An automatic test and error calibration system for blood glucose meters

Abstract

This invention relates to the field of blood glucose calibration technology, specifically to an automatic testing and error calibration system for blood glucose meters. It solves the technical problem in existing technologies where the lack of continuous, high-precision reference data makes it difficult to promptly and accurately identify and correct measurement errors in home blood glucose meters during long-term use. The system includes: a data acquisition module that acquires the original blood glucose measurement sequence and obtains a reference blood glucose value and its corresponding reference timestamp; a drift analysis module that determines the device drift estimate based on the fluctuation characteristics of the original blood glucose measurement sequence at different time scales; an anchor point fusion module that determines the associated blood glucose measurement data corresponding to the reference timestamp from the original blood glucose measurement sequence and corrects the device drift estimate based on the deviation between the associated blood glucose measurement data and the reference blood glucose value; and a calibration output module that corrects the measurement value of the home blood glucose meter based on the corrected device drift estimate and outputs the corrected blood glucose value.

Description

Technical Field

[0001] This invention relates to the field of blood glucose calibration technology, and more specifically to an automatic testing and error calibration system for blood glucose meters. Background Technology

[0002] Blood glucose testing is a key part of daily health management for diabetic patients. Currently, most home blood glucose meters are based on electrochemical methods, which calculate blood glucose values ​​by detecting the reaction current of blood samples on test strips. However, these devices are susceptible to the effects of batch differences in test strips, changes in electrode performance, equipment aging, and fluctuations in environmental and operating conditions during long-term use. The measurement results may gradually show systematic deviations, affecting the stability and consistency of the test.

[0003] Current home blood glucose meter calibration methods mainly rely on factory-calibrated parameters or periodic calibration using standard solutions. These calibrations are static and cannot adapt to the dynamic changes that occur during long-term use. Furthermore, some calibration methods require manual calibration by the user or the use of external high-precision equipment, which is complex and highly dependent. In actual home use scenarios, the device often cannot continuously obtain reliable reference data, resulting in insufficient practicality and timeliness of calibration.

[0004] In the absence of continuous, high-precision reference data, existing technologies struggle to effectively identify systematic errors in blood glucose meter measurement data introduced by changes in device performance. They also cannot accurately distinguish between the user's actual blood glucose changes and device measurement drift, easily misinterpreting physiological changes as device errors. Furthermore, given the low sampling frequency and incomplete time synchronization of blood glucose test data in hospital laboratories, it is difficult to reasonably utilize such low-frequency reference data for error correction, and it may also disrupt the continuity and stability of high-frequency measurement data from home blood glucose meters. Summary of the Invention

[0005] To address the technical problem in existing technologies where the lack of continuous high-precision reference data makes it difficult to promptly and accurately identify and correct measurement errors in home blood glucose meters during long-term use, the present invention aims to provide an automatic testing and error calibration system for blood glucose meters. The specific technical solution adopted is as follows: In a first aspect, an automatic testing and error calibration system for a blood glucose meter is provided, comprising: a data acquisition module for acquiring the original blood glucose measurement sequence from a home blood glucose meter, and acquiring a reference blood glucose value and a corresponding reference timestamp; a drift analysis module for determining a device drift estimate based on the fluctuation characteristics of the original blood glucose measurement sequence at different time scales; the device drift estimate is used to characterize the systematic measurement error of the home blood glucose meter; an anchor point fusion module for determining the associated blood glucose measurement data corresponding to the reference timestamp from the original blood glucose measurement sequence, and correcting the device drift estimate based on the deviation between the associated blood glucose measurement data and the reference blood glucose value; and a calibration output module for correcting the measurement value of the home blood glucose meter based on the corrected device drift estimate, and outputting the corrected blood glucose value.

[0006] In conjunction with the first aspect above, in one possible implementation, the drift analysis module is specifically used for: dividing the original blood glucose measurement sequence into time windows to obtain multiple analysis windows of different time lengths; smoothing the blood glucose measurement data within each analysis window to obtain a smoothed blood glucose sequence corresponding to each analysis window; using the smoothed blood glucose sequence to reduce short-term noise interference; determining the confidence level of each analysis window based on the consistency of changes in the smoothed blood glucose sequence; identifying stable windows with stable blood glucose changes from multiple analysis windows; and determining the device drift estimate based on the long-term cumulative offset trend of the smoothed blood glucose sequence corresponding to the stable window, combined with the confidence level.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the time window division adopts a sliding window approach, with different time lengths of analysis windows corresponding to different sliding step sizes to generate continuous smooth blood glucose sequences; the smooth blood glucose sequences corresponding to multiple analysis windows include: high-frequency data reflecting short-term physiological fluctuations, medium-frequency data reflecting medium-term trends, and low-frequency data reflecting long-term shifts.

