Continuous blood glucose monitoring methods and systems based on individual body temperature compensation and aging correction

Abstract

This application relates to the field of medical monitoring technology and discloses a continuous blood glucose monitoring method and system based on individual body temperature compensation and aging correction. The method includes: acquiring raw current signals and measured body temperature data and aligning them in time to generate synchronous data pairs; calculating the rate of change of body temperature and inputting it into a glucose diffusion temperature sensitivity model to output a theoretical permeation rate reference sequence; using this sequence as a reference input, employing adaptive filtering to remove high-frequency noise from the raw current to generate a base signal; determining the degree of aging of the semi-permeable membrane based on the statistical characteristics of the base signal and calculating the aging attenuation coefficient; generating calibration parameter values ​​based on this coefficient; fusing the calibration parameter values ​​with real-time body temperature data to generate a dynamic compensation factor, correcting the base signal point-by-point to compensate for temperature sensitivity drift caused by aging; and converting the corrected current signal into a glucose concentration value. This application solves the problem of blood glucose signal calibration under the coupling interference of temperature and aging, improving the long-term accuracy of continuous blood glucose monitoring.

Description

Technical Field

[0001] This application relates to the field of medical monitoring technology, and in particular to a continuous blood glucose monitoring method and system based on individual body temperature compensation and aging correction. Background Technology

[0002] Continuous glucose monitoring (CGM) systems measure glucose concentration in tissue fluid in real time using an electrochemical sensor implanted under the skin. The sensor's output current signal is positively correlated with glucose concentration. However, fluctuations in body temperature affect the diffusion rate of glucose in tissue fluid and the reaction kinetics of the sensor electrode surface, causing the measured value to deviate from the true concentration. Simultaneously, the sensor's semi-permeable membrane ages over long-term use, gradually decreasing its permeability, further exacerbating the impact of temperature on measurement accuracy. Therefore, effectively compensating for measurement errors caused by body temperature changes and semi-permeable membrane aging in CGM has become a pressing technical problem to be solved in this field.

[0003] In existing technologies, calibration methods for the influence of body temperature typically employ static correction strategies. This involves pre-establishing empirical formulas for temperature and current output, and then applying a fixed coefficient to the current value based on the current body temperature during actual measurement. These methods do not consider the dynamic impact of body temperature change rates on glucose diffusion rates. When body temperature changes rapidly (e.g., during patient exercise, fever, or sudden changes in ambient temperature), static correction cannot track the changes in permeation rate caused by instantaneous temperature gradients, resulting in significant hysteresis errors. Another common method is to use low-pass filtering to smooth the raw current signal, attempting to filter out high-frequency noise caused by temperature fluctuations. However, simple filtering also filters out the true signal components during rapid changes in blood glucose, leading to response delays and peak flattening. Regarding the aging problem of the sensor's semi-permeable membrane, existing methods often rely on factory-preset attenuation curves for periodic calibration. However, differences in the physiological environments of different patients cause the actual aging rate to deviate from the preset curve. Adjusting calibration parameters too early or too late introduces systematic biases. Some systems use two-point or multi-point finger-prick blood calibration to correct aging drift, but this method requires frequent blood sampling, resulting in a poor user experience and failing to achieve continuous automatic compensation. In addition, existing technologies often process body temperature compensation and aging calibration separately, lacking a mechanism to integrate the two, which leads to a decrease in the aging compensation effect when the temperature changes, or a mismatch in temperature sensitivity compensation after aging.

[0004] To address the above deficiencies, this application combines a glucose diffusion temperature sensitivity model with adaptive filtering to filter out high-frequency noise caused by body temperature changes. It also integrates the semi-permeable membrane aging attenuation coefficient with real-time body temperature data to generate a dynamic compensation factor, thus solving the problem of accurate calibration of blood glucose signals under temperature and aging coupling interference and improving the long-term accuracy and environmental adaptability of continuous blood glucose monitoring. Summary of the Invention

[0005] This application provides a continuous blood glucose monitoring method and system based on individual body temperature compensation and aging correction, which solves the problem of accurate calibration of blood glucose signals under the coupling interference of temperature and aging, and improves the long-term accuracy and environmental adaptability of continuous blood glucose monitoring.

[0006] In a first aspect, this application provides a continuous blood glucose monitoring method based on individual body temperature compensation and aging correction, the method comprising: Step S1: Obtain the original current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, and perform time alignment processing on the original current signal sequence and the measured body temperature data sequence to generate a synchronized data pair; Step S2: Calculate the body temperature change rate based on the measured body temperature data sequence in the synchronized data pair, input the body temperature change rate into the glucose diffusion temperature sensitivity model, and output the theoretical glucose permeation rate reference sequence. Step S3: Using an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as the reference input, the original current signal sequence is filtered to remove high-frequency noise caused by body temperature changes, and a base signal sequence is generated. Step S4: Calculate statistical characteristic values ​​based on the substrate signal sequence, compare the statistical characteristic values ​​with the benchmark characteristic values, distinguish between aging attenuation and physiological fluctuations based on the comparison results, determine the degree of aging of the sensor semipermeable membrane, and calculate the aging attenuation coefficient. Step S5: Obtain the mapping relationship between sensor semi-permeable membrane aging and calibration parameters, determine the calibration parameter adjustment amount based on the aging attenuation coefficient and the mapping relationship, and generate calibration parameter values; Step S6: The calibration parameter value and real-time body temperature data are fused and calculated to generate a dynamic compensation factor. The base signal sequence is corrected point by point according to the dynamic compensation factor to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment. Step S7: Convert the corrected current signal sequence into glucose concentration values, and generate a calibration report based on the glucose concentration values.

[0007] Secondly, this application provides a continuous blood glucose monitoring system based on individual body temperature compensation and aging correction, used to implement the aforementioned continuous blood glucose monitoring method based on individual body temperature compensation and aging correction, the system comprising: The data synchronization module is used to acquire the raw current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, perform time alignment processing on the raw current signal sequence and the measured body temperature data sequence, and generate a synchronized data pair. The rate calculation module is used to calculate the body temperature change rate based on the measured body temperature data sequence in the synchronous data pair, input the body temperature change rate into the glucose diffusion temperature sensitivity model, and output a theoretical glucose permeation rate reference sequence. The signal filtering module is used to use an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as a reference input, to filter the original current signal sequence to remove high-frequency noise caused by body temperature changes and generate a base signal sequence. The aging judgment module is used to calculate statistical feature values ​​based on the substrate signal sequence, compare the statistical feature values ​​with the benchmark feature values, distinguish between aging attenuation and physiological fluctuations based on the comparison results, determine the degree of aging of the sensor semipermeable membrane, and calculate the aging attenuation coefficient. The parameter adjustment module is used to obtain the mapping relationship between the aging of the sensor semipermeable membrane and the calibration parameters, determine the adjustment amount of the calibration parameters according to the aging attenuation coefficient and the mapping relationship, and generate calibration parameter values. The dynamic compensation module is used to fuse the calibration parameter value with the real-time body temperature data to generate a dynamic compensation factor. Based on the dynamic compensation factor, the base signal sequence is corrected point by point to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment. The concentration conversion module is used to convert the corrected current signal sequence into glucose concentration values ​​and generate a calibration report based on the glucose concentration values.

[0008] Thirdly, this application provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the continuous blood glucose monitoring method based on individual body temperature compensation and aging correction.

[0009] This application proposes a continuous blood glucose monitoring method and system based on individual body temperature compensation and aging correction, which solves the problem of accurate calibration of blood glucose signals under the coupling interference of temperature and aging, and improves the long-term accuracy and environmental adaptability of continuous blood glucose monitoring. Compared with the prior art, the beneficial effects of the technical solution of this application are at least as follows: First, by calculating the rate of change in body temperature and inputting it into the glucose diffusion temperature sensitivity model, a theoretical glucose permeation rate reference sequence is output, which realizes the quantification of the dynamic influence of temperature gradient and avoids the lag error caused by the inability of static temperature correction to track instantaneous changes.

