near-infrared spectroscopy, and dielectric constant data to extract shared features reflecting the optical and electrical properties of human subcutaneous tissue; subsequently, the two task decoders use these shared features to synchronously and in parallel infer insulin concentration and blood glucose concentration.

[0053] It should be understood that insulin resistance refers to a pathological state in which peripheral tissues (such as muscle and fat) of the body have reduced sensitivity to insulin. Insulin concentration values ​​alone are significantly affected by eating patterns and fluctuate considerably (for example, insulin spikes after meals). Relying solely on a single insulin concentration makes it difficult to distinguish between normal insulin elevations caused by factors such as eating and pathological insulin elevations due to decreased cellular sensitivity. This step introduces blood glucose concentration as a reference benchmark to calculate the ratio of insulin concentration to blood glucose concentration, providing a visual representation of insulin resistance. When this ratio is significantly elevated, it means that the body needs to secrete higher concentrations of insulin to maintain the same blood glucose level. Therefore, it can accurately reflect whether the body is in the insulin resistance stage, providing users with an early risk warning that is more clinically valuable than a single indicator.

[0054] It should be understood that when blood glucose concentration is measured and combined with model inference, the dielectric constant of subcutaneous tissue fluid and physiological state play the same role as when insulin concentration is calculated using surface-enhanced Raman scattering spectroscopy signals. They can compensate for errors in the obtained blood glucose concentration value and improve its accuracy.

[0055] like Figure 3 As shown, in one embodiment of the first aspect of this disclosure, the non-invasive method for detecting insulin concentration further includes:

[0056] Step S41: The multimodal fusion calculation model compensates for the time difference between the concentrations of insulin and / or blood glucose in tissue fluid and blood in real time based on the blood perfusion index, so as to obtain the concentrations of insulin and / or blood glucose in blood.

[0057] It should be understood that there is a certain time lag (phase difference) in the diffusion of insulin and glucose molecules from the capillary network to the subcutaneous tissue fluid. This lag time is not a constant but is highly nonlinearly correlated with the hemodynamic state of the local microcirculation. This phase difference is related to the blood perfusion index. Specifically, when the blood perfusion index is high, fresh blood flows rapidly through the capillaries, quickly replenishing high concentrations of insulin or glucose molecules, keeping the concentration difference between the inside and outside of the blood vessels at its maximum. According to Fick's law, the greater the concentration difference, the faster the diffusion, and therefore the shorter the delay (phase difference). Conversely, when the blood perfusion index is low, blood flow is slow, and molecule replenishment is not timely. According to Fick's law, the diffusion rate slows down, leading to a significantly longer delay. On the other hand, a high blood perfusion index is usually accompanied by higher capillary hydrostatic pressure. A higher pressure value also accelerates the entry of water and insulin / glucose molecules into the subcutaneous tissue fluid, thus shortening the phase difference to some extent. Furthermore, when the blood perfusion index is high, it can cause previously closed and dormant capillaries to open, which increases the contact surface area between blood and tissue fluid, thereby further shortening the time difference for insulin or blood glucose molecules to diffuse from the blood into the subcutaneous tissue fluid.

[0058] The multimodal fusion computing model constructs a relationship between the blood perfusion index and the time phase difference between the blood perfusion index and the insulin and / or blood glucose concentrations in the blood and subcutaneous tissue fluid, thereby compensating for the time phase difference. Combined with the insulin and / or blood glucose concentrations in the subcutaneous tissue fluid, it obtains the insulin and / or blood glucose concentrations in the blood in real time.

[0059] In one embodiment of the first aspect of this disclosure, the insulin concentration and / or blood glucose concentration in the blood are first calibrated; wherein the method of the first calibration includes: correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model based on reference insulin concentration values ​​and / or reference blood glucose concentration values ​​at several reference times input by the user.

