Automatic calibration system and automatic calibration method of continuous glucose monitoring system

Abstract

The present disclosure describes an automatic calibration system and an automatic calibration method for a continuous glucose monitoring system. The automatic calibration system comprises a glucometer, a detection unit and a parameter calibration unit, the glucometer is used for measuring the second concentration, the detection unit detects whether the continuous glucose monitoring system has an abnormal condition or not, and if the detection unit detects that the abnormal condition exists, the parameter calibration unit is started; the parameter calibration unit obtains a loss function related to the target parameter based on the second concentration, the sensing signal and a reference value of the target parameter, and calibrates the target parameter by optimizing the loss function. In this case, the target parameters can be calibrated based on reasonable opportunities, so that the accuracy of the continuous glucose monitoring system can be improved, and the usability can be improved.

Description

Technical Field

[0001] The present disclosure relates generally to the field of medical devices, and more particularly to an automatic calibration system and method for a continuous glucose monitoring system. Background Art

[0002] Continuous Glucose Monitoring (CGM) system is a type of system used to monitor glucose levels in the human body. This type of system can continuously monitor the glucose level in the human body over a certain period of time, and monitor and alarm abnormal glucose levels in the human body. For diabetic patients, using a continuous glucose monitoring system to monitor glucose levels in the body can better track changes in glucose levels in the body.

[0003] During use, the parameters of the continuous glucose monitoring system (such as delay coefficient, sensitivity coefficient and deviation coefficient) may change due to various factors, resulting in inaccurate glucose readings. At this time, the continuous glucose monitoring system needs to be calibrated. At present, the commonly used continuous glucose calibration method is the finger blood calibration method, that is, the user uses a fingertip blood glucose meter to obtain blood glucose readings and inputs the blood glucose readings into the continuous glucose monitoring system. The continuous glucose monitoring system calibrates the parameters based on the blood glucose readings in real time.

[0004] However, users of continuous glucose monitoring systems generally lack relevant knowledge of glucose monitoring systems and find it difficult to determine the appropriate time to calibrate the continuous glucose monitoring system. If the parameters are calibrated immediately based on blood glucose readings before the continuous glucose monitoring system has a deviation that affects the use of the system, the calibrated continuous glucose monitoring system may instead make the readings of the glucose concentration monitored by the calibrated continuous glucose monitoring system even more inaccurate. Summary of the invention

[0005] The present disclosure is proposed in view of the above-mentioned prior art conditions, and its purpose is to provide an automatic calibration system and automatic calibration method for a continuous glucose monitoring system that can automatically determine the calibration timing and thereby improve the accuracy of the glucose concentration readings of the continuous glucose monitoring system.

[0006] To this end, the first aspect of the present disclosure provides an automatic calibration system for a continuous glucose monitoring system, the automatic calibration system is used to calibrate a target parameter of the continuous glucose monitoring system, the target parameter is used to convert a sensor signal of a target object obtained by the continuous glucose monitoring system into a corresponding glucose concentration as a first concentration; the automatic calibration system includes a blood glucose meter, a detection unit and a parameter calibration unit, the blood glucose meter is used to measure a second concentration of the target object, the second concentration is the glucose concentration of the target object's blood, the detection unit detects whether an abnormal situation occurs in the continuous glucose monitoring system based on the second concentration of the detection point and the first concentration of the plurality of detection points, the detection point is a time point for obtaining the second concentration, the abnormal situation includes that there are not less than a preset number of the detection points among the plurality of detection points that meet a preset deviation condition, the preset deviation condition includes that the difference between the first concentration and the second concentration of the detection point is not less than a first preset value or not greater than a second preset value, and if the detection unit detects that the abnormal situation exists, the parameter calibration unit obtains a loss function related to the target parameter based on the second concentrations of the plurality of detection points, the sensor signals of the plurality of detection points and the reference value of the target parameter, and calibrates the target parameter by optimizing the loss function.

[0007] In the first aspect of the present disclosure, the target parameter is calibrated based on abnormal situation detection. In this case, the target parameter can be calibrated based on a reasonable timing, thereby improving the accuracy and availability of the continuous glucose monitoring system. At the same time, because the first concentration will be continuously greater than or continuously less than the second concentration when a system deviation occurs, therefore, when an abnormal situation is detected, the system deviation can be effectively measured and the continuous glucose monitoring system can be calibrated based on the size relationship of the difference between the first concentration and the second concentration, and the influence of random deviation on the calibration timing can be reduced. In addition, the abnormal situation detection can be performed automatically, thereby realizing automatic calibration of the continuous glucose monitoring system. In addition, optimizing the loss function based on data from multiple detection points can keep the loss function at a small value during the optimization process, reducing the influence of the error of the blood glucose meter.