[0010] Second, an adaptive filtering algorithm is adopted, using the theoretical glucose osmotic rate reference sequence as the reference input, to filter out high-frequency noise caused by body temperature changes from the original current signal and generate a base signal sequence, which retains the real signal components when blood glucose changes rapidly, and avoids response delay and peak clipping caused by simple filtering.

[0011] Third, the statistical characteristic value within the sliding window is calculated based on the base signal sequence, and compared with the benchmark characteristic value to determine the degree of aging of the semipermeable membrane and calculate the aging attenuation coefficient. This enables dynamic tracking of the actual aging rate of the sensor and avoids systematic deviations caused by the deviation between the factory preset attenuation curve and individual differences.

[0012] Fourth, the mapping relationship between semipermeable membrane aging and calibration parameters is obtained, and calibration parameter values ​​are generated based on the aging attenuation coefficient, realizing adaptive adjustment of calibration parameters without requiring patients to frequently draw blood for finger-prick blood calibration.

[0013] Fifth, the calibration parameter values ​​are fused with real-time body temperature data to generate a dynamic compensation factor, and the base signal sequence is corrected point by point to compensate for the temperature sensitivity drift caused by aging, thus solving the coupling mismatch problem when body temperature compensation and aging calibration are processed separately. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the continuous blood glucose monitoring method based on individual body temperature compensation and aging correction in this application. Figure 2 This is a graph showing the 72-hour glucose concentration monitoring results in this application; Figure 3 Box plot of glucose concentration monitoring error distribution in this application; Figure 4 This is a schematic diagram of the continuous blood glucose monitoring system based on individual body temperature compensation and aging correction in this application. Detailed Implementation

[0016] This application provides a method and system for continuous blood glucose monitoring based on individual body temperature compensation and aging correction. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application 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 a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, 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.

[0017] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the continuous blood glucose monitoring method based on individual body temperature compensation and aging correction in this application includes: Step S1: Obtain the original current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, and perform time alignment processing on the original current signal sequence and the measured body temperature data sequence to generate a synchronized data pair.

[0018] In one specific embodiment, performing step S1 includes the following steps: The working electrode of the continuous glucose monitoring sensor is implanted under the skin to collect the raw current signal sequence in real time during the monitoring period. The raw current signal sequence is composed of microcurrents generated by the redox reaction of glucose molecules on the electrode surface. Synchronously acquire the measured body temperature data sequence with the timestamp corresponding to the original current signal sequence; Using the timestamp of the original current signal sequence as the main reference point, adjacent temperature sampling points are retrieved in the measured body temperature data sequence, and the fitted temperature value is calculated using a linear interpolation algorithm. The signal values ​​in the original current signal sequence and the corresponding fitted temperature values ​​are encapsulated by timestamps to generate one-to-one synchronous data pairs.

[0019] Specifically, the continuous glucose monitoring sensor is implanted in the subcutaneous tissue layer via skin implantation. The working electrode surface is equipped with a glucose oxidase catalytic layer. After glucose molecules in the tissue fluid diffuse to the electrode surface, they undergo an oxidation-reduction reaction under the action of the catalytic layer and release electrons. The electrons form a microcurrent in the external circuit. This microcurrent is continuously collected at a fixed sampling frequency to form a raw current signal sequence. The sampling frequency is set to once per minute, and the current signal value is at the nanoampere level. The signal is positively correlated with the glucose concentration. The raw current signal sequence is equipped with a timestamp. The timestamp uses an absolute time encoding method, accurate to the second. The analog current signal is amplified by a factor of 100 through a signal conditioning circuit, and then converted into a digital signal by a 12-bit resolution analog-to-digital converter. The digital signal and the timestamp are stored synchronously.

[0020] The temperature sensing element uses a thermistor, integrated into the sensor base and close to the implantation site. Its temperature measurement range covers the fluctuation range of human body temperature, with a measurement accuracy of 0.1 degrees Celsius. The temperature sensing element samples synchronously with the raw current signal at a sampling frequency of once per minute. The analog voltage signal output by the temperature sensing element is converted into a voltage change signal by a bridge circuit, then processed by an amplification circuit before being input to an analog-to-digital converter. The resulting digital temperature value is the measured body surface temperature. This measured body surface temperature value is then converted into an equivalent subcutaneous temperature value through a pre-calibrated body surface-to-subcutaneous temperature mapping relationship. The conversion model is as follows: ,in The measured body surface temperature at the current moment. The rate of change of body surface temperature is calculated from the difference between adjacent sampling points. , , The calibration constant is obtained by performing multiple linear regression fitting through synchronous acquisition of surface temperature and subcutaneous temperature data; the converted equivalent subcutaneous temperature value is stored synchronously as the measured body temperature data and the corresponding timestamp to form a measured body temperature data sequence, which maintains the time dimension acquisition consistency with the original current signal sequence.

[0021] A unified time reference axis is constructed using the timestamps of the original current signal sequence as the primary reference point. In the measured body temperature data sequence, a linear traversal algorithm is used to retrieve two temperature sampling points adjacent to the primary reference point. These two sampling points are located before and after the timestamp of the primary reference point. Based on the premise that temperature changes are linear within a short time interval, a linear interpolation algorithm is used to calculate the fitted temperature value at the corresponding time point of the primary reference point. The formula for the linear interpolation algorithm is as follows: , in the formula Represents the fitted temperature value. The values ​​are the temperature sampling points with the earliest timestamps. The values ​​are from temperature sampling points with later timestamps. Main reference point timestamp, This is the timestamp of the previous temperature sampling point. For example, the timestamp of the next temperature sampling point is the 100th hour. In the measured body temperature data sequence, the previous temperature sampling point is 36.5 degrees Celsius at 99 hours and 59 minutes, and the next temperature sampling point is 37.2 degrees Celsius at 101 minutes. The calculated fitted temperature value for the 100th hour is 36.85 degrees Celsius.

[0022] The current signal values ​​in the original current signal sequence are associated and encapsulated with the fitted temperature values ​​obtained through linear interpolation according to the same timestamp, forming a synchronous data pair. Each synchronous data pair contains three types of data: timestamp, current signal value, and fitted temperature value. The three types of data maintain a one-to-one correspondence. The synchronous data pairs are arranged in chronological order according to their timestamps to form a continuous sequence and stored in a circular buffer. The circular buffer can hold 72 hours of synchronous data pairs. When new data is written, the earliest data is overwritten. The synchronous data pairs provide a time-dimensional matching data basis for calculating the rate of change of body temperature. The current signal value and the fitted temperature value are bound together through timestamps to eliminate time deviations caused by hardware clock differences and system wake-up delays.

[0023] Step S2: Calculate the body temperature change rate based on the measured body temperature data sequence in the synchronized data pair, input the body temperature change rate into the glucose diffusion temperature sensitivity model, and output the theoretical glucose permeation rate reference sequence.

[0024] In one specific embodiment, performing step S2 includes the following steps: For the measured body temperature data sequence in the synchronous data pair, the ratio of the body temperature difference between adjacent time points to the time interval is calculated to obtain the body temperature change rate sequence; The body temperature change rate sequence was input into a pre-calibrated glucose diffusion temperature sensitivity model to calculate the effect of the theoretical temperature gradient on the glucose permeation rate at each time point. Based on the influence values, a smooth theoretical glucose permeation rate reference sequence is generated using cubic spline interpolation.