[0008] In conjunction with the first aspect above, in one possible implementation, the drift analysis module is specifically used to: for each analysis window, analyze the data fluctuation of the smoothed blood glucose sequence and the offset of the smoothed blood glucose sequence relative to the mean of the smoothed blood glucose within the analysis window; obtain the confidence level of the analysis window based on the data fluctuation and the consistency between the offset and the historical offset trend; the greater the data fluctuation or the lower the consistency, the lower the confidence level.

[0009] In conjunction with the first aspect above, in one possible implementation, the drift analysis module is specifically used to: statistically analyze the changes between adjacent data points in the smoothed blood glucose sequence, and determine the analysis window corresponding to the continuous data segment with a change less than a preset change threshold as a stable window.

[0010] In conjunction with the first aspect above, in one possible implementation, the drift analysis module is specifically used for: determining the cumulative offset based on the smoothed blood glucose sequence for each stable window; weighting and fusing the cumulative offset according to the confidence level of the stable window to obtain a comprehensive offset; and determining the device drift estimate at the current moment based on the comprehensive offset and the device drift estimate at the previous moment through a preset recursive update model.

[0011] In conjunction with the first aspect above, in one possible implementation, the aforementioned anchor point fusion module is specifically used to: analyze the deviation between the associated blood glucose measurement data and the reference blood glucose value as the anchor point correction amount; and use the anchor point correction amount as input to adjust the output of the recursive update model to correct the device drift estimate.

[0012] In conjunction with the first aspect above, in one possible implementation, the calibration output module is further configured to: statistically analyze the rate of change of the corrected device drift estimate relative to the previous device drift estimate; if the rate of change is greater than a preset drift rate threshold, then adjust the corrected device drift estimate to the corresponding value of the preset drift rate threshold.

[0013] In conjunction with the first aspect above, in one possible implementation, the drift analysis module is specifically used to: perform mean filtering on the blood glucose measurement data within each analysis window to generate a smoothed blood glucose sequence corresponding to the analysis window.

[0014] In conjunction with the first aspect above, in one possible implementation, the data acquisition module is specifically used to: acquire blood glucose values ​​including timestamps measured by a home blood glucose meter to form an original blood glucose measurement sequence; and receive reference blood glucose values ​​and reference timestamps input by the user or synchronized from a medical data platform.

[0015] The present invention has the following beneficial effects: By acquiring the original blood glucose measurement sequence of a home blood glucose meter and the reference blood glucose value associated with the reference timestamp, and combining the fluctuation characteristics of the original blood glucose data at different time scales, the device drift estimate characterizing systematic measurement error is accurately determined. Then, the drift estimate is corrected by using the deviation between the associated blood glucose measurement data corresponding to the reference timestamp and the reference blood glucose value. Finally, the blood glucose meter measurement value is effectively calibrated, which can effectively distinguish between the user's physiological blood glucose fluctuations and device measurement drift, give full play to the anchoring role of low-frequency reference data, ensure the reliability and continuity of the calibration process, and significantly improve the measurement accuracy and stability of home blood glucose meters in long-term use scenarios. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a system structure diagram of an automatic blood glucose meter testing and error calibration system provided in one embodiment of the present invention; Figure 2 A flowchart illustrating an automatic blood glucose meter testing and error calibration method according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the hardware structure of an automatic blood glucose meter testing and error calibration device according to an embodiment of the present invention. Detailed Implementation

[0018] The specific solution of the automatic blood glucose meter testing and error calibration system provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0019] Please see Figure 1 The diagram illustrates a system structure of an automatic blood glucose meter testing and error calibration system according to an embodiment of the present invention. The automatic blood glucose meter testing and error calibration system includes: a data acquisition module 1, a drift analysis module 2, an anchor point fusion module 3, and a calibration output module 4.

[0020] Among them, the data acquisition module 1 is the basic source of data for the system, which includes two sub-modules: blood glucose measurement sub-module 11 and reference data receiving sub-module 12.

[0021] The blood glucose measurement submodule 11 is implemented through the test strip detection unit and the timestamp recording unit of the blood glucose meter. After the test strip detection unit collects the user's blood glucose value, the timestamp recording unit synchronously records the time information of each collection. The two are combined to form the original blood glucose measurement sequence. The reference data receiving submodule 12 can be implemented through the wired transmission interface of the blood glucose meter, the wireless communication module, or the medical data platform docking unit. It can receive the reference blood glucose value and reference timestamp manually entered by the user, and can also synchronize the above reference data from the associated medical data platform.

[0022] The raw blood glucose measurement sequence output by data acquisition module 1 is transmitted to drift analysis module 2, while the reference blood glucose value and reference timestamp are simultaneously sent to anchor fusion module 3, providing basic data support for subsequent error analysis and drift correction.