[0008] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the detection points of not less than the preset number are arranged in the plurality of detection points in a front-to-back manner. In this case, since the system deviation is biased, detection based on the front-to-back adjacent detection points can more accurately determine whether the system deviation occurs, and thus can determine a more accurate calibration timing.

[0009] In addition, in the calibration method involved in the first aspect of the present disclosure, optionally, a data receiving and storage unit is further included, and the data receiving and storage unit is used to receive and store the second concentration of the multiple detection points, the first concentration of the multiple detection points, and the sensor signals of the multiple detection points. In this case, the data of multiple detection points can be stored, so that the detection unit can conveniently use and operate the stored data, and then can judge whether there is an abnormality based on the relationship between the first concentration and the second concentration obtained in the previous period of time.

[0010] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the data receiving and storing unit is also used to detect whether there are abnormal data points in the second concentrations of the multiple detection points, and the abnormal data points include abnormal second concentrations. If the abnormal data points are detected, the abnormal data points are removed from the second concentrations of the multiple detection points. In this case, removing the abnormal data points can reduce the impact of the erroneous second concentration.

[0011] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the first preset value is not less than 0, and the second preset value is the opposite of the first preset value. In this case, the first preset value can be used for the case where the first concentration is greater than the second concentration, and the second preset value can be used for the case where the first concentration is less than the second concentration, and the sensitivity of detecting abnormal conditions can be adjusted by the size of the first preset value and the second preset value.

[0012] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the iterative formula for optimizing the loss function is:

[0013]

[0014] Among them, p k+1 represents the target value of the target parameter obtained in the k+1th iteration, p k represents the target value of the target parameter obtained at the kth iteration, represents the gradient of the loss function with respect to the target parameter, f represents the loss function, H represents the Hesser matrix, H -1 represents the inverse matrix of H, and k represents the index of the number of iterations. In this case, by iterating and optimizing the target parameters using the iterative formula, the target parameters that meet the conditions defined by the loss function can be obtained for calibration. In addition, based on the iterative formula of the Heather matrix, the convergence speed can reach the second order, the convergence is fast, and the optimization efficiency is high, which is conducive to the deployment of the optimization system in some low-performance hardware environments, integration in the monitoring system, and setting a higher optimization frequency.

[0015] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the target parameter includes one or more of a delay coefficient, a sensitivity coefficient, and a deviation coefficient. In this case, since multiple coefficients can be taken into account, the accuracy of glucose concentration calibration can be effectively improved compared to the current naive algorithm that only considers the sensitivity coefficient. Since the glucose concentration of blood and the glucose concentration of interstitial fluid are not exactly the same, and the changes are not completely synchronized, considering the delay coefficient can make the calibrated continuous blood glucose monitor more in line with the physiological characteristics of the target.

[0016] In addition, in the automatic calibration system of the first aspect of the present disclosure, optionally, the reference value of the target parameter includes at least one of the factory parameters or historical target parameters of the continuous glucose monitoring system. In this case, the existing parameter values ​​with high confidence can be effectively utilized.

[0017] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the parameter calibration unit includes a loss calculation unit, the loss calculation unit obtains the error term of the loss function based on the second concentration of the multiple detection points and the sensor signals of the multiple detection points, the sensor signals are current signals, the regularization term of the loss function is obtained based on the reference value of the target parameter, and the loss function is obtained based on the error term and the regularization term. In this case, the estimated value of the target parameter can be made as close to the true value as possible and does not deviate from the reference value, thereby obtaining a more suitable target parameter for calibration.