[0025] Specifically, a sequence of measured body temperature data is extracted from the synchronized data pairs. Each data point in this sequence corresponds to a specific timestamp and its corresponding body temperature measurement value. For the measured body temperature data sequence in the synchronized data pairs, the rate of change of body temperature needs to be calculated. This is done by taking the body temperature value at the current time point and the body temperature value at the immediately preceding time point, calculating the difference between the two, and then dividing this difference by the time interval between the two time points. This time interval is determined by the sensor's sampling frequency; for example, if the sampling period is set to 1 minute, then the time interval between adjacent time points is 1 minute. For the first time point in the entire sequence, since there is no previous time point to refer to, the rate of change cannot be calculated, so its rate of change is set to 0. Following this process, starting from the second time point, the above operation of dividing the difference by the interval is performed sequentially for each time point. The results of each calculation are recorded in chronological order, forming a sequence of body temperature change rates. Each value in this sequence represents the instantaneous rate of increase or decrease of body temperature at the corresponding moment, and its unit is degrees Celsius per minute. For example, when a patient starts exercising from a resting state, their body temperature may rise from 36.5 degrees Celsius to 37.2 degrees Celsius within 10 minutes. At the beginning of the exercise, the temperature difference between adjacent sampling points is relatively large, and the calculated rate of change of body temperature will show a positive peak, which is about 0.07 degrees Celsius per minute. This peak accurately reflects the instantaneous state of rapid rise in body temperature.

[0026] After obtaining the body temperature change rate sequence, it is input into a pre-calibrated glucose diffusion temperature sensitivity model. The glucose diffusion temperature sensitivity model is a mathematical relationship specifically designed to describe the influence of temperature gradients on the diffusion rate of glucose molecules in a sensor's semi-permeable membrane. It is constructed based on Fick's first diffusion law and introduces a temperature gradient compensation term to handle the nonlinear effects of rapid body temperature changes. The model contains three core parameters: the basic diffusion coefficient... Temperature gradient response coefficient and nonlinear adjustment factors Basic diffusion coefficient This represents the inherent diffusion capacity of glucose molecules across a semipermeable membrane under standard constant body temperature conditions; temperature gradient response coefficient. Used to quantify the linear regulation of glucose osmotic rate by a unit change in body temperature; nonlinear regulation factor This is used to correct the nonlinear response characteristics of the diffusion rate during rapid body temperature fluctuations, avoiding deviations in calculated values ​​from physical reality caused by extreme body temperature changes. The values ​​of the above three parameters are obtained by fitting the temperature-changing experimental data of the newly manufactured sensor under simulated physiological conditions in vitro. In a specific embodiment, the calibration process is as follows: The working electrode of the continuous blood glucose monitoring sensor from the same batch is placed in a phosphate buffer solution. The glucose concentration of the buffer solution is set to 5.5 mmol / L to simulate normal human blood glucose levels, and the ionic strength and pH value are adjusted to a physiological range similar to that of subcutaneous tissue fluid. The experimental environment temperature is precisely controlled by a programmable thermostat, with a temperature measurement accuracy of 0.05 degrees Celsius. During the experiment, the thermostat temperature is controlled within the range of 35.0 degrees Celsius to 39.0 degrees Celsius, and the temperature is cycled up and down at three constant rates of 0.1 degrees Celsius per minute, 0.3 degrees Celsius per minute, and 0.5 degrees Celsius per minute, while the original current signal sequence output by the sensor is recorded simultaneously. With the rate of temperature change as the independent variable and the deviation ratio of the current signal from the reference temperature as the dependent variable, a quadratic polynomial fitting is performed using the least squares method. The fitting results show that the quadratic polynomial can well describe the nonlinear relationship between the rate of temperature change and the change in permeation rate. The basic diffusion coefficient was verified through repeated experiments with multiple batches of sensors. The value ranges from 0.82 to 0.88, with a typical value of 0.85; temperature gradient response coefficient. The value ranges from 0.04 to 0.06, with a typical value of 0.05; nonlinear adjustment factor. The value ranges from 0.01 to 0.03, with a typical value of 0.02. The rate of change at each time point in the body temperature change rate sequence is substituted into the model to calculate the effect of the theoretical temperature gradient at that time point on the glucose osmotic rate. The calculation formula is: ,in, This represents the rate of change at each time point in the body temperature change rate sequence, with the quadratic term representing the value. This method is used to capture the nonlinear effects of rapid changes in body temperature. When the absolute value of the rate of change in body temperature is less than 0.1 degrees Celsius per minute, the quadratic term contributes little; when the absolute value of the rate of change in body temperature is greater than 0.3 degrees Celsius per minute, the quadratic term produces a significant correction. Through this calculation process, a corresponding theoretical temperature gradient influence value is obtained for each time point. This value quantifies the instantaneous promoting or inhibiting effect of a temperature gradient on glucose molecules crossing a semipermeable membrane at the current rate of body temperature change. For example, with a baseline diffusion coefficient of 0.85, a temperature gradient response coefficient of 0.05, a nonlinear regulation factor of 0.02, and a body temperature change rate of 0.1 degrees Celsius per minute, the calculated value is... Increased body temperature leads to a slight increase in the osmosis rate.

[0027] After calculation using the glucose diffusion temperature sensitivity model, each time point in the body temperature change rate sequence yields a set of discrete theoretical temperature gradient influence values, each corresponding to its respective timestamp. Since the sampling period for the sensor acquiring the raw current signal is fixed at 1 minute, and the time interval in the body temperature change rate calculation process is consistent with the sampling period, the resulting influence value sequence exhibits an evenly spaced distribution on the time axis. To obtain a continuous and smooth theoretical glucose osmosis rate reference sequence, a cubic spline interpolation algorithm is used to fit the discrete influence values. The basic idea of ​​cubic spline interpolation is to construct a cubic polynomial between every two adjacent discrete data points, ensuring that the entire sequence has continuous first and second derivatives at the connection points, thus generating a smooth curve. In specific implementation, discrete time points are set... and their corresponding impact values In the interval Constructing a cubic polynomial ,in These are undetermined coefficients. The constraints include: , , , Furthermore, natural boundary conditions are set at both endpoints. and , and They represent the first i The cubic polynomial constructed on each interval The first and second derivatives are obtained. By solving the linear equations formed by these constraints, the coefficients of the cubic polynomial in each interval can be uniquely determined. Then, for each time point corresponding to each timestamp in the original current signal sequence, the time point is substituted into the cubic polynomial corresponding to the interval to calculate the interpolation result at that moment. By traversing all time points in chronological order and arranging the interpolation results corresponding to each time point in sequence, a smooth theoretical glucose osmotic rate reference sequence is generated. Each value in this sequence represents the theoretical glucose osmotic rate driven entirely by the current body temperature fluctuation under ideal conditions of the sensor's semi-permeable membrane. For example, if the discrete influence value gradually increases from 0.95 to 1.02 and then decreases to 0.98 over a certain period of time, cubic spline interpolation will generate a smooth curve in the transition region between the increase and decrease, avoiding the broken line shape that may be produced by linear interpolation, making the change in the theoretical osmotic rate more consistent with the continuity of the physiological process.

[0028] Step S3: Using an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as the reference input, the original current signal sequence is filtered to remove high-frequency noise caused by body temperature changes, and a base signal sequence is generated.

[0029] In one specific embodiment, performing step S3 includes the following steps: The theoretical glucose osmotic rate reference sequence is used as the reference input signal for the least mean square adaptive filtering algorithm. Based on the reference input signal, the weight coefficient vector is iteratively adjusted point by point for the original current signal sequence to identify high-frequency noise components related to body temperature changes. The identified high-frequency noise components are subtracted point by point from the original current signal sequence to generate a substrate signal sequence that characterizes the inherent permeability resistance of the sensor's semipermeable membrane.

[0030] Specifically, a least mean square adaptive filtering algorithm is used to process the original current signal sequence and the theoretical glucose permeation rate reference sequence. The original current signal sequence is acquired by the implanted working electrode of the continuous glucose monitoring sensor. Each current value in this sequence corresponds to a timestamp, reflecting the microcurrent intensity generated by the redox reaction of glucose molecules on the electrode surface. The theoretical glucose permeation rate reference sequence is generated in step S2, and each value represents the theoretical permeation rate that glucose should achieve under the current measured rate of body temperature change and when the semipermeable membrane is in an ideal, unaged state. The two sequences are in one-to-one correspondence in time, forming a synchronized input-output relationship.