[0060] It should be understood that the reference time refers to the key metabolic time points corresponding to insulin or blood glucose concentrations, such as fasting, 1 hour postprandial, and 2-3 hours postprandial. Preferably, 3-5 key metabolic time points can be selected for the detection of reference insulin concentration values ​​and / or reference blood glucose concentration values, and the detection results can be input into the multimodal fusion calculation model to correct the model output results. The reference insulin concentration value and / or reference blood glucose concentration value refers to the accurate insulin concentration and / or blood glucose concentration in the blood directly measured by the user through invasive or other non-invasive detection methods.

[0061] Because the pre-trained multimodal fusion computing model outputs the average patterns of a large sample population, and for each individual user, factors such as skin color, body weight, and skin type can cause deviations in their physiological state, dielectric constant, and the fundamental response characteristics of surface-enhanced Raman scattering (SERS) signals compared to the pre-trained model for that large sample population, the same input signal intensity corresponds to different output results for different users. Therefore, by introducing accurate insulin and / or blood glucose concentration values ​​from the blood, personalized parameter corrections are made to the mapping relationship between the model input and output, thereby constructing a more personalized and accurate multimodal fusion computing model specific to that user. It should be understood that the first calibration usually only needs to be performed when the user first uses the model. This is because skin color, subcutaneous fat thickness, etc., are relatively stable physiological characteristics that do not fluctuate drastically in the short term like sweat or body temperature. Once the initial calibration is completed, the model can maintain long-term adaptability to the user in subsequent uses.

[0062] In one embodiment of the first aspect of this disclosure, a second calibration is performed on the insulin concentration and / or blood glucose concentration in the blood; wherein the method of the second calibration includes: generating a corresponding correction factor based on the systematic characteristic deviation between the multimodal sensing signal at the reference time of the day and the pre-stored multimodal baseline signal, and systematically correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model within the day according to the correction factor.

[0063] It should be understood that the reference time is the time period during which the user is in a basal metabolic state. Preferably, the reference time is the time period corresponding to when the user is in a fasting state, such as after waking up in the morning, before eating, or before strenuous exercise (it should be understood that at this time, the concentration of insulin and blood glucose in the body is at a relatively constant basal level, and is least affected by external interference). The multimodal sensing signals are the dielectric constant of subcutaneous tissue fluid, physiological state parameters, and surface-enhanced Raman scattering spectral signals. The multimodal baseline signal is the average or standard value of historical reference time signals measured and stored at the reference time of the user during the first calibration (i.e., the first personalized calibration) or at reference times of the past several days. The systematic feature deviation is the difference between the relevant features of the multimodal baseline signal extracted by the model and the relevant features of the multimodal sensing signal of the day. The correction factor is the overall offset or gain scaling factor calculated based on the systematic feature deviation. This correction factor will be applied to the measurement results of all subsequent times of the day to eliminate systematic errors caused by daytime instrument drift.

[0064] During long-term daily use, minor deviations in the signals measured by the sensing module due to slight offsets or changes in the degree of contact with the user's skin, as well as fluctuations in the daily hydration of the skin's stratum corneum (the degree of skin dryness), can all affect the results obtained by the model inference. Without correction, these signal fluctuations can be misinterpreted by the highly sensitive multimodal model as changes in insulin or blood glucose concentrations, leading to persistent systematic errors in the daily measurement results. This embodiment compares the systematic characteristic deviations between the multimodal sensing signals at the daily baseline time and the pre-stored multimodal baseline signals to generate a corresponding correction factor. This correction factor is then applied to subsequent model calculations for the day, thereby mitigating the output deviations of the multimodal fusion calculation model caused by the aforementioned factors and making its output results more accurate. Furthermore, since this method can automatically identify and acquire multimodal sensing signals at the reference time of the day, automatically compare them with the systematic characteristic deviations between them and the pre-stored multimodal baseline signals, and generate correction factors to correct the model output results, it can achieve user-unobtrusive daily self-calibration. This reduces long-term operation drift, improves the accuracy of insulin concentration and / or blood glucose concentration output results, and enhances the user experience.