[0018] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the error term includes a first type of error term, and the first type of error term is used to simultaneously calibrate the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameter; or the error term includes a second type of error term, and the second type of error term is used to simultaneously calibrate the sensitivity coefficient and deviation coefficient in the target parameter, and calibrate the delay coefficient in the target parameter based on the calibrated sensitivity coefficient and deviation coefficient. In this case, the delay coefficient, sensitivity coefficient and deviation coefficient are optimized simultaneously, and the mutual influence between the three coefficients is comprehensively considered, which is more in line with the actual situation, and the target value obtained is also more in line with the physiological characteristics of the target object. In addition, for the simultaneous optimization of the sensitivity coefficient and the deviation coefficient, but the delay coefficient does not participate in the optimization, the mutual influence between the sensitivity coefficient and the deviation coefficient is comprehensively considered and the situation of the instability of the delay coefficient is considered at the same time, which can support the situation of the instability of the delay coefficient, which is conducive to improving the stability of the performance of the monitoring system. In addition, since multiple coefficients are taken into consideration, the accuracy of the first concentration can be effectively improved compared to the current solution that only considers the sensitivity coefficient. At the same time, since the second concentration and the first concentration may not be exactly the same and the changes may not be completely synchronized, considering the deviation coefficient and the delay coefficient can better fit the physiological characteristics of the target object, thereby obtaining a more accurate first concentration.

[0019] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the first type of error term is expressed as:

[0020]

[0021] Among them, E1 represents the first type of error term, I(t i +τ) represents the tth of the multiple detection points i +τ time sensor signal, BG(t i ) represents the tth i The second concentration at the time, N represents the number of the plurality of detection points, i represents the index of the plurality of detection points, t i represents the detection time of the i-th detection point, τ represents the delay coefficient, s represents the sensitivity coefficient, and b represents the deviation coefficient. In this case, the delay coefficient, sensitivity coefficient, and offset coefficient obtained by the loss function can match the current-glucose concentration model, and the delay coefficient, sensitivity coefficient, and offset coefficient obtained in this way can make the loss function smaller.

[0022] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the second type of error term is expressed as:

[0023]

[0024] Where E2 represents the second type of error term, I(t i +τ i ) represents the tth of the multiple detection points i +τ i The sensor signal at the moment, BG2(t i ) represents the tth i The second concentration at the time, N represents the number of the plurality of detection points, i represents the index of the plurality of detection points, t i represents the detection time of the i-th detection point, τ i represents the delay coefficient corresponding to the i-th detection point, s represents the sensitivity coefficient, and b represents the deviation coefficient. In this case, different detection points can be allowed to have different delays, so as to further fit the actual situation and thus improve the accuracy of calibration.

[0025] In addition, in the automatic calibration system involved in the first aspect of the present disclosure, optionally, the regularization term is:

[0026] N=λ(p-p0) 2 ,

[0027] Wherein, N represents the regularization term, λ represents the regularization coefficient, p represents the value of the target parameter obtained after calibration, and p0 represents the reference value of the target parameter. Since the target parameter will not fluctuate too much in a short period of time, the introduction of the regularization term can make the loss function more reasonable and can further improve the accuracy of the target value of the target parameter obtained after iteration. In this case, by introducing the regularization coefficient, the preset values ​​of different target parameters can have different weight coefficients, thereby increasing the weight of the preset value with high confidence and reducing the weight of the preset value with low confidence.

[0028] The second aspect of the present disclosure also provides an automatic calibration method for a continuous glucose monitoring system, the automatic calibration method is used to calibrate the target parameters of the continuous glucose monitoring system, the target parameters are used to convert the sensor signals of the target object acquired by the continuous glucose monitoring system into corresponding glucose concentrations as first concentrations, the automatic calibration method comprises: acquiring second concentrations of the multiple detection points, the second concentration being the glucose concentration of the target object's blood; acquiring first concentrations corresponding to the multiple detection points; detecting whether an abnormal situation occurs in the continuous glucose monitoring system based on the second concentrations of the multiple detection points and the first concentrations of the multiple detection points, the abnormal situation comprising that there are no less than a preset number of detection points among the multiple detection points that satisfy a preset deviation condition, the preset deviation condition comprising that the difference between the first concentration and the second concentration of the detection point is no less than a first preset value or no greater than a second preset value; and if the abnormal situation is detected, acquiring a loss function related to the target parameter based on the second concentrations of the multiple detection points, the sensor signals of the multiple detection points and the reference value of the target parameter, and calibrating the target parameter by optimizing the loss function.

[0029] In the second aspect of the present disclosure, the target parameter is calibrated based on abnormal situation detection. In this case, the target parameter can be calibrated based on a reasonable timing, thereby improving the accuracy and availability of the continuous glucose monitoring system. At the same time, because the first concentration will be continuously greater than or continuously less than the second concentration when a system deviation occurs, therefore, when an abnormal situation is detected, the system deviation can be effectively measured and the continuous glucose monitoring system can be calibrated based on the size relationship of the difference between the first concentration and the second concentration, and the influence of random deviation on the calibration timing can be reduced. In addition, it can be performed automatically based on abnormal situation detection, thereby realizing automatic calibration of the continuous glucose monitoring system. In addition, optimizing the loss function based on data from multiple detection points can keep the loss function at a small value during the optimization process, reducing the influence of the error of the blood glucose meter.