[0031] Before running the least mean square adaptive filtering algorithm, the filter order and convergence factor need to be set. The filter order determines the number of historical points the algorithm remembers. Based on the sensor sampling period of once per minute and the main cycle of body temperature fluctuations being several minutes to tens of minutes, the filter order is set to fifth order. This means that each iteration uses the current time point and the reference input values ​​from the previous four time points to form the input vector. The convergence factor controls the adjustment step size of the weight coefficient vector and is set to the reciprocal of the reference input signal power multiplied by 0.1. The reference input signal power is estimated by the square mean within a sliding window. This way, the step size automatically decreases when the signal fluctuation is large to ensure stability, and the step size is moderately increased when the signal is stable to accelerate convergence.

[0032] The algorithm iterates point-by-point from the first time point of the original current signal sequence. At time point t, the values ​​of that time point and the previous four time points in the theoretical glucose osmotic rate reference sequence are extracted and arranged into an input vector X(t). The current weight coefficient vector W(t) is a vector with the same dimension as the input vector, with all components initially set to zero. The algorithm calculates and estimates the noise value. The estimated noise value represents the high-frequency noise component in the original current signal that the algorithm believes should be caused by the temperature gradient under the current body temperature change conditions. Subsequently, the actual current value d(t) at the same time point in the original current signal sequence is read, and the error signal is calculated. The error signal e(t) represents the portion of the actual current not explained by the theoretical osmotic rate, i.e., the remaining component after excluding the influence of body temperature fluctuations. Theoretically, it should mainly reflect the current basis corresponding to the inherent osmotic resistance of the semipermeable membrane. Based on the least mean square criterion, the algorithm updates the weight coefficient vector using the error signal e(t) and the input vector X(t), with the update formula as follows: ,in The convergence factor is used. This update process moves the weight coefficient vector along the negative gradient direction of the error performance surface, gradually approximating the Wiener solution. As time progresses, the algorithm repeatedly performs the above calculations: constructing the input vector, calculating the estimated noise, calculating the error, and updating the weight coefficients. After several iterations, the weight coefficient vector converges to the optimal value, at which point the estimated noise value y(t) can accurately fit the high-frequency noise components related to body temperature changes in the original current signal.

[0033] After calculating the estimated noise value at each time point, the algorithm subtracts the estimated noise value y(t) from the original current signal value d(t) to obtain the base signal value. Arranging the baseline signal values ​​at all time points in chronological order forms the baseline signal sequence. Each value in this sequence represents the current component contributed solely by the inherent osmotic resistance of the sensor's semipermeable membrane, after eliminating instantaneous osmotic rate fluctuations caused by body temperature changes. The inherent osmotic resistance of the semipermeable membrane is affected by long-term factors such as the degree of hydration of the membrane material, protein adsorption, and changes in micropore structure. Its frequency of change is much lower than the high-frequency noise caused by body temperature fluctuations. Therefore, the baseline signal sequence shows a slow trend of change on the time axis, while the high-frequency components have been effectively suppressed by adaptive filtering. For example, when a patient changes from a resting state to an active state, the body temperature rises from 36.6 degrees Celsius to 37.3 degrees Celsius within ten minutes, and a sharp rise independent of blood glucose concentration appears in the original current signal sequence. The theoretical glucose osmotic rate reference sequence calculates the theoretical increase in osmotic rate during this time period based on the measured body temperature change rate. This sequence exhibits a similar rising waveform within the corresponding time interval. The least mean square adaptive filtering algorithm uses this theoretical sequence as a reference and iteratively adjusts the weight coefficient vector point by point to make the estimated noise value approximate the rising peak in the original current signal. After subtracting this estimated noise from the original current signal, the base signal sequence remains stable within this time interval, avoiding misinterpreting the increase in current caused by body temperature as an increase in blood sugar. If the sensor's semi-permeable membrane has aged, the base signal sequence will show an overall slow downward trend, which is independent of body temperature fluctuations.

[0034] Step S4: Calculate the statistical characteristic value based on the base signal sequence, compare the statistical characteristic value with the benchmark characteristic value, distinguish between aging attenuation and physiological fluctuations based on the comparison result, determine the degree of aging of the sensor semipermeable membrane, and calculate the aging attenuation coefficient.

[0035] In one specific embodiment, performing step S4 includes the following steps: A sliding window of a preset time length is selected to traverse the base signal sequence, and the statistical characteristic values ​​of the signal within each window are calculated. The statistical characteristic values ​​include at least the signal mean and the signal variance. Determine whether the signal variance is within the preset variance range. If so, determine that there is no abnormal interference in the current window signal and use the signal mean of the window as the effective mean. The effective mean is compared with the baseline characteristic value during the initial calibration phase of the sensor to determine whether the signal strength of the base signal sequence has decreased. If not, then the original aging degradation coefficient remains unchanged; If so, then further obtain the effective mean sequence of N consecutive windows and calculate the rate of change of signal strength, where N is an integer greater than 5; If the absolute value of the rate of change is less than a preset change threshold and the duration of the decrease exceeds a preset duration, it is determined that the decrease in signal strength is caused by the aging of the semipermeable membrane, and the aging attenuation coefficient is calculated based on the difference between the effective mean and the reference characteristic value. If the absolute value of the rate of change is greater than or equal to the preset change threshold and a rebound occurs in the subsequent window, it is determined that the decrease in signal intensity is caused by a physiological decrease in blood glucose concentration, and the original aging attenuation coefficient remains unchanged.

[0036] Specifically, the base signal sequence is a current signal sequence characterizing the inherent permeability resistance of the sensor's semipermeable membrane after adaptive filtering to remove high-frequency noise caused by changes in body temperature. Each data point in the sequence is the current signal value at the corresponding timestamp, measured in nanoamperes. The sampling frequency is 1 time / minute, and the sequence length covers the duration of real-time monitoring by the sensor. First, the parameters of the sliding window are set. The preset time length of the sliding window is set to 30 minutes, and the window step size is set to 5 minutes. These parameters are determined based on the fluctuation cycle of glucose concentration in human subcutaneous tissue fluid and the time scale of changes in the aging signal of the semipermeable membrane. This ensures coverage of short-term signal fluctuations while avoiding delays in aging feature extraction due to an excessively long window and deviations in statistical feature calculation due to an excessively short window. The sliding window starts from the initial data point of the base signal sequence and moves sequentially along the time axis. After each movement, all current signal data points within the window are extracted to form independent signal data segments, until the entire base signal sequence has been traversed. After completing a full coverage traversal of the base signal sequence, for each signal data segment extracted from a sliding window, the signal mean and signal variance are calculated. The signal mean is the arithmetic mean of all current signal values ​​within the window, reflecting the average current level corresponding to the inherent permeability resistance of the semipermeable membrane within the window. The signal variance is the average of the sum of squares of the differences between each current signal value and the signal mean within the window, characterizing the degree of signal fluctuation within the window, and is used to identify abnormal interference in the signal, such as instantaneous signal disturbances caused by movement or changes in body position.