[0065] In one embodiment of the first aspect of this disclosure, a pre-trained quality control model is used to evaluate the physical quality of each signal measured by each sensing module in real time to obtain a first confidence level, wherein the evaluation criteria for the physical quality include at least one of the signal-to-noise ratio and stability of the signal; the multimodal fusion computing model is used to evaluate the degree of matching between its inference result and the preset physiological logic in real time to obtain a second confidence level; when the first confidence level and the second confidence level simultaneously meet their respective preset thresholds, the inference result of the multimodal fusion computing model is output.

[0066] It should be understood that when the model output is insulin concentration, the signals measured by each sensing module are the dielectric constant of subcutaneous tissue fluid, physiological state, and surface-enhanced Raman scattering (SERS) spectral signals; when the model output is blood glucose concentration, the signals measured by each sensing module are the dielectric constant of subcutaneous tissue fluid, physiological state, and near-infrared spectral signals; when the model output is insulin concentration, blood glucose concentration, and insulin resistance level, the signals measured by each sensing module are the dielectric constant of subcutaneous tissue fluid, physiological state, SERS spectral signals, and near-infrared spectral signals. The physiological logic refers to the reasonable boundary conditions constructed based on the laws of human metabolic kinetics to constrain the model's reasoning results. For example, based on the continuity of biochemical processes, the concentration change between adjacent sampling points should conform to a smooth, continuous curve; or, based on the survival range of vital signs, an effective concentration range is set; or, based on physiological facts, the maximum physiological rate limit of insulin or blood glucose concentration changes in human blood is set; and so on.

[0067] When a user is in motion, the mechanical displacement of their limbs can cause micrometer-level gaps or pressure fluctuations at the contact interface between the sensor module and the skin, resulting in severe oscillations in contact impedance. This introduces random noise, degrades the signal-to-noise ratio of each signal, and worsens the stability of the signals. Therefore, by monitoring the signal-to-noise ratio and stability of the signals in real time, it is possible to distinguish the sensor decoupling from the actual fluctuations in insulin and / or blood glucose concentration values ​​at the physical source, preventing misjudgments caused by loose wearing.

[0068] This embodiment employs a dual verification mechanism to simultaneously determine the signal-to-noise ratio, stability, and the degree of matching between the model's output and the preset physiological logic of each signal. This ensures that the final output is both physically credible and medically logically consistent, thereby maximizing the accuracy of the output.

[0069] like Figure 4 As shown, a second aspect of this disclosure provides a non-invasive insulin concentration detection device 100, comprising: a flexible housing 101, a flexible battery 102, a flexible substrate 103, and a flexible adhesive layer 104 stacked from top to bottom.

[0070] The flexible adhesive layer 104 is used to attach to the user's skin surface, and a number of first windows are provided at corresponding positions on it so that the sensing module on the flexible substrate 103 can detect the user's skin.

[0071] Preferably, the flexible adhesive layer 104 is made of medical-grade liquid silicone or thermoplastic elastomer. Both have excellent chemical stability and bioinertness, giving the flexible adhesive layer 104 good biocompatibility with the skin and avoiding rejection reactions during long-term wear.

[0072] Preferably, the flexible adhesive layer 104 has a gradient structure in which the thickness gradually decreases from the center to the edge. This design eliminates the step effect by thinning the edges, effectively preventing edge lifting caused by clothing friction and improving the mechanical stability of wearing. On the other hand, this structure provides better surface conformability, improving the user's wearing comfort and skin fit.

[0073] Preferably, a medical-grade hydrocolloid adhesive is coated on the skin-facing side of the flexible adhesive layer 104. This not only ensures safe contact with sensitive skin due to the low allergenicity of the hydrocolloid, but also achieves a stable connection between the flexible adhesive layer 104 and the skin through its strong adhesion properties. Furthermore, it also has moisture-wicking properties, further maintaining contact stability during long-term wear.