[0023] The drift analysis module 2 receives the original blood glucose measurement sequence from the data acquisition module 1 and is the core module for realizing device drift estimation. It includes four sub-modules: window partitioning and smoothing sub-module 21, confidence calculation sub-module 22, stable window identification sub-module 23, and drift estimation sub-module 24.

[0024] The window division and smoothing submodule 21 is implemented by the processor of the blood glucose meter using a preset algorithm. First, the original blood glucose measurement sequence is divided into analysis windows of different time lengths according to the time sequence to obtain multiple analysis windows. Then, the mean filtering operation is performed on the blood glucose measurement data in each window to generate a smooth blood glucose sequence, thereby reducing short-term noise interference. At the same time, it separates high-frequency data that reflects short-term physiological fluctuations, medium-frequency data that reflects medium-term trends, and low-frequency data that reflects long-term deviations. The confidence calculation submodule 22 is implemented by the data processing chip. For each analysis window, it analyzes the degree of data fluctuation of the smoothed blood glucose sequence and the consistency between the offset of the smoothed blood glucose sequence relative to the mean within the window and the historical offset trend. Based on these two indicators, the confidence of each analysis window is determined. The confidence is used to evaluate the reliability of the data within the window. The stable window identification submodule 23 is implemented by the processor running a filtering algorithm, which statistically analyzes the changes between adjacent data points in the smoothed blood glucose sequence, and determines the analysis window corresponding to the continuous data segment with a change less than a preset change threshold as a stable window. The drift estimation submodule 24 is implemented by the algorithm execution unit. First, it calculates the cumulative offset based on the smooth blood glucose sequence corresponding to the stable window. Then, it weights and fuses the cumulative offset according to the confidence level to obtain the comprehensive offset. Finally, it combines the comprehensive offset with the device drift estimate of the previous moment and determines the device drift estimate of the current moment through a preset recursive update model (this estimate is used to characterize the systematic measurement error of the home blood glucose meter).

[0025] The device drift estimate, stable window data, and confidence scores of each window output by drift analysis module 2 will be transmitted to anchor point fusion module 3 to provide data support for accurate correction of drift estimate.

[0026] Anchor point fusion module 3 is a key module connecting reference data and drift estimation, and it includes three sub-modules: correlation data extraction sub-module 31, anchor point correction quantum module 32, and drift correction sub-module 33.

[0027] The associated data extraction submodule 31 is implemented by the data retrieval unit. After receiving the reference timestamp transmitted by the data acquisition module 1, it extracts the associated blood glucose measurement data corresponding to the reference timestamp from the stable window data output by the drift analysis module 2. The anchor point correction quantum module 32 is implemented by a computing chip, which calculates the deviation between the associated blood glucose measurement data and the reference blood glucose value transmitted by the data acquisition module 1, and uses the deviation as the anchor point correction amount. The drift correction submodule 33 is implemented by the processor. It takes the anchor point correction amount as input, adjusts the output parameters of the recursive update model used in the drift analysis module 2, corrects the device drift estimate transmitted by the drift analysis module 2, and obtains the corrected device drift estimate.

[0028] The corrected device drift estimate output by anchor point fusion module 3 is transmitted to calibration output module 4, providing the core basis for the final blood glucose value calibration.

[0029] The calibration output module 4 is the system's result output module, which contains three sub-modules: drift rate verification sub-module 41, blood glucose calibration sub-module 42, and result output sub-module 43.

[0030] The drift rate verification submodule 41 is implemented by the processor. It receives the corrected device drift estimate transmitted by the anchor point fusion module 3, and counts the rate of change of the estimate relative to the previous device drift estimate. If the rate of change is greater than the preset drift rate threshold, the corrected device drift estimate is adjusted to the value corresponding to the preset drift rate threshold to avoid misjudging physiological changes as device drift. The blood glucose calibration submodule 42 is implemented by the calculation unit. It calls the smoothed blood glucose sequence output by the drift analysis module 2, combines it with the corrected device drift estimate after rate verification, and performs calibration calculation to eliminate the influence of systematic measurement errors on the measurement results. The result output submodule 43 is implemented by the blood glucose meter's display screen, built-in storage chip, and mobile terminal communication module (such as Bluetooth module). It stores the calibrated blood glucose value with the corresponding timestamp, displays the calibration result in real time on the display screen, and can also transmit the calibrated blood glucose value to the associated mobile terminal or medical data platform as needed for user health management analysis.

[0031] Please see Figure 2 The diagram illustrates a flowchart of an automatic blood glucose meter testing and error calibration method performed by an automatic blood glucose meter testing and error calibration system according to an embodiment of the present invention. The automatic blood glucose meter testing and error calibration method includes: S1. Obtain the original blood glucose measurement sequence from the home blood glucose meter, and obtain the reference blood glucose value and the corresponding reference timestamp.