[0037] The preset variance interval is determined through sensor factory calibration experiments. The calibration process is carried out under a standard physiological environment of constant temperature and rest. The base signal sequence of the brand-new sensor is collected, and the signal variance within different windows is calculated. The variance range corresponding to the 95% confidence interval is taken as the preset variance interval. The signal fluctuation within this interval is only generated by the sensor's inherent noise and there is no external abnormal interference. For example, for a certain model of continuous blood glucose monitoring sensor, 30 brand-new sensor samples are continuously monitored for 24 hours under a standard constant temperature environment of 37 degrees Celsius. The base signal sequence is collected and the signal variance within the sliding window is calculated. The statistical results show that under the condition of no external interference, the 95% confidence interval of the base signal variance is 0.02 to 0.08. Based on this, the preset variance interval is set to [0.02, 0.08]. The signal variance of the current window is compared with a preset variance interval. If the signal variance falls within the preset variance interval (e.g., 0.03), it is determined that there is no abnormal interference within the current window, and the signal mean of the window is taken as the effective mean. If the signal variance exceeds the preset variance interval (e.g., 0.1), it is determined that there is abnormal interference within the current window, and the statistical characteristic value of the window is discarded and not included in subsequent aging judgment. The baseline characteristic value is the mean value of the base signal obtained during the initial calibration phase of the sensor. The initial calibration phase is the 24-hour hydration stabilization period after the sensor is implanted subcutaneously. During this phase, the semipermeable membrane has not aged and is in a brand-new state. Base signal sequences are collected within the standard body temperature range, and the signal mean within the sliding window is calculated. The average value of multiple calculations is taken as the baseline characteristic value. This value is the mean value of the current signal corresponding to the inherent permeation resistance of the semipermeable membrane in the unaged state, and is used as the reference benchmark for aging judgment.

[0038] The effective mean is compared with the benchmark characteristic value. If the effective mean is not less than the benchmark characteristic value, the original aging degradation coefficient is maintained.

[0039] If the effective mean is less than the baseline characteristic value, it indicates a decrease in the signal intensity of the base signal sequence. This cannot be directly attributed to semipermeable membrane aging, as physiological decreases in blood glucose concentration (such as nocturnal hypoglycemia or fasting hypoglycemia) can also lead to a decrease in the current signal. Therefore, it is necessary to further obtain an effective mean sequence for six consecutive windows (corresponding to a 30-minute time span), perform linear fitting on this sequence using the least squares method, calculate the rate of change of signal intensity (unit: nanoamperes per minute), and record the duration of continuous decline in signal intensity from the first time it falls below the baseline characteristic value (unit: minutes), while simultaneously monitoring whether a rebound in the signal mean occurs in subsequent windows. The absolute value of the rate of change is determined to be less than a preset threshold and the duration of continuous decline exceeds a preset duration. The rate of change threshold is determined through an offline calibration experiment: 72-hour continuous blood glucose monitoring data are collected from 30 subjects, and finger-prick blood glucose reference values ​​are simultaneously obtained. Hypoglycemic event windows (blood glucose concentration below 3.9 mmol / L) and normal blood glucose windows are labeled, and the distribution of the base signal change rate within the two types of windows is statistically analyzed. The results showed that the signal decline rate within the hypoglycemic event window was typically greater than 0.5 nanoamps per minute, while the signal decline rate caused by semipermeable membrane aging was typically less than 0.1 nanoamps per minute. Based on this, a rapid change threshold of 0.5 nanoamps per minute was set, a slow change threshold of 0.1 nanoamps per minute was set, and a duration threshold of 4 hours was set. In the first case: if the absolute value of the change rate is less than the slow change threshold of 0.1 nanoamps per minute, and the duration reaches more than 4 hours, then the signal intensity decline is determined to be caused by slow aging of the semipermeable membrane. Semipermeable membrane aging manifests as micropore shrinkage, decreased hydrophilicity, and protein adsorption, leading to increased glucose osmotic resistance and weakened current signal generated by the redox reaction on the electrode surface. At this point, the degree of aging increases, and the aging attenuation coefficient is calculated as the ratio of the effective mean to the baseline characteristic value: the aging attenuation coefficient equals 1 minus the effective mean divided by the baseline characteristic value. For example, the baseline characteristic value of the sensor in the initial calibration stage is 100 nanoamps. The variance of the base signal within a certain sliding window falls within a preset variance range, with an effective mean of 85 nanoamps. The calculated aging attenuation coefficient is 0.15, which indicates that after aging, the permeability of the semipermeable membrane is only 85% of its initial state, a 15% decrease compared to the initial state. In the second scenario: if the absolute value of the rate of change is greater than or equal to the rapid change threshold of 0.5 nanoamps per minute, and a rebound occurs in subsequent windows (the rebound exceeds 50% of the decrease), then the signal intensity decrease is determined to be caused by a physiological decrease in blood glucose concentration. The previously effective aging attenuation coefficient is maintained, and the calibration report is marked "Possible hypoglycemia" to alert the user.The third scenario: If the absolute value of the rate of change is between the slow change threshold of 0.1 nanoamps per minute and the fast change threshold of 0.5 nanoamps per minute, or if the duration is less than 4 hours, then the cause of the signal decline cannot be clearly distinguished. In this case, a conservative strategy is adopted: do not update the aging attenuation coefficient, keep the original coefficient unchanged, and wait for more window data before making a judgment.

[0040] Step S5: Obtain the mapping relationship between the aging of the sensor semi-permeable membrane and the calibration parameters, determine the adjustment amount of the calibration parameters based on the aging attenuation coefficient and the mapping relationship, and generate the calibration parameter values.

[0041] In one specific embodiment, performing step S5 includes the following steps: Obtain a pre-calibrated mapping table between the aging degree of the sensor semi-permeable membrane and the initial calibration parameters. The mapping table records the sensitivity correction factor corresponding to different aging attenuation coefficient ranges. Using the aging attenuation coefficient as an index, retrieve the corresponding sensitivity correction factor; Multiply the sensitivity correction factor by the initial calibration parameters to generate calibration parameter values ​​that correct for aging errors.

[0042] Specifically, the mapping relationship between the aging degree of the sensor's semi-permeable membrane and the initial calibration parameters is a two-dimensional data set calibrated before the continuous blood glucose monitoring system leaves the factory. The initial calibration parameters are the current signal and glucose concentration conversion parameters obtained by electrochemical calibration of the continuous blood glucose monitoring sensor in a brand-new state, including the sensitivity coefficient and baseline offset. The sensitivity coefficient represents the amplitude of the current signal change corresponding to a unit glucose concentration, and the baseline offset is the inherent current output value of the sensor under glucose-free conditions. The calibration of the mapping table was carried out in a standard physiological simulation environment at a constant temperature of 37 degrees Celsius. Sensor samples were obtained using gradient accelerated aging. The aging attenuation coefficient ranges were divided into 0 to 0.1, 0.1 to 0.2, 0.2 to 0.3, 0.3 to 0.4, 0.4 to 0.5, and 0.5 to 1.0. Fifty sensor samples were selected for each range. The baseline signal sequence of each sample was collected in a standard glucose concentration solution. The effective mean of the samples was calculated and combined with the baseline characteristic value to determine the corresponding aging attenuation coefficient. The sensitivity correction factor required to maintain measurement accuracy under each aging state was simultaneously determined. The sensitivity correction factor is a dimensionless value used to adjust the sensitivity coefficient of the initial calibration parameters. The second-order least squares method was used to fit the ranges and the sensitivity correction factor. The fitting residual was controlled within 0.001. The fitted aging attenuation coefficient ranges and the sensitivity correction factors were established in a one-to-one correspondence. The mapping table was constructed in the form of storing the aging attenuation coefficient ranges in rows and the sensitivity correction factors in columns.

[0043] The aging attenuation coefficient is a dimensionless value calculated from the effective mean of the base signal sequence and the benchmark characteristic value. It directly characterizes the degree of glucose permeability degradation caused by the aging of the semipermeable membrane. After the system calls the mapping table, it uses an interval traversal matching algorithm to match the real-time calculated aging attenuation coefficient with the intervals in the table. The matching process follows the left-closed and right-open interval determination rule, that is, when the aging attenuation coefficient is equal to the lower limit of the interval, it belongs to the current interval, and when it is equal to the upper limit of the interval, it belongs to the next adjacent interval. The calculation precision of the traversal matching is set to four decimal places. The response time of the system digital signal processor to complete a single matching does not exceed 10 milliseconds. After the matching is completed, the sensitivity correction factor corresponding to the target aging attenuation coefficient interval is directly extracted. This retrieval process establishes a direct data correspondence between the aging attenuation coefficient and the sensitivity correction factor, ensuring that the correction factor and the real-time aging status of the sensor maintain data-level consistency.