[0030] According to the present disclosure, the calibration timing can be automatically determined and the accuracy of the glucose concentration readings of the continuous glucose monitoring system can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The present disclosure will now be explained in further detail by way of example only with reference to the accompanying drawings, in which:

[0032] Figure 1 is a schematic diagram showing a glucose monitoring scenario involved in the examples of the present disclosure.

[0033] Figure 2 is a system block diagram showing an automatic calibration system involved in examples of the present disclosure.

[0034] Figure 3 is a system block diagram showing another example of an automatic calibration system according to an example of the present disclosure.

[0035] Figure 4 1 is a flow chart showing the overall process of calibrating target parameters of the calibration system involved in the present disclosure.

[0036] Figure 5 It is a schematic diagram showing the process of determining abnormal situations involved in the present disclosure.

[0037] Figure 6 is a block diagram showing a detection unit involved in the example of the present disclosure.

[0038] Figure 7 is a block diagram showing a parameter calibration unit involved in the example of the present disclosure.

[0039] Figure 8 is a block diagram showing another example of a parameter calibration unit involved in the examples of the present disclosure.

[0040] Fig. 9 is a block diagram showing a second type of optimization unit involved in the examples of the present disclosure.

[0041] Description of reference numerals:

[0042] Monitoring system…1, processing system…2, automatic calibration system…3, second concentration measurement unit…30, data receiving and storage unit…31, detection unit…32, difference calculation unit…321, judgment unit…322, parameter calibration unit…33, loss calculation unit…331, first type optimization unit…332, second type optimization unit…333, delay unit…3331, blood glucose meter…4. DETAILED DESCRIPTION

[0043] Hereinafter, the preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same symbols are assigned to the same components, and repeated descriptions are omitted. In addition, the accompanying drawings are only schematic diagrams, and the ratio of the dimensions of the components or the shapes of the components may be different from the actual ones.

[0044] It should be noted that the terms "including" and "having" and any variations thereof in the present disclosure, such as a process, method, system, product or device that includes or has a series of steps or units, are not necessarily limited to those steps or units clearly listed, but may include or have other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0045] The present disclosure example relates to a continuous glucose monitoring system (hereinafter referred to as the monitoring system), which calibrates the glucose concentration based on a calibrated target parameter, thereby improving the accuracy of the glucose concentration reading of the continuous glucose monitoring system and improving the usability. The processing system involved in the present disclosure example can be applied to the scenario of calibrating the monitoring system.

[0046] The monitoring system involved in the example of the present disclosure can be used to continuously monitor the glucose concentration of the body fluid of the target object and the change of the glucose concentration over a period of time. In addition, the monitoring system can obtain the sensor signal of the target object, the sensor signal can match the real glucose concentration of the body fluid of the target object, and the glucose concentration obtained by processing the sensor signal can be used as the glucose concentration of the body fluid of the target object. That is, the change of the sensor signal can reflect the change of the real glucose concentration of the target object, and the glucose concentration obtained by processing the sensor signal can be regarded as the glucose concentration of the body fluid of the target object.

[0047] In some examples, the monitoring system has a target parameter, and the target parameter can be used to convert the sensor signal of the target object obtained by the monitoring system into a corresponding glucose concentration. In some examples, the glucose concentration obtained by the monitoring system can be calibrated by calibrating the target parameter.

[0048] In addition, the sensing signal may be a signal related to an electrochemical reaction. Preferably, the sensing signal may be a current signal. Thus, the sensing signal may be converted into a glucose concentration more conveniently.

[0049] In addition, the examples disclosed herein may not be limited to glucose concentration. For example, by changing the sensors of the monitoring system, data of other body fluid components other than glucose may also be obtained. The body fluid components here may be, for example, acetylcholine, amylase, bilirubin, cholesterol, chorionic gonadotropin, creatine kinase, creatine, creatinine, DNA, fructosamine, glucose, glutamine, growth hormone, hormone, ketone body, lactate, oxygen, peroxide, prostate-specific antigen, prothrombin, RNA, thyroid stimulating hormone, troponin, etc. In other examples, the concentration of drugs in body fluids may also be monitored, for example, antibiotics (e.g., gentamicin, vancomycin, etc.), digitoxin, digoxin, theophylline, and warfarin, etc.