[0032] In some implementations, the system acquires blood glucose values, including timestamps, measured by a home blood glucose meter to form an original blood glucose measurement sequence; and receives reference blood glucose values ​​and reference timestamps input by the user or synchronized from a medical data platform.

[0033] Specifically, when a user uses a home blood glucose meter, the blood glucose level is collected by the meter's detection unit through the reaction between the test strip and the blood sample. At the same time, the built-in time recording module of the blood glucose meter automatically records the time information corresponding to each sampling, i.e., the sampling timestamp. These blood glucose values ​​containing sampling timestamps are automatically integrated in the order of sampling to form the original blood glucose measurement sequence, ensuring that each blood glucose data corresponds to a clear sampling time, providing a clear time basis for distinguishing blood glucose changes and device drift at different time dimensions.

[0034] Reference blood glucose values ​​and corresponding reference timestamps can be obtained in two ways: one is user input. After a user completes a physical examination at a hospital and obtains blood glucose data from the hospital laboratory, they can manually enter the hospital's blood glucose data as a reference value through the user interface (such as a touchscreen or physical button) of the home blood glucose meter. At the same time, they can also enter the time of the blood glucose test at the hospital as a reference timestamp. This adapts to user scenarios where the data is not bound to a medical data platform, ensuring the feasibility of manually entering reference data. The other way is synchronization from a medical data platform. The home blood glucose meter can connect to the user's pre-bound medical and health platform account via Bluetooth or other wireless means. At this time, a preset identity matching algorithm will be used, such as matching the user's account with the bound hospital visit information, to synchronize the user's hospital physical examination blood glucose data stored in the platform as a reference blood glucose value, and simultaneously synchronize the corresponding hospital test time as a reference timestamp. A preset timeout threshold will be set during the synchronization process, such as 30 seconds. If the data synchronization is not completed within 30 seconds, the blood glucose meter will issue a prompt message, realizing automatic synchronization of reference data, reducing the tediousness of manual operation for users, and ensuring a smooth experience during the synchronization process through the timeout threshold.

[0035] S2. Determine the device drift estimate based on the fluctuation characteristics of the original blood glucose measurement sequence at different time scales.

[0036] Device drift estimates are used to characterize the systematic measurement error of home blood glucose meters.

[0037] In one possible implementation, the method for determining the device drift estimate can be specifically implemented through the following steps S21 to S25, which are explained in detail below: S21. Divide the original blood glucose measurement sequence into time windows to obtain multiple analysis windows of different time lengths.

[0038] In some implementations, after acquiring and sorting the original blood glucose measurement sequences by timestamp, the next step is to perform a sliding time window division operation on the original blood glucose measurement sequences. The sorted original blood glucose measurement sequences do not need to meet the requirement of uniform sampling intervals; they only need to be concatenated in chronological order of timestamps.

[0039] During the segmentation, different time lengths are set as window sizes, and the window slides along the time axis with a certain step size to cover the entire time range of the original sequence, resulting in multiple analysis windows with overlapping regions: for example, a short-term window with a length of 2-3 days and a sliding step size of 1 hour is set to capture rapid changes in blood glucose within a short period; a medium-term window with a length of 5-7 days and a sliding step size of 1 day is set to reflect blood glucose change trends over several hours to several days; and a long-term window with a length of 15-30 days and a sliding step size of 3-5 days is set to observe overall data shift over a longer period. Blood glucose data within each window are included in the calculation according to timestamps, and each slide generates a smoothed blood glucose value within a window, thus forming a continuous smoothed blood glucose sequence. This method can provide trend curves at multiple time scales for subsequent analysis, making it easier to distinguish between user physiological fluctuations and long-term device drift.

[0040] S22. Smooth the blood glucose measurement data in each analysis window to obtain the smoothed blood glucose sequence corresponding to each analysis window.

[0041] Smoothing blood glucose sequences is used to reduce short-term noise interference.

[0042] In some implementations, after dividing the time window, smoothing is performed on the blood glucose measurement data within each analysis window. Specifically, a mean filtering method is used: taking a short-term analysis window as an example, if the window contains 5 blood glucose measurement data (the sampling size is not fixed), the arithmetic mean of these 5 data is calculated, and this mean is used as the smoothed blood glucose data for that window. After processing each analysis window in this way, the smoothed blood glucose data from each window are integrated in chronological order to obtain the smoothed blood glucose sequence corresponding to each analysis window. This step can reduce short-term noise interference, such as instantaneous errors caused by hand tremors, abnormal blood collection operations, etc., which are canceled out by the mean calculation, making the data closer to the actual blood glucose changes or device offset.