[0044] The calibration parameter value is generated by multiplying the sensitivity correction factor with the initial calibration parameter. The operation is performed only on the sensitivity coefficient in the initial calibration parameter, while the baseline offset remains unchanged from the initial calibration value. The calculation formula is U=E×F, where E is the sensitivity coefficient in the initial calibration parameter, F is the retrieved sensitivity correction factor, and U is the generated calibration parameter value. The operation is performed using a 32-bit floating-point data format to avoid precision loss caused by integer operations. For example, if the initial sensitivity coefficient E is 12.5 nA / mmol / L and the matched sensitivity correction factor F is 0.85, the calculated calibration parameter value U is 10.625 nA / mmol / L. The generated calibration parameter value is directly transmitted to the dynamic compensation stage as the benchmark parameter for body temperature data fusion calculation, forming a parameter linkage relationship with the real-time body temperature data, and providing a fixed calibration benchmark for the subsequent generation of the dynamic compensation factor.

[0045] Step S6: The calibration parameter values ​​and real-time body temperature data are fused and calculated to generate a dynamic compensation factor. The base signal sequence is corrected point by point according to the dynamic compensation factor to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment.

[0046] In one specific embodiment, performing step S6 includes the following steps: Obtain the instantaneous body temperature value at the current moment as real-time body temperature data; Calculate the instantaneous change in the current temperature gradient based on real-time body temperature data; The calibration parameter value is used as the compensation reference. The instantaneous change of temperature gradient is multiplied by the preset temperature sensitivity coefficient to obtain the dynamic adjustment term. The compensation reference and the dynamic adjustment term are added to obtain the dynamic compensation factor. The preset temperature sensitivity coefficient is a pre-calibrated constant. The substrate signal sequence is corrected point by point based on the dynamic compensation factor to compensate for the temperature sensitivity drift caused by the aging of the sensor semipermeable membrane, and a corrected current signal sequence is generated.

[0047] Specifically, it should be noted that steps S3 and S6 are not redundant compensations for the same disturbance source. Step S3 filters out high-frequency dynamic noise caused by instantaneous fluctuations in body temperature. This filtering process is based on the ideal temperature sensitivity model of the sensor before aging, which is pre-calibrated and fixed at the factory. When the sensor's semi-permeable membrane ages, its actual temperature sensitivity coefficient (i.e., the current response amplitude corresponding to a unit change in body temperature) drifts, deviating from the factory-calibrated ideal model. At this point, the filtering process in step S3 cannot completely eliminate the temperature effect due to model mismatch, and some error components related to body temperature changes still remain in the filtered base signal. The dynamic compensation factor in step S6 corrects this "temperature sensitivity drift caused by aging," rather than performing secondary compensation for the current body temperature change itself. The two processes have different physical objects: S3 deals with dynamic interference caused by body temperature changes, while S6 deals with model parameter deviations caused by aging. They work together; the former is responsible for eliminating instantaneous noise, and the latter is responsible for calibrating long-term drift. There is no logical risk of redundant compensation or overcompensation.

[0048] Real-time body temperature data is the instantaneous body temperature value obtained by the continuous glucose monitoring sensor at the moment of data acquisition after conversion based on the surface-to-subcutaneous temperature mapping relationship. This value is collected by a temperature sensing element integrated into the sensor base, and after signal conditioning circuitry and analog-to-digital conversion, it becomes a digital body temperature value, maintaining a one-to-one correspondence with the base signal sequence timestamp. The sampling frequency of the temperature sensing element is consistent with the sampling frequency of the current signal, both being once per minute. The temperature measurement accuracy is controlled within 0.1 degrees Celsius, covering the common temperature fluctuation range of human subcutaneous tissue, ensuring that the real-time body temperature data can accurately reflect the local temperature state of the sensor implantation site, with instantaneous temperature gradient changes. The instantaneous change in body temperature is used to characterize the magnitude of change in body temperature per unit time. It is calculated by dividing the difference between the instantaneous body temperature at the current moment and the instantaneous body temperature at the previous sampling moment by the time interval between the two samplings. The time interval is determined by the fixed sampling period of the sensor and is 60 seconds. This calculation process is based on the linear difference algorithm and is performed in 32-bit floating-point format, retaining four significant digits after the decimal point to avoid calculation errors caused by data truncation. The instantaneous change in temperature gradient can quantify the dynamic trend of body temperature changes, distinguish between different states such as rapid rise, slow rise, constant, and fall, and provide a basis for the generation of dynamic adjustment terms.

[0049] The preset temperature sensitivity coefficient is a constant calibrated on a constant-temperature variable-temperature experimental platform before the continuous blood glucose monitoring sensor leaves the factory. The calibration process is carried out within the common human body temperature range of 35°C to 40°C. By gradually adjusting the ambient temperature, the response amplitude of the sensor's output current signal to temperature changes is measured. A linear regression algorithm is used to fit the correspondence between the temperature change and the current signal change. The coefficient of determination of the regression fitting is controlled above 0.99, and a fixed value of the temperature sensitivity coefficient is finally determined. This coefficient is stored in the system storage medium. In a specific embodiment, for a typical sensor using a platinum-iridium alloy working electrode and a glucose oxidase membrane layer, the preset temperature sensitivity coefficient obtained by linear regression fitting within the calibration range of 35°C to 40°C ranges from 0.008 to 0.015 nA / °C, with a typical value of 0.012 nA / °C. This coefficient represents the change in sensor output current (unit: nanoamperes) caused by a unit change in body temperature of 1°C under conditions without aging effects. The dynamic adjustment term is directly invoked during the fusion calculation process and does not change with usage time or sensor status. It is obtained by multiplying the instantaneous change in temperature gradient with a preset temperature sensitivity coefficient. The calculation process maintains the same data accuracy as the instantaneous change in temperature gradient. The sign and magnitude of the dynamic adjustment term change synchronously with the body temperature change state. The dynamic adjustment term is positive when the body temperature rises, negative when the body temperature falls, and approaches zero when the body temperature remains constant. It can reflect the instantaneous impact of temperature changes on glucose diffusion rate in real time. The dynamic compensation factor is generated by adding the calibration parameter value to the dynamic adjustment term. The compensation benchmark is the calibration parameter value generated in step S5, representing the fixed calibration benchmark under the aging state of the semipermeable membrane. The dynamic adjustment term is superimposed on the compensation benchmark to form a composite compensation parameter that takes into account both aging calibration and the dynamic influence of body temperature. There is no data truncation or rounding in the addition process, which fully preserves the calculation accuracy. The dynamic compensation factor changes synchronously with real-time body temperature data and aging state, realizing real-time dynamic adjustment of calibration parameters.

[0050] The base signal sequence is a current signal sequence characterizing the inherent osmotic resistance of the sensor's semipermeable membrane after adaptive filtering to remove high-frequency noise related to body temperature. Each signal point in the sequence carries an independent timestamp. The dynamic compensation factor and the base signal sequence are matched point by point according to the timestamp. The point-by-point correction process uses multiplication, multiplying the base signal value corresponding to each timestamp by the dynamic compensation factor at the same timestamp. The resulting value is the current signal value after removing the coupling interference of aging and temperature. After all time points have been processed, they are arranged in chronological order to form the corrected current signal sequence. This sequence eliminates the temperature sensitivity drift caused by semipermeable membrane aging and retains the true signal components corresponding to changes in glucose concentration.

[0051] Step S7: Convert the corrected current signal sequence into glucose concentration values, and generate a calibration report based on the glucose concentration values.

[0052] In one specific embodiment, performing step S7 includes the following steps: The corrected current signal sequence is substituted into the pre-calibrated formula to complete the electrochemical conversion and obtain the glucose concentration value; Compare the glucose concentration value with the preset normal range; if it exceeds the range, mark the abnormal range and trigger an alert. Integrate glucose concentration value sequences, anomaly markers, warning information, and timestamps to generate a calibration report containing blood glucose change curves.