[0050] For the convenience of description below, the glucose concentration of the body fluid of the target object obtained by the monitoring system is referred to as the first concentration, and the blood glucose concentration of the target object is referred to as the second concentration.

[0051] The target object involved in the examples of the present disclosure may be a human body. For example, the target object may be a diabetic patient.

[0052] The body fluid involved in the examples of the present disclosure may refer to the liquid substance in the body of the target object, which is specifically related to the source of the sensing signal. In some examples, body fluids can be divided into intracellular fluid, plasma, and interstitial fluid, etc. Preferably, the body fluid used to monitor the glucose concentration may be interstitial fluid (i.e., body fluid in the interstitial space). In this case, since the change in glucose concentration of the interstitial fluid can better reflect the cell metabolism, the glucose concentration of the interstitial fluid can better reflect the change in glucose concentration. For diabetic patients, monitoring the change in glucose concentration is more important than monitoring the high and low values ​​of glucose concentration.

[0053] In some examples, the target parameters may include one or more of a delay coefficient, a sensitivity coefficient, and a deviation coefficient. In some examples, the delay coefficient may represent the time delay in the measurement of glucose concentration by the monitoring system (e.g., a sensor of the monitoring system), the sensitivity coefficient may represent the sensitivity of the monitoring system, and the deviation coefficient may represent the deviation of the monitoring system. In some examples, the deviation of the monitoring system may come from the systematic deviation of the hardware of the monitoring system itself or the deviation between the actual glucose concentration and the glucose concentration of the body fluid at the location of the monitoring system caused by physiological reasons of the target object itself. In addition, it should be noted that the scheme involved in the examples of the present disclosure may not be limited to the above-mentioned target parameters, and can be adjusted according to the target parameters to be adjusted.

[0054] Examples of the present disclosure will be described in detail below with reference to the accompanying drawings. Figure 1 It is a schematic diagram showing the glucose monitoring scenario involved in the example of the present disclosure. It should be noted that the scenario described in the example of the present disclosure is to more clearly illustrate the technical solution of the example of the present disclosure, and does not constitute a limitation on the technical solution provided by the example of the present disclosure.

[0055] refer to Figure 1 , taking the interstitial fluid as an example of body fluid, in a glucose monitoring environment, the monitoring system 1 can continuously collect glucose data of the target object (e.g., sensor signals or glucose concentrations corresponding to sensor signals) and transmit them to corresponding processing components. Specifically, the sensor in the monitoring system 1 can be partially or completely placed under the skin of the target object and react with glucose in the interstitial fluid to obtain the sensor signal. The glucose concentration obtained by processing the sensor signal can be transmitted to the corresponding processing component (e.g., processing system 2) via a receiver of the monitoring system 1 located outside the body of the target object (e.g., on the surface of the body).

[0056] In some examples, the monitoring system 1 may collect the sensor signal of the target object and transmit it to the processing system 2 for processing to obtain the glucose concentration. In some examples, the processing system 2 may process the sensor signal based on the target parameter of the monitoring system 1 to obtain the glucose concentration.

[0057] In some examples, an automatic calibration system 3 for calibrating a target parameter of the monitoring system 1 may be included in the glucose monitoring environment. The automatic calibration system 3 may calibrate the target parameter based on the sensor signal acquired by the monitoring system 1 and the second concentration measured by the blood glucose meter 4. In this case, through the calibration operation, the target parameter can adapt to changes in the environment related to the target object (such as changes in the monitoring system 1 itself, changes in the physical environment of the target object, or changes in the external environment, etc.), thereby improving the accuracy of obtaining the glucose concentration using the target parameter. In addition, the blood glucose meter 4 may be part of the automatic calibration system 3 or may be independent of the automatic calibration system 3. In some examples, the blood glucose meter 4 may be a fingertip blood glucose meter.

[0058] Figure 2 is a system block diagram showing an automatic calibration system 3 according to an example of the present disclosure. Figure 3 is a system block diagram showing another example of the automatic calibration system 3 involved in the example of the present disclosure.

[0059] For some examples, see Figure 1 , the blood glucose meter 4 can be independent of the automatic calibration system 3; see Figure 2 and Figure 3 The blood glucose meter 4 may also be a part of the automatic calibration system 3 , in which case the second concentration measuring unit 30 may be used to represent the blood glucose meter 4 .