[0043] Meanwhile, the smoothed blood glucose sequences corresponding to different analysis windows of varying lengths represent different scales of change: the smoothed sequences for short-term windows reflect high-frequency data reflecting short-term physiological blood glucose fluctuations, the smoothed sequences for medium-term windows reflect mid-frequency data reflecting blood glucose trends over several hours to several days, and the smoothed sequences for long-term windows reflect low-frequency data reflecting long-term systematic device offsets. This step filters out transient interference data, allowing sequences at different scales to correspond to physiological changes and device offsets respectively, providing cleaner basic data for subsequent error differentiation and analysis.

[0044] S23. For each analysis window, determine the confidence level of the analysis window based on the consistency of changes in the smoothed blood glucose sequence.

[0045] In some implementations, for each analysis window, the data volatility of the smoothed blood glucose sequence and the offset of the smoothed blood glucose sequence relative to the mean smoothed blood glucose within the analysis window are analyzed. The confidence level of the analysis window is obtained based on the data volatility and the consistency between the offset and the historical offset trend. The greater the data volatility or the lower the consistency, the lower the confidence level.

[0046] Specifically, the first step is to analyze the volatility of the smoothed blood glucose series: calculate the variance of the smoothed blood glucose series within the current window (for example, if there are 7 data points in a medium-term window, calculate the average of the sum of squared deviations of these 7 data points from the window mean). The larger the variance, the more drastic the data volatility within the window, and the lower the data reliability. Next, calculate the offset of the smoothed blood glucose series relative to the mean of the smoothed blood glucose series within the window: subtract the mean of the smoothed blood glucose series in historical windows of the same length from the mean of the smoothed blood glucose series in the current window to obtain the offset value for that window; simultaneously, retrieve the offset values ​​of historical analysis windows of the same length and calculate the average of these historical offset values ​​as a reference for historical offset trends.

[0047] The confidence level is then calculated using the following formula: In the formula, W is the length of the current window; The variance of the smoothed blood glucose sequence within the window characterizes the degree of data fluctuation. Set a preset benchmark standard deviation parameter (e.g., 0.5 (mmol / L), which is the typical variance of normal blood glucose fluctuations). This is the offset value of the current window, representing the degree of deviation of the current window; The average offset value of historical windows of the same length represents the historical offset trend; exp is the natural exponential function used to map the input to positive real numbers. In the item, The larger the value (the more drastic the data fluctuation), the smaller the value of this item, resulting in a negative penalty for the confidence level; In the item, and The greater the difference (the lower the consistency), the smaller the value of this item, which negatively penalizes the confidence score. The arithmetic mean of the two products is taken to comprehensively reflect the reliability of the window data, thus obtaining the confidence score. The larger the summation result, the smaller the data fluctuation within the window and the higher the consistency with historical windows. The more reliable the data, the higher the confidence level.

[0048] The core functions of the natural exponential function are: first, to achieve non-negativity mapping, transforming the negative penalty term reflecting data fluctuations and deviation consistency in the formula into a positive real number, ensuring that the confidence level is always non-negative, consistent with its physical meaning of representing data reliability; second, to construct a non-linear penalty mechanism, where the more drastic the data fluctuations and the lower the consistency with historical deviation trends, the corresponding exponential term decreases non-linearly and rapidly, amplifying the impact of bad data and significantly reducing the confidence level, accurately realizing the screening logic of "the more unreliable the data, the lower the confidence level"; and third, to ensure numerical smoothness, its continuous differentiability avoids sudden changes in confidence level due to small changes in input parameters, making the results more consistent with the actual scenario of continuous fluctuations in blood glucose data, and stably supporting subsequent equipment error correction analysis.

[0049] S24. Identify stable windows with steady blood glucose changes from multiple analysis windows.

[0050] In some implementations, the change between adjacent data points in the blood glucose sequence is statistically smoothed, and the analysis window corresponding to the continuous data segment with a change less than a preset change threshold is determined as a stable window.

[0051] Specifically, after generating and calculating the confidence scores for the smoothed blood glucose sequences in each analysis window, the next step is to identify stable windows where blood glucose changes are relatively stable. First, for the smoothed blood glucose sequence within each analysis window, the change between adjacent data points is statistically analyzed. Specifically, the absolute value of the difference between the next smoothed blood glucose data point and the previous smoothed blood glucose data point is calculated to quantify the degree of blood glucose fluctuation at adjacent time points. Then, a preset change threshold is set, for example, 0.5 mmol / L (this threshold is based on the normal physiological blood glucose fluctuation range). Analysis windows corresponding to continuous data segments with changes less than this threshold are initially identified as candidate stable windows.