[0053] Specifically, the corrected current signal sequence is a current signal sequence that has been dynamically compensated point by point and has eliminated the coupling interference from semi-permeable membrane aging and body temperature fluctuations. This sequence retains the true electrochemical signal directly related to glucose concentration, with signal values ​​in nanoamperes. Each value in the sequence is bound to a unique timestamp, consistent with the sensor's sampling period. The pre-calibrated formula is derived from the standard solution calibration experiment during the sensor's factory manufacturing stage. The experiment uses gradient concentration glucose standard solutions and is conducted under a standard environment with constant temperature and humidity. By collecting stable current values ​​output by the sensor at different concentrations, the least squares method is used for linear fitting to determine the linear transformation relationship. The calibration formula adopts... Where C is the glucose concentration value, K is the sensitivity coefficient calibrated by the sensor at the factory, I is the corrected current signal value, and B is the baseline offset calibrated by the factory. K and B are fixed constants calibrated through multiple sets of repeated experiments. The calibration process is repeated no less than 30 times, and the fitting determination coefficient is no less than 0.995. The conversion process is executed point by point by the processor in 32-bit floating-point arithmetic format. Each current value in the corrected current signal sequence is substituted into the formula to calculate the glucose concentration value at the corresponding time. The concentration value unit is uniformly in millimoles per liter, and the result is retained to two decimal places to meet the accuracy specifications of clinical blood glucose monitoring.

[0054] After the glucose concentration value is calculated, it is compared with the preset normal blood glucose concentration range in the system. The preset normal range is set according to clinical diagnostic standards. The comparison logic adopts the numerical range judgment rule, comparing the current concentration value with the upper limit and lower limit of the normal range in turn. When the concentration value is higher than the upper limit, the interval where the current timestamp is located is marked as a hyperglycemic abnormal interval. When the concentration value is lower than the lower limit, it is marked as a hypoglycemic abnormal interval. The concentration value is between the upper and lower limits and is marked as a normal interval. The abnormality mark and the corresponding timestamp are stored synchronously, and a graded warning information is triggered at the same time. Hyperglycemia and hypoglycemia correspond to different warning levels. The warning information includes the abnormality type, the occurrence timestamp, and the deviation range. When there is no abnormality, the warning information field is set to empty. The warning information and the abnormal interval mark maintain a strict temporal correspondence.

[0055] The calibration report is generated based on the integration and visualization of multi-dimensional time-series data. The integrated data types include glucose concentration value sequences arranged in chronological order, point-by-point matched anomaly markers, warning information at corresponding time nodes, and complete timestamp sequences. All data are aligned with a unified time axis to form a structured time-series data set. On this basis, the curve generation module is called to draw a continuous blood glucose change curve with timestamps as the horizontal axis and glucose concentration values ​​as the vertical axis. The curve is simultaneously marked with the upper and lower limits of the normal blood glucose range, the location of abnormal intervals, and warning trigger nodes. The report also includes a time-series concentration value list, anomaly statistics, and warning records. The entire report is packaged in a standardized format. The calibration report can be used for real-time display, historical review, clinical reference, and verification of equipment calibration effects, fully presenting the monitoring accuracy and data reliability of the sensor after body temperature compensation and aging calibration.

[0056] Please see Figure 2 , Figure 2 The graph shows a comparison of 72-hour glucose concentration monitoring results, illustrating the trends of actual blood glucose levels, values ​​monitored by the traditional static correction method, and values ​​monitored by this method over time. This demonstrates that the monitoring curve of this method closely matches the actual blood glucose levels, accurately tracking blood glucose changes even under scenarios involving combined body temperature fluctuations and sensor aging, while the traditional static correction method exhibits significant offsets and fluctuations. This application improves long-term monitoring accuracy through dynamic body temperature compensation and aging-adaptive calibration.

[0057] Please see Figure 3 , Figure 3 The box plot of the glucose concentration monitoring error distribution shows the differences in the monitoring error distribution range, dispersion, and median between the traditional static correction method and the proposed method. This demonstrates that the proposed method has smaller errors, a more concentrated distribution, and stronger stability, while the traditional static correction method exhibits large error fluctuations and high deviations. This application effectively suppresses temperature and aging coupling interference, reducing monitoring errors and improving data consistency and reliability.

[0058] Please see Figure 4 The following describes the continuous blood glucose monitoring system based on individual body temperature compensation and aging correction in the embodiments of this application. The continuous blood glucose monitoring system based on individual body temperature compensation and aging correction includes: The data synchronization module is used to acquire the raw current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, perform time alignment processing on the raw current signal sequence and the measured body temperature data sequence, and generate synchronized data pairs. The rate calculation module is used to calculate the rate of change of body temperature based on the measured body temperature data sequence in the synchronous data pair, input the rate of change of body temperature into the glucose diffusion temperature sensitivity model, and output the theoretical glucose permeation rate reference sequence. The signal filtering module is used to filter the original current signal sequence using an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as the reference input, to remove high-frequency noise caused by body temperature changes and generate a base signal sequence. The aging judgment module is used to calculate statistical characteristic values ​​based on the substrate signal sequence, compare the statistical characteristic values ​​with the benchmark characteristic values, determine the degree of aging of the sensor semi-permeable membrane based on the comparison results, and calculate the aging attenuation coefficient. The parameter adjustment module is used to obtain the mapping relationship between the aging of the sensor semipermeable membrane and the calibration parameters, determine the adjustment amount of the calibration parameters based on the aging attenuation coefficient and the mapping relationship, and generate calibration parameter values. The dynamic compensation module is used to fuse the calibration parameter values ​​with the real-time body temperature data to generate a dynamic compensation factor. Based on the dynamic compensation factor, the base signal sequence is corrected point by point to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment. The concentration conversion module is used to convert the corrected current signal sequence into glucose concentration values ​​and generate a calibration report based on the glucose concentration values.

[0059] Through the collaborative efforts of the aforementioned components, this system constructs an integrated system for dynamic body temperature compensation and adaptive calibration of semi-permeable membrane aging. This achieves high-precision, long-term, and automated calibration of glucose monitoring data under temperature and aging coupling interference. The data synchronization module aligns the original current signal with the measured body temperature data, providing a unified time reference for the entire calculation process and eliminating calculation errors caused by time deviations. The rate calculation module, based on the body temperature change rate and glucose diffusion temperature sensitivity model, outputs a theoretical permeation rate reference sequence, accurately quantifying the impact of dynamic body temperature changes on glucose diffusion. The signal filtering module performs adaptive filtering with the theoretical permeation rate as a reference input, accurately removing high-frequency noise related to body temperature and outputting a clean sensor substrate signal sequence. The aging judgment module compares the statistical characteristics of the substrate signal with the baseline characteristics to determine the degree of aging of the semipermeable membrane in real time and calculates the aging attenuation coefficient, achieving dynamic tracking of the aging state. The parameter adjustment module matches the aging attenuation coefficient with a preset mapping relationship and automatically generates calibration parameters adapted to the current aging state, replacing traditional manual blood sampling calibration. The dynamic compensation module integrates calibration parameters and real-time body temperature data to generate a dynamic compensation factor, correcting the substrate signal point by point and compensating for temperature sensitivity drift caused by aging. The concentration conversion module converts the corrected current signal into a standard glucose concentration value, completes abnormal marking and early warning, and outputs a complete calibration report and blood glucose change curve. The entire system is model-driven and algorithm-adaptive, fundamentally solving the problem of accuracy degradation in traditional CGM under scenarios of body temperature fluctuations and sensor aging, and significantly improving the accuracy, stability and user compliance of continuous blood glucose monitoring.

[0060] This application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the continuous blood glucose monitoring method based on individual body temperature compensation and aging correction.