[0074] The flexible substrate 103 is stacked on top of the flexible adhesive layer 104, and is provided with a signal acquisition module and an insulin concentration generation module containing several sensing modules. The signal acquisition module includes a first sensing module, a second sensing module, and a third sensing module. The first sensing module acquires the dielectric constant of the user's subcutaneous tissue fluid, measured by a sensing module attached to the user's skin surface. The second sensing module acquires the user's physiological state, which includes at least one of resting, exercise, and stress. The physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature. The third sensing module acquires the user's surface-enhanced Raman scattering (SERS) spectral signal, obtained under near-infrared light source excitation. The insulin concentration generation module collaboratively determines the insulin concentration in the user's subcutaneous tissue fluid based on the dielectric constant of the subcutaneous tissue fluid, the physiological state, and the SERS spectral signal.

[0075] In one embodiment of the second aspect of this disclosure, the signal acquisition module further includes a fourth sensing module for acquiring near-infrared spectral signals measured by a sensing module attached to the surface of the user's skin; the third and fourth sensing modules are disposed at the center of the flexible substrate 103, and the first sensing module is distributed around the center of the flexible substrate 103.

[0076] Preferably, the first sensing module comprises an array of complementary open-loop resonators of different sizes directly printed on the flexible substrate 103, the array being distributed around the center of the flexible substrate 103. Using the antenna formed by this array, electromagnetic field energy can be focused onto a region 2-3 mm below the skin rich in interstitial fluid, enabling the detection of the dielectric constant of the subcutaneous fluid.

[0077] In one embodiment of the second aspect of this disclosure, a second window is provided in the third sensing module region of the flexible substrate 103. When the flexible substrate 103 and the flexible adhesive layer 104 are stacked, the second window coincides with a first window (i.e., they are coaxially aligned in space), (thereby forming a connected optical channel) so that the near-infrared light source passes through the second window and the first window to irradiate the skin or the Raman enhancement substrate. A Raman enhancement substrate is provided on the side of the second window facing the skin, and the Raman enhancement substrate is detachably disposed on the flexible substrate 103.

[0078] Preferably, the first window and the second window are light-transmitting windows filled with a material that is transparent to near-infrared light (such as quartz, optical glass or transparent polymer film) (in this case, the Raman-enhancing substrate can be directly coated on the side of the second window facing the skin).

[0079] Preferably, the reinforcing substrate is a monolayer array of gold nanofibers or silver nanorods fixed by vapor deposition or self-assembly techniques. It should be understood that such a reinforcing substrate can exhibit good Raman enhancement for near-infrared light sources.

[0080] Preferably, the near-infrared light source includes a near-infrared laser emitter with a center wavelength of 785 nm and a narrow linewidth, positioned perpendicular to the flexible substrate 103. More preferably, the near-infrared light source further includes a micro-optical lens and a filter disposed on the laser emitter. It should be understood that the wavelength is set to 785 nm to maximize the suppression of autofluorescence interference from human skin tissue while ensuring the Raman signal intensity. The perpendicular positioning is also more conducive to achieving short-distance optical transmission to reduce light loss. The use of lenses and filters can focus the laser and filter out Rayleigh scattering light.

[0081] Preferably, the third sensing module further includes an avalanche photodiode or a silicon photomultiplier tube to capture weak Raman scattering signals.

[0082] In one embodiment of the second aspect of this disclosure, the third sensing module includes a Raman signal acquisition unit, the Raman signal acquisition unit being covered by a flexible metal shielding foil. Since the first sensing module emits a broadband microwave sweep signal to measure the dielectric constant during operation, this sweep signal can easily cause electromagnetic interference to the third sensing module. By providing the flexible metal shielding foil, the electrical noise interference of the microwave sweep signal emitted by the first sensing module to the Raman spectroscopy acquisition circuit is effectively blocked.