[0060] As described above, the automatic calibration system 3 can be used to calibrate the target parameters of the monitoring system 1. To this end, the example of the present disclosure also provides an example of an automatic calibration system 3 for calibrating the target parameters of the monitoring system 1. The automatic calibration system 3 can also be called an automatic update system, automatic setting system or automatic adjustment system of the target parameters. Figure 2 The automatic calibration system 3 may include a second concentration measurement unit 30 , a detection unit 32 and a parameter calibration unit 33 .

[0061] In some examples, the second concentration measurement unit 30 (i.e., the blood glucose meter 4) can be used to measure the second concentration of the target object. In some examples, the detection unit 32 can detect whether an abnormal situation occurs in the monitoring system 1 based on the second concentration of multiple detection points and the first concentration of multiple detection points. In some examples, the detection point can be a time point for obtaining the second concentration. In some examples, the abnormal situation includes that there are no less than a preset number of detection points in the multiple detection points that meet the preset deviation condition. In some examples, the preset deviation condition includes that the difference between the first concentration and the second concentration of the detection point is not less than the first preset value or not greater than the second preset value.

[0062] In some examples, the number of multiple detection points can be set. In some examples, the number of multiple detection points can be set to twice the number of target parameters to be calibrated. Thus, the target value of the target parameter can be obtained using a relatively small number of detection points. For example, if the number of target parameters to be calibrated is 3, the number of multiple detection points can be set to 6. It should be noted that the number of multiple detection points can also be adjusted according to the allowable error range of the first concentration. For example, if the accuracy requirement is not too high, the number of multiple detection points can also be set to be less than twice the number of target parameters to be optimized.

[0063] In some examples, the first preset value may be not less than 0, and the second preset value may be the opposite of the first preset value. In this case, the first preset value can be for the case where the first concentration is greater than the second concentration, and the second preset value can be for the case where the first concentration is less than the second concentration.

[0064] In some examples, the arrangement of the detection points of not less than the preset number in the plurality of detection points may be front-to-back adjacent. In this case, since the system deviation is biased, detection based on the front-to-back adjacent detection points can more accurately determine whether the system deviation occurs, and thus can determine a more accurate calibration timing. In some examples, the preset number is not greater than the number of the plurality of detection points set. In addition, in some examples, the preset number may be 3. Thus, abnormal situations can be detected when the amount of data is small.

[0065] In some examples, such as Figure 3 As shown, the automatic calibration system 3 may further include a data receiving and storing unit 31. The data receiving and storing unit 31 may be used to receive and store the second concentrations of the plurality of detection points, the first concentrations of the plurality of detection points, and the sensing signals of the plurality of detection points.

[0066] In some examples, the detection point may be a time point at which the second concentration of the target object is measured using the blood glucose meter 4. In this case, data of multiple detection points can be stored, so that the detection unit 32 can conveniently use and operate the stored data, and then can determine whether there is an abnormality based on the relationship between the first concentration and the second concentration obtained in the previous period of time.

[0067] In some examples, the data receiving and storing unit 31 is further used to detect whether there are abnormal data points in the second concentration of the multiple detection points. In some examples, the abnormal data points include abnormal second concentrations. If the presence of abnormal data points is detected, the abnormal data points are removed from the second concentrations of the multiple detection points. In this case, removing the abnormal data points can reduce the impact of the erroneous second concentration.

[0068] In some examples, the parameter calibration unit 33 may acquire a loss function related to the target parameter based on the second concentrations of the plurality of detection points, the sensing signals of the plurality of detection points, and the reference value of the target parameter, and calibrate the target parameter by optimizing the loss function.

[0069] Figure 4 1 is a flow chart showing the overall process of calibrating target parameters of the calibration system involved in the present disclosure.

[0070] Below, with Figure 3 The calibration system 3 shown is taken as an example, and combined with Figure 4 The steps of calibrating the target parameters of the monitoring system 1 by the automatic calibration system 3 are described in detail.

[0071] like Figure 4 As shown, the steps of calibrating the target parameters of the monitoring system 1 by the calibration system 3 include: receiving and storing data from multiple detection points (step S100); detecting whether there is an abnormal situation based on the data from multiple detection points (step S200); and calibrating the target parameters based on the data from multiple detection points and the reference value of the target parameters (step S300).