[0052] Simultaneously, considering the specific characteristics of blood glucose data within a given timeframe, the system further filters data from the candidate stable window to the period between 3:00 AM and 5:00 AM daily. During this period, users are in a fasting state and are minimally affected by factors such as diet and exercise; therefore, the long-term trend of their blood glucose changes more closely reflects the device's own drift rather than fluctuations caused by physiological factors. If no measurement data is available within the target timeframe, the time window is expanded (e.g., 2:00 AM to 6:00 AM) or the fasting measurement data closest to the target timeframe is selected as a substitute.

[0053] The effect of calculating adjacent changes is to accurately capture the instantaneous fluctuation of blood glucose. The effect of setting a threshold is to filter out non-stationary data caused by instantaneous noise or sudden physiological fluctuations. The effect of filtering data in the early morning period is to further eliminate interference from non-equipment factors, so that the final stable window is dominated by equipment drift, providing a cleaner basic data for subsequent equipment error analysis.

[0054] S25. Based on the long-term cumulative shift trend of the smoothed blood glucose sequence corresponding to the stable window, and in combination with the confidence level, determine the device drift estimate.

[0055] In some implementations, for each stable window, the cumulative offset is determined based on the smoothed blood glucose sequence; the cumulative offset is weighted and fused based on the confidence level of the stable window to obtain the comprehensive offset; and the device drift estimate at the current time is determined based on the comprehensive offset and the device drift estimate at the previous time step through a preset recursive update model.

[0056] Specifically, after identifying the stable window, the device drift estimate is determined based on the long-term cumulative shift trend of the smoothed blood glucose sequence corresponding to the stable window, combined with the confidence level. First, for each stable window, the fasting baseline data sequence from 3:00 AM to 5:00 AM each day is extracted, and the linear regression slope of the fasting baseline data sequence is calculated. The product of the slope and the time span of the stable window is used as the cumulative shift corresponding to that stable window. Focusing on fasting data can minimize the interference of physiological factors such as diet and exercise, making the cumulative shift closer to the true drift trend of the device.

[0057] Subsequently, the cumulative offsets of each stable window are weighted and fused according to the confidence level of each stable window: the confidence level of each stable window is used as the weight of the cumulative offset of that window (the higher the confidence level, the greater the weight ratio), and the comprehensive offset is calculated by the weighted average method. The sum of the weights of all stable windows is 1, so that the impact of the window data with higher reliability on the comprehensive offset is more significant, and the interference of low confidence data on the results is avoided.

[0058] Then, the estimated device drift value at the current moment is determined using a pre-defined recursive update model: In the formula, is the estimated device drift value at time k (in mmol / L); a is a dimensionless update coefficient (e.g., 0.2, to match the slow change characteristics of device drift). The device drift estimate (in mmol / L) for the previous time step k is the total systematic bias of the device estimated based on historical data. The total offset represents the total drift re-estimated based on observations from multiple current stable windows (in mmol / L). and The units are consistent, both being absolute values ​​of blood glucose concentration deviations, thus allowing for smooth updates through weighted averaging; the product of (1-a) and the previous time-series estimate preserves error continuity, while the product of a and the overall offset introduces the current offset data. Both are positively correlated with the result, resulting in... The numerical representation of systematic equipment error takes into account both error continuity and the timeliness of new data, making the estimate more realistic and used to quantify long-term equipment drift.

[0059] During system initialization, the device drift estimate is set to the factory calibration parameter (usually 0) by default to ensure that the analysis starts from the initial state of the device.

[0060] S3. Determine the associated blood glucose measurement data corresponding to the reference timestamp from the original blood glucose measurement sequence, and correct the device drift estimate based on the deviation between the associated blood glucose measurement data and the reference blood glucose value.

[0061] In some implementations, the deviation between the correlated blood glucose measurement data and the reference blood glucose value is analyzed and used as the anchor correction amount; the anchor correction amount is used as input to adjust the output of the recursive update model and correct the device drift estimate.

[0062] Specifically, the first step is to determine the associated blood glucose measurement data from the original blood glucose measurement sequence: extract a stable, smoothed blood glucose sequence within a 2-3 day window around the reference time stamp, and calculate the mean of this sequence as the associated data. The 2-3 day window is chosen to cover the stable interval before and after the reference time stamp, avoiding random interference from single data points.

[0063] Then, the deviation between the associated blood glucose measurement data and the reference blood glucose value (i.e., the difference between the associated blood glucose measurement data and the reference blood glucose value) is calculated and used as the anchor correction amount. This directly reflects the direction and magnitude of the difference between the home measurement value and the accurate value in the hospital. It is the key anchor point for calibrating the device error. Moreover, the hospital data is not directly substituted here, but is introduced in the form of deviation to avoid disrupting the continuity of the home data.