[0083] In one embodiment of the second aspect of this disclosure, the second sensing module includes a skin conductivity detection unit; the skin conductivity detection unit includes two opposing electrodes.

[0084] Preferably, the electrode is fabricated using an immersion gold (nickel immersion gold) process and comes into contact with the skin. Because the metal is a bioinert metal, it significantly reduces the risk of causing skin allergies or contact dermatitis compared to bare copper or tin-plated surfaces.

[0085] In one embodiment of the second aspect of this disclosure, the third sensing module is provided with a near-infrared light source; when the skin conductivity detection unit or the fifth sensing module included in the signal acquisition module of the second sensing module detects that the flexible adhesive layer 104 and the skin are not tightly adhered, the near-infrared light source in the third sensing module is turned off.

[0086] Preferably, the fifth sensing module is a capacitive contact sensor.

[0087] It should be understood that, as another optional or parallel detection method, the skin conductivity detection unit on the second sensing module can also be used to determine whether the skin is in close contact.

[0088] Because the 785nm near-infrared laser used in the third sensing module is invisible or weakly visible light, the human eye lacks a natural blink avoidance reflex for this wavelength. If the laser continues to operate in the event of accidental detachment of the patch, the high-energy-density laser beam may directly irradiate the eyes of the user or bystanders, focusing on the retina and causing irreversible thermal or photochemical damage. By introducing the aforementioned contact-state-based automatic shutdown mechanism, this safety risk can be avoided, improving the safety performance of device 100.

[0089] The flexible battery 102 is positioned above the flexible substrate 103 (i.e., on the side furthest from the user's skin) to power the flexible substrate 103 and shield it from external environmental interference with the circuit signals on the flexible substrate 103. It should be understood that the flexible battery 102 covers key circuit areas on the flexible substrate 103. On one hand, it acts as a power module, providing operating voltage to the flexible substrate 103 and various sensing modules; on the other hand, utilizing the conductivity of its internal metal current collector (such as copper foil or aluminum foil), it serves as an electromagnetic shielding layer, to a certain extent blocking high-frequency electromagnetic noise from the external environment from interfering with the weak analog signals on the flexible substrate 103, thereby improving the signal-to-noise ratio.

[0090] Preferably, the flexible battery 102 is a thin flexible lithium polymer battery. It should be understood that this type of battery has high energy density and excellent mechanical bending resistance, and can deform synchronously with the flexible substrate 103 and the user's limb without breaking the circuit or bulging, thus ensuring the power supply stability of the device 100 under dynamic wearing and the overall thin and light design.

[0091] The flexible housing 101 is positioned above the flexible battery 102 (i.e., the side of the flexible battery 102 furthest from the user's skin), providing protection for the device 100. Specifically, the flexible housing 101 covers the flexible battery 102 and its flexible circuit board. Simultaneously, the edge region of the flexible housing 101 is sealed and fixed to the edge region of the bottom flexible adhesive layer 104, thereby forming a sealed enclosure that encapsulates the flexible battery 102 and the flexible substrate 103 within it.

[0092] Preferably, the flexible housing 101 is an injection-molded soft silicone shell. It can provide sealed protection for the internal components of the device 100, achieving a waterproof and dustproof rating of IP67 or higher. Due to the integral molding characteristics and high density of silicone, it can provide comprehensive sealed protection for the internal components of the device 100, preventing external moisture or dust from intruding, thus enabling the device 100 to achieve an overall waterproof and dustproof rating of IP67 or higher.

[0093] A third aspect of this disclosure provides an electronic terminal 200, comprising: a memory; and a processor coupled to the memory, the processor being configured to execute the non-invasive insulin concentration detection method described in any of the preceding claims based on instructions stored in the memory. Figure 5 As shown, the various components in electronic terminal 200 are coupled together via bus system 204. It can be understood that bus system 204 is used to enable communication between these components. In addition to a data bus, bus system 204 also includes a power bus, a control bus, and a status signal bus.