[0061] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the methods and systems described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0062] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A continuous blood glucose monitoring method based on individual body temperature compensation and aging correction, characterized in that, Includes the following steps: Step S1: Obtain the original current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, and perform time alignment processing on the original current signal sequence and the measured body temperature data sequence to generate a synchronized data pair; Step S2: Calculate the body temperature change rate based on the measured body temperature data sequence in the synchronized data pair, input the body temperature change rate into the glucose diffusion temperature sensitivity model, and output the theoretical glucose permeation rate reference sequence. Step S3: Using an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as the reference input, the original current signal sequence is filtered to remove high-frequency noise caused by body temperature changes, and a base signal sequence is generated. Step S4: Calculate statistical characteristic values ​​based on the substrate signal sequence, compare the statistical characteristic values ​​with the benchmark characteristic values, distinguish between aging attenuation and physiological fluctuations based on the comparison results, determine the degree of aging of the sensor semipermeable membrane, and calculate the aging attenuation coefficient. Step S5: Obtain the mapping relationship between sensor semi-permeable membrane aging and calibration parameters, determine the calibration parameter adjustment amount based on the aging attenuation coefficient and the mapping relationship, and generate calibration parameter values; Step S6: The calibration parameter value and real-time body temperature data are fused and calculated to generate a dynamic compensation factor. The base signal sequence is corrected point by point according to the dynamic compensation factor to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment. Step S7: Convert the corrected current signal sequence into glucose concentration values, and generate a calibration report based on the glucose concentration values.

2. The method according to claim 1, characterized in that, Step S1 includes: The working electrode of the continuous glucose monitoring sensor is implanted under the skin to collect the raw current signal sequence during the monitoring period in real time. The raw current signal sequence is composed of microcurrents generated by the redox reaction of glucose molecules on the electrode surface. Synchronously acquire the measured body temperature data sequence with the timestamp corresponding to the original current signal sequence; Using the timestamp of the original current signal sequence as the main reference point, adjacent temperature sampling points are retrieved in the measured body temperature data sequence, and a linear interpolation algorithm is used to calculate the fitted temperature value. The signal values ​​in the original current signal sequence and the corresponding fitted temperature values ​​are encapsulated with timestamps to generate one-to-one synchronous data pairs.

3. The method according to claim 1, characterized in that, Step S2 includes: For the measured body temperature data sequence in the synchronous data pair, the ratio of the body temperature difference between adjacent time points to the time interval is calculated to obtain the body temperature change rate sequence. The body temperature change rate sequence is input into a pre-calibrated glucose diffusion temperature sensitivity model to calculate the effect of the theoretical temperature gradient at each time point on the glucose permeation rate. Based on the aforementioned influence values, a smooth theoretical glucose permeation rate reference sequence is generated using cubic spline interpolation.

4. The method according to claim 1, characterized in that, Step S3 includes: The theoretical glucose permeation rate reference sequence is used as the reference input signal for the least mean square adaptive filtering algorithm. Based on the reference input signal, the weight coefficient vector is iteratively adjusted point by point for the original current signal sequence to identify high-frequency noise components related to body temperature changes. The identified high-frequency noise components are subtracted point by point from the original current signal sequence to generate a substrate signal sequence characterizing the inherent permeability resistance of the sensor's semipermeable membrane.

5. The method according to claim 1, characterized in that, Step S4 includes: A sliding window of a preset time length is selected to traverse the base signal sequence, and the statistical characteristic values ​​of the signal within each window are calculated. The statistical characteristic values ​​include at least the signal mean and the signal variance. Determine whether the signal variance is within a preset variance range. If so, determine that there is no abnormal interference in the current window signal and use the signal mean of the window as the effective mean. The effective mean value is compared with the reference characteristic value during the initial calibration phase of the sensor to determine whether the signal strength of the base signal sequence has decreased. If not, then maintain the original aging degradation coefficient unchanged; If so, then further obtain the effective mean sequence of N consecutive windows and calculate the rate of change of signal strength, where N is an integer greater than 5; If the absolute value of the rate of change is less than a preset change threshold and the duration of the decrease exceeds a preset duration, it is determined that the decrease in signal strength is caused by the aging of the semipermeable membrane, and the aging attenuation coefficient is calculated based on the difference between the effective mean and the reference characteristic value. If the absolute value of the rate of change is greater than or equal to the preset change threshold and a rebound occurs in the subsequent window, it is determined that the decrease in signal intensity is caused by a physiological decrease in blood glucose concentration, and the original aging attenuation coefficient remains unchanged.

6. The method according to claim 5, characterized in that, Step S5 includes: Obtain a pre-calibrated mapping table between the aging degree of the sensor semi-permeable membrane and the initial calibration parameters. The mapping table records the sensitivity correction factor corresponding to different aging attenuation coefficient ranges. Using the aging attenuation coefficient as an index, retrieve the corresponding sensitivity correction factor; The sensitivity correction factor is multiplied by the initial calibration parameters to generate calibration parameter values ​​that correct for aging errors.

7. The method according to claim 6, characterized in that, Step S6 includes: Obtain the instantaneous body temperature value at the current moment as real-time body temperature data; Calculate the instantaneous change in the current temperature gradient based on real-time body temperature data; The calibration parameter value is used as the compensation benchmark. The instantaneous change in temperature gradient is multiplied by a preset temperature sensitivity coefficient to obtain a dynamic adjustment term. The compensation benchmark and the dynamic adjustment term are added to obtain a dynamic compensation factor. The preset temperature sensitivity coefficient is a pre-calibrated constant. The substrate signal sequence is corrected point by point according to the dynamic compensation factor to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, thereby generating a corrected current signal sequence.

8. The method according to claim 1, characterized in that, Step S7 includes: The corrected current signal sequence is substituted into the pre-calibrated formula to complete the electrochemical conversion and obtain the glucose concentration value. The glucose concentration value is compared with a preset normal range. If it exceeds the range, the abnormal range is marked and an early warning is triggered. Integrate glucose concentration value sequences, anomaly markers, warning information, and timestamps to generate a calibration report containing blood glucose change curves.

9. A continuous blood glucose monitoring system based on individual body temperature compensation and aging correction, used to implement the method described in any one of claims 1-8, characterized in that, include: The data synchronization module is used to acquire the raw current signal sequence and the measured body temperature data sequence through the continuous blood glucose monitoring sensor, perform time alignment processing on the raw current signal sequence and the measured body temperature data sequence, and generate a synchronized data pair. The rate calculation module is used to calculate the body temperature change rate based on the measured body temperature data sequence in the synchronous data pair, input the body temperature change rate into the glucose diffusion temperature sensitivity model, and output a theoretical glucose permeation rate reference sequence. The signal filtering module is used to use an adaptive filtering algorithm, with the theoretical glucose osmotic rate reference sequence as a reference input, to filter the original current signal sequence to remove high-frequency noise caused by body temperature changes and generate a base signal sequence. The aging judgment module is used to calculate statistical feature values ​​based on the substrate signal sequence, compare the statistical feature values ​​with the benchmark feature values, distinguish between aging attenuation and physiological fluctuations based on the comparison results, determine the degree of aging of the sensor semipermeable membrane, and calculate the aging attenuation coefficient. The parameter adjustment module is used to obtain the mapping relationship between the aging of the sensor semipermeable membrane and the calibration parameters, determine the adjustment amount of the calibration parameters according to the aging attenuation coefficient and the mapping relationship, and generate calibration parameter values. The dynamic compensation module is used to fuse the calibration parameter value with the real-time body temperature data to generate a dynamic compensation factor. Based on the dynamic compensation factor, the base signal sequence is corrected point by point to compensate for the temperature sensitivity drift caused by the aging of the sensor semi-permeable membrane, and a corrected current signal sequence is generated. The real-time body temperature data is the instantaneous body temperature value at the current moment. The concentration conversion module is used to convert the corrected current signal sequence into glucose concentration values ​​and generate a calibration report based on the glucose concentration values.

10. A computer-readable storage medium having instructions stored thereon, characterized in that, When the instruction is executed by the processor, it implements the continuous blood glucose monitoring method based on individual body temperature compensation and aging correction as described in any one of claims 1-8.

Images