[0072] In some examples, in step S100, data of multiple detection points may be received and stored. Specifically, the second concentration measurement unit 30 may obtain the second concentration of the target object, which may be input to the data receiving and storage unit 31. In some examples, when the data receiving and storage unit 31 receives and stores the second concentration of a detection point, it may also simultaneously receive and store the sensing signal and the first concentration of the monitoring system 1 at the same or adjacent time point. That is, each detection point may correspond to three types of data.

[0073] In some examples, in step S100, as described above, the data receiving and storage unit 31 may detect whether there is an abnormal data point in the second concentrations of the plurality of data points. In some examples, the data receiving and storage unit 31 may detect whether the second concentration corresponding to a single data point is an abnormal data point based on the statistical values ​​(including the mean, median, mode, etc.) of the second concentrations of the plurality of data points. In some examples, the data receiving and storage unit 31 may detect whether the second concentration corresponding to a single data point is an abnormal data point based on the state of the target object at the detection point (e.g., eating and fasting state).

[0074] In some examples, after detecting an abnormal data point, the data receiving and storage unit 31 may remove the abnormal data point from the plurality of data points. In some examples, the sensor signal and the first concentration of the related detection point (i.e., the detection point corresponding to the abnormal detection point) may also be removed together when removing the abnormal data point.

[0075] In some examples, in step S100 , first concentrations corresponding to a plurality of detection points may be acquired.

[0076] In some examples, the frequency of acquiring the sensing signal and the first concentration may be once every 5 minutes. In some examples, the frequency of measuring the second concentration may be once a day, that is, there may be one detection point every day. Therefore, the acquisition frequency of the first concentration is higher than that of the second concentration, and among the multiple first concentrations obtained by the monitoring system 1, the first concentration whose measurement time is closest to the detection point can be selected as the first concentration of the detection point.

[0077] In some examples, the measurement frequency of the second concentration may be greater than once a day. In some examples, the measurement frequency of the second concentration may be no less than three times a day. Thus, calibration can be performed using multiple detection points per day. In some examples, the time point for measuring the second concentration daily may be a fixed time of the day.

[0078] Figure 5 It is a schematic diagram showing the process of determining abnormal situations involved in the present disclosure.

[0079] In some examples, in step S200, it is possible to detect whether there is an abnormal situation based on data from multiple detection points. Figure 5 , step S200 may include: obtaining a preset number of continuous detection points (step S210), calculating the difference between the first concentration and the second concentration of each detection point (step S220), determining whether each difference satisfies a preset deviation condition (step S230), and determining whether an abnormal situation exists (step S240).

[0080] In some examples, the detection unit 32 can detect whether an abnormality occurs in the detection system 1 based on the second concentration of the plurality of detection points and the first concentration of the plurality of detection points. In some examples, the abnormality includes that no less than a preset number of detection points among the plurality of detection points satisfy the deviation condition.

[0081] In some examples, in step S210, a preset number of consecutive detection points may be acquired.

[0082] In some examples, in step S200 and step S210, as described above, the preset number may be 3 to 10, for example, the preset number may be 3, 4, 5, 6, or 7, etc., but the present disclosure is not limited thereto, and the preset number may be any value. In some examples, as described above, the arrangement of the detection points not less than the preset number in the plurality of detection points may be front-to-back adjacent (in some examples, it may also be said that the detection points arranged according to the above conditions are continuous).

[0083] In some examples, in step S220, the difference between the first concentration and the second concentration at each detection point may be calculated. In some examples, the difference between the first concentration and the second concentration is obtained by subtracting the second concentration from the first concentration.

[0084] In some examples, in step S230, it can be determined whether each difference satisfies a preset deviation condition. In some examples, as described above, the preset deviation condition can include that the difference between the first concentration and the second concentration of the detection point is not less than a first preset value or not greater than a second preset value.

[0085] In some examples, in step S240, it can be determined whether an abnormal situation exists.

[0086] For example, if the first preset value is 0.5, the second preset value is -0.5. For three consecutive detection points, the corresponding first concentrations are 4.5, 4.3 and 4.6, and the second concentrations are 5.3, 5.0 and 5.3, respectively. The difference values ​​corresponding to each of the three detection points are -0.8, -0.7 and -0.7, respectively (all concentrations are in millimoles / liter, and the relevant description below also applies). Since the above differences are all less than 0, they can be compared with the second preset value. It can be seen that the difference values ​​corresponding to each of the above detection points are all less than the second preset value. At this time, the detection unit 32 can determine that an abnormal situation exists.