[0064] Next, a hierarchical drift estimation model is used to correct the equipment drift estimate. This model decomposes equipment drift into two parts: Part One This is the long-term benchmark drift established and maintained by the anchor point correction. When the anchor point correction is obtained, the anchor point correction... The drift is used as the updated long-term baseline; otherwise, it remains unchanged. Part Two It uses a recursive model to update short-term dynamic drift in real time, reflecting short-term fluctuations in the equipment: In the formula, the system calculates at time k-1. Then, it is stored in memory / non-volatile memory, and read from memory when it is necessary to calculate the short-term dynamic drift at time k; When the anchor point correction amount is obtained At that time, in addition to updating the long-term baseline drift, the short-term dynamic drift is also reset to zero, i.e. =0. This means that the system restarts accumulating short-term dynamic drift from the new calibration baseline.

[0065] This hierarchical model ensures that the calibration benchmark established by the anchor point correction is permanently retained (until new reference data arrives), while short-term recursive drift is dynamically adjusted on the benchmark, achieving both long-term stable calibration and adapting to short-term changes in the equipment.

[0066] S4. Correct the measured value of the home blood glucose meter based on the corrected device drift estimate, and output the corrected blood glucose value.

[0067] In some implementations, after obtaining the corrected device drift estimate, a calibration operation is performed on the home blood glucose meter's measurement value. First, the smoothed blood glucose sequence corresponding to the stable window is called, and the corrected device drift estimate is subtracted from the smoothed blood glucose data in the sequence to obtain the calibrated blood glucose value. This eliminates the influence of systematic device drift and makes the output value closer to the user's actual blood glucose level.

[0068] In some implementations, to avoid misjudging physiological changes as equipment drift, a rate check needs to be performed on the corrected estimate. This method also includes: statistically analyzing the rate of change of the corrected equipment drift estimate relative to the previous equipment drift estimate (e.g., calculating the monthly rate of change based on the system operating frequency (e.g., once a day), i.e., multiplying the daily rate of change by the total number of days in the month to convert it into a monthly rate of change), setting a preset drift rate threshold (e.g., 0.1 mmol / L / month, ensuring that the unit time for threshold comparison is the same and matching the normal aging rate of the equipment); if the rate of change is greater than the preset drift rate threshold, the corrected equipment drift estimate is adjusted to the corresponding value of the preset drift rate threshold to limit the magnitude of error estimation, prevent calibration anomalies caused by sudden physiological fluctuations, and ensure operational safety.

[0069] Finally, the corrected blood glucose value is output, completing the automatic error correction process of the blood glucose meter. This value is then stored in association with the sampling timestamp and displayed in real time via the blood glucose meter's display module. It can also be transmitted to mobile terminals or medical platforms.

[0070] Based on the above technical solution, by acquiring the original blood glucose measurement sequence of the home blood glucose meter and the reference blood glucose value associated with the reference timestamp, and combining the fluctuation characteristics of the original blood glucose data at different time scales, the device drift estimate characterizing the systematic measurement error is accurately determined. Then, the drift estimate is corrected by using the deviation between the associated blood glucose measurement data corresponding to the reference timestamp and the reference blood glucose value. Finally, the blood glucose meter measurement value is effectively calibrated, which can effectively distinguish between the user's physiological blood glucose fluctuations and the device measurement drift, give full play to the anchoring role of low-frequency reference data, ensure the reliability and continuity of the calibration process, and significantly improve the measurement accuracy and stability of the home blood glucose meter in long-term use scenarios.

[0071] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0072] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0073] In this embodiment of the invention, the automatic blood glucose meter testing and error calibration device can be divided into functional units according to the above method example. For example, each function can be divided into its own functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0074] This invention also provides a hardware structure diagram of an automatic blood glucose meter testing and error calibration device, see below. Figure 3 The automatic blood glucose meter testing and error calibration device 300 includes a processor 301, and optionally, a memory 302 connected to the processor 301.

[0075] In the first possible implementation, see Figure 3The automatic blood glucose meter testing and error calibration device 300 also includes a transceiver 303. The processor 301, memory 302, and transceiver 303 are connected via a bus. The transceiver 303 is used to communicate with other devices or communication networks. Optionally, the transceiver 303 may include a transmitter and a receiver. The device in the transceiver 303 that implements the receiving function can be considered as a receiver, which is used to perform the receiving steps in the embodiments of the present invention. The device in the transceiver 303 that implements the transmitting function can be considered as a transmitter, which is used to perform the transmitting steps in the embodiments of the present invention.

[0076] Based on the first possible implementation method Figure 3 The schematic diagram shown can be used to illustrate the structure of the automatic blood glucose meter testing and error calibration device involved in the above embodiments.