[0094] The user interface 205 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0095] It is understood that memory 202 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this disclosure are intended to include, but are not limited to, these and any other suitable categories of memory.

[0096] The memory 202 in this embodiment is used to store various types of data to support the operation of the electronic terminal 200. Examples of this data include: any executable program for operation on the electronic terminal 200, such as the operating system 2021 and application programs 2022; the operating system 2021 contains various system programs, such as the framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. The application programs 2022 may contain various applications, such as media players, browsers, etc., for implementing various application services. The methods provided in this embodiment may be included in the application programs 2022.

[0097] The methods disclosed in the above embodiments of this disclosure can be applied to processor 201, or implemented by processor 201. Processor 201 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 201 or by instructions in the form of software. The processor 201 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 201 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. General-purpose processor 201 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of this disclosure can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0098] In an exemplary embodiment, the electronic terminal 200 may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to execute the aforementioned method.

[0099] In summary, this disclosure effectively overcomes the shortcomings of the prior art and has high industrial applicability.

Claims

1. A non-invasive method for detecting insulin concentration, characterized in that, include: The dielectric constant, physiological state, and surface-enhanced Raman scattering (SERS) spectral signal reflecting the specific vibrational characteristics of insulin molecules in the user's subcutaneous tissue fluid are obtained by a sensing module attached to the user's skin surface; wherein the SERS spectral signal is obtained under excitation by a near-infrared light source. The insulin concentration in the user's subcutaneous tissue fluid is determined collaboratively based on the dielectric constant of the subcutaneous tissue fluid, physiological state, and surface-enhanced Raman scattering spectral signal; wherein the physiological state includes at least one of resting, exercise, and stress; and the physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature.

2. The non-invasive method for detecting insulin concentration according to claim 1, characterized in that, The method for collaboratively determining the insulin concentration in a user's subcutaneous tissue fluid based on the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid includes: using the dielectric constant, physiological state, and surface-enhanced Raman scattering spectral signal of the subcutaneous tissue fluid as input features, inputting them into a pre-trained multimodal fusion computing model, and inferring the insulin concentration in the user's subcutaneous tissue fluid through the multimodal fusion computing model.

3. The non-invasive method for detecting insulin concentration according to claim 2, characterized in that, Also includes: Acquire near-infrared spectral signals measured by a sensing module attached to the user's skin surface; The blood glucose concentration in the user's subcutaneous tissue fluid is determined based on the near-infrared spectral signal; wherein, the method for determining the blood glucose concentration includes: the input features of the multimodal fusion computing model also include the near-infrared spectral signal, and the result obtained by the multimodal fusion computing model inference also includes the blood glucose concentration in the user's subcutaneous tissue fluid; the multimodal fusion computing model is a multi-task learning model; Based on the ratio of insulin concentration to blood glucose concentration in subcutaneous tissue fluid at each time point, the changing trend of this ratio is obtained by fitting, and the user's insulin resistance level is determined based on the changing trend.

4. The non-invasive method for detecting insulin concentration according to claim 3, characterized in that, Also includes: The multimodal fusion calculation model compensates for the time difference between the concentrations of insulin and / or blood glucose in tissue fluid and blood in real time based on the blood perfusion index, so as to obtain the concentrations of insulin and / or blood glucose in blood.

5. The non-invasive method for detecting insulin concentration according to claim 4, characterized in that, The insulin concentration and / or blood glucose concentration in the blood are first calibrated; wherein the first calibration method includes: correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model based on reference insulin concentration values ​​and / or reference blood glucose concentration values ​​at several reference times input by the user.

6. The non-invasive method for detecting insulin concentration according to claim 3, characterized in that, A second calibration is performed on the insulin concentration and / or blood glucose concentration in the blood; wherein the second calibration method includes: generating a corresponding correction factor based on the systematic characteristic deviation between the multimodal sensing signal at the reference time of the day and the pre-stored multimodal baseline signal, and systematically correcting the insulin concentration and / or blood glucose concentration in the blood output by the multimodal fusion calculation model within the day according to the correction factor.

7. The non-invasive method for detecting insulin concentration according to claim 3, characterized in that, It also includes using a pre-trained quality control model to evaluate the physical quality of each signal measured by each sensing module in real time to obtain a first confidence level, wherein the evaluation criteria for the physical quality include at least one of the signal-to-noise ratio and stability of the signal; using the multimodal fusion computing model to evaluate the degree of matching between its inference result and the preset physiological logic in real time to obtain a second confidence level; and outputting the inference result of the multimodal fusion computing model when the first confidence level and the second confidence level simultaneously meet their respective preset thresholds.

8. A non-invasive insulin concentration detection device, characterized in that, include: The flexible housing, flexible battery, flexible substrate, and flexible adhesive layer are stacked from top to bottom; among them, The flexible adhesive layer is used to attach to the user's skin surface, and several first windows are provided at corresponding positions on it so that the sensing module on the flexible substrate can detect the user's skin. The flexible substrate is stacked on top of the flexible adhesive layer, and a signal acquisition module and an insulin concentration generation module, comprising several sensing modules, are disposed thereon. The signal acquisition module includes a first sensing module, a second sensing module, and a third sensing module. The first sensing module acquires the dielectric constant of the user's subcutaneous tissue fluid, measured by a sensing module attached to the user's skin surface. The second sensing module acquires the user's physiological state, which includes at least one of resting, exercise, and stress. The physiological state is determined based on at least one of heart rate, heart rate variability, perfusion index, skin conductivity, and body surface temperature. The third sensing module acquires the user's surface-enhanced Raman scattering (SERS) spectral signal, obtained under near-infrared light source excitation. The insulin concentration generation module collaboratively determines the insulin concentration in the user's subcutaneous tissue fluid based on the dielectric constant of the subcutaneous tissue fluid, the physiological state, and the SERS spectral signal. The flexible battery is placed above the flexible substrate to supply power to the flexible substrate and shield the circuit signals on the flexible substrate from interference from the external environment. The flexible casing is placed over the flexible battery to protect the device.

9. The non-invasive insulin concentration detection device according to claim 8, characterized in that, The signal acquisition module further includes a fourth sensing module, which is used to acquire near-infrared spectral signals measured by a sensing module attached to the user's skin surface; the third and fourth sensing modules are disposed at the center of the flexible substrate, and the first sensing module is distributed around the center of the flexible substrate.

10. The non-invasive insulin concentration detection device according to claim 8, characterized in that, The second sensing module includes a skin conductivity detection unit; the skin conductivity detection unit includes two opposing electrodes.

11. The non-invasive insulin concentration detection device according to claim 10, characterized in that, The third sensing module is equipped with a near-infrared light source; when the skin conductivity detection unit on the second sensing module or the fifth sensing module included in the signal acquisition module detects that the flexible adhesive layer and the skin are not tightly adhered, the near-infrared light source in the third sensing module is turned off.

12. The non-invasive insulin concentration detection device according to claim 8, characterized in that, A second window is provided in the third sensing module area of ​​the flexible substrate. When the flexible substrate and the flexible adhesive are stacked, the second window coincides with a first window so that the near-infrared light source passes through the second window and the first window to irradiate the skin. A Raman enhancement substrate is provided on the side of the second window facing the skin. The Raman enhancement substrate is detachably disposed on the flexible substrate.

13. The non-invasive insulin concentration detection device according to claim 8, characterized in that, The third sensing module includes a Raman signal acquisition unit, and a flexible metal shielding foil is provided above the Raman signal acquisition unit.

14. An electronic terminal, characterized in that, include: Memory; and a processor coupled to the memory, the processor being configured to execute the noninvasive insulin concentration detection method according to any one of claims 1-7 based on instructions stored in the memory.

Images