[0077] in, Figure 3 The diagram can also illustrate the system chip in the automatic testing and error calibration device of the blood glucose meter. In this case, the actions performed by the aforementioned automatic testing and error calibration device can be implemented by this system chip. The specific actions performed can be found above and will not be repeated here.

Claims

1. An automatic blood glucose meter testing and error calibration system, characterized in that, include: The data acquisition module is used to acquire the raw blood glucose measurement sequence from the home blood glucose meter, and to obtain the reference blood glucose value and the corresponding reference timestamp; The drift analysis module is used to determine the device drift estimate based on the fluctuation characteristics of the original blood glucose measurement sequence at different time scales. The device drift estimate is used to characterize the systematic measurement error of the home blood glucose meter; An anchor point fusion module is used to determine the associated blood glucose measurement data corresponding to the reference timestamp from the original blood glucose measurement sequence, and to correct the device drift estimate based on the deviation between the associated blood glucose measurement data and the reference blood glucose value. The calibration output module is used to correct the measured value of the home blood glucose meter based on the corrected device drift estimate and output the corrected blood glucose value. The drift analysis module is specifically used for: The original blood glucose measurement sequence is divided into time windows to obtain multiple analysis windows of different time lengths; The blood glucose measurement data within each analysis window are smoothed to obtain a smoothed blood glucose sequence corresponding to each analysis window; the smoothed blood glucose sequence is used to reduce short-term noise interference. For each analysis window, the confidence level of the analysis window is determined based on the consistency of the changes in the smoothed blood glucose sequence; Identify stable windows with steady blood glucose changes from multiple analysis windows; The device drift estimate is determined based on the long-term cumulative shift trend of the smoothed blood glucose sequence corresponding to the stable window, combined with the confidence level.

2. The automatic blood glucose meter testing and error calibration system according to claim 1, characterized in that, The time window is divided using a sliding window method, with different time lengths of analysis windows corresponding to different sliding step sizes to generate continuous smooth blood glucose sequences. The smooth blood glucose sequences corresponding to multiple analysis windows include: high-frequency data reflecting short-term physiological fluctuations, medium-frequency data reflecting medium-term trends, and low-frequency data reflecting long-term shifts.

3. The automatic blood glucose meter testing and error calibration system according to claim 1, characterized in that, The drift analysis module is specifically used for: For each analysis window, the degree of data fluctuation of the smoothed blood glucose sequence and the offset of the smoothed blood glucose sequence relative to the mean smoothed blood glucose within the analysis window are analyzed. The confidence level of the analysis window is obtained based on the degree of data fluctuation and the degree of consistency between the offset and the historical offset trend; the greater the degree of data fluctuation or the lower the degree of consistency, the lower the confidence level.

4. The automatic blood glucose meter testing and error calibration system according to claim 1, characterized in that, The drift analysis module is specifically used for: The changes between adjacent data points in the smoothed blood glucose sequence are statistically analyzed, and the analysis window corresponding to the continuous data segment with a change less than a preset change threshold is determined as the stable window.

5. The automatic blood glucose meter testing and error calibration system according to claim 1, characterized in that, The drift analysis module is specifically used for: For each stable window, the cumulative offset is determined based on the smoothed blood glucose sequence; Based on the confidence level of the stable window, the cumulative offset is weighted and fused to obtain the comprehensive offset; Based on the combined offset and the device drift estimate from the previous moment, the device drift estimate for the current moment is determined using a preset recursive update model.

6. The automatic blood glucose meter testing and error calibration system according to claim 5, characterized in that, The anchor point fusion module is specifically used for: The deviation between the correlated blood glucose measurement data and the reference blood glucose value is analyzed and used as the anchor point correction amount; Using the anchor point correction as input, the output of the recursive update model is adjusted to correct the device drift estimate.

7. The automatic blood glucose meter testing and error calibration system according to claim 6, characterized in that, The calibration output module is also used for: The rate of change of the statistically corrected equipment drift estimate relative to the equipment drift estimate at the previous time step; If the rate of change is greater than a preset drift rate threshold, the corrected device drift estimate is adjusted to the corresponding value of the preset drift rate threshold.

8. The automatic blood glucose meter testing and error calibration system according to claim 1, characterized in that, The drift analysis module is specifically used for: Perform mean filtering on the blood glucose measurement data within each analysis window to generate a smoothed blood glucose sequence corresponding to the analysis window.

9. The automatic blood glucose meter testing and error calibration system according to any one of claims 1-8, characterized in that, The data acquisition module is specifically used for: The blood glucose values, including timestamps, measured by the home blood glucose meter are obtained to form the original blood glucose measurement sequence; Receive the reference blood glucose value and the reference timestamp input by the user or synchronized from the medical data platform.

Images