[0087] For another example, if the first preset value is 0.5, the second preset value is -0.5. For three consecutive detection points, the corresponding first concentrations are 6.6, 6.3, and 6.4, and the second concentrations are 5.5, 5.3, and 5.6, respectively. The difference values ​​corresponding to each of the three detection points are 1.1, 1.0, and 0.8, respectively. Since the above differences are all greater than 0, they are compared with the first preset value. It can be seen that the difference values ​​corresponding to each of the above detection points are greater than the first preset value. At this time, the detection unit can determine that an abnormal situation exists.

[0088] For another example, the first preset value is 0.5, and the second preset value is -0.5. For three consecutive detection points, the corresponding first concentrations are 6.6, 5.7, and 6.4, and the second concentrations are 5.5, 5.8, and 5.6, respectively. Then the difference values ​​corresponding to each of the three detection points are 1.1, -0.1, and 0.8, respectively. Since the above differences are not all greater than 0 or less than 0, it can be considered that the deviation between the first concentration and the second concentration is a random deviation, and the detection unit 32 can determine that the abnormal situation does not exist.

[0089] Figure 6 is a block diagram showing the detection unit 32 to which the present disclosure example relates.

[0090] In some examples, such as Figure 6As shown, the detection unit 32 may include a difference calculation unit 321 and a judgment unit 322. The difference calculation unit 321 may perform the difference calculation as exemplified above; and the judgment unit 322 may judge whether there is an abnormality according to the calculation result of the difference calculation.

[0091] In some examples, after the number of received and stored detection points reaches the number of set detection points, it can be detected whether there is an abnormality in the group of multiple detection points. In other examples, it can be determined whether the detection point meets the deviation condition when a single detection point is received and stored, and it can be determined that an abnormality exists if it is detected that no less than a preset number of continuous detection points meet the deviation condition.

[0092] In some examples, in step S300, the target parameter is calibrated based on the data of the multiple detection points and the reference value of the target parameter. Specifically, in step S200, if the detection unit 32 detects that an abnormal situation exists, the parameter calibration unit 33 can obtain a loss function related to the target parameter based on the second concentrations of the multiple detection points, the sensor signals of the multiple detection points and the reference value of the target parameter, and calibrate the target parameter by optimizing the loss function.

[0093] In some examples, in step S300, the parameter calibration unit 33 may obtain a calibration result (also referred to as a target value or a calibration value) of the target parameter by minimizing a loss function.

[0094] In some examples, data of multiple detection points used to obtain target values ​​may also be referred to as calibration data.

[0095] In various examples of the present disclosure, in step S300, obtaining the target value of the target parameter by optimizing the loss function may be directly obtaining the target values ​​of all target parameters by optimizing the loss function (that is, all target parameters are the target parameters to be optimized) or directly obtaining the target values ​​of a part of the target parameters by optimizing the loss function, and then obtaining the target values ​​of another part of the target parameters based on the target values ​​of the target parameters that have been obtained. The specific method may be adjusted according to actual conditions.

[0096] In addition, the loss function can represent the degree of difference (also referred to as the degree of inconsistency) between the second concentration of the detection point in the calibration data and the first concentration of the corresponding sensor signal obtained through processing by the monitoring system 1. That is, the loss function can represent the degree of difference between the true value of the detection point in the calibration data (i.e., the second concentration obtained by the blood glucose meter 4) and the estimated value (i.e., the first concentration corresponding to the sensor signal obtained by the monitoring system 1). In this case, based on the loss function, the estimated value can be made as close to the true value as possible and finally a more appropriate target value of the target parameter can be obtained. That is, a more appropriate target value can be obtained to reduce the error between the first concentration obtained through processing by the monitoring system 1 and the corresponding second concentration to meet the corresponding requirements.

[0097] In some examples, the loss function can be obtained based on the second concentration of multiple detection points in the calibration data, the sensor signal, and the reference value of the target parameter. In this case, the target value of the target parameter is optimized and obtained using the glucose concentration (i.e., blood sugar) and the sensor signal of the blood at multiple detection points. By optimizing the data of multiple detection points to keep the loss function at a small value during the optimization process, the influence of the finger blood error can be weakened, thereby obtaining a more accurate target value of the target parameter, and then the accuracy of obtaining the first concentration using the target value of the target parameter can be improved. In addition, combined with the reference value of the target parameter, the influence of the finger blood error can be further weakened.

[0098] Specifically, the above-mentioned degree of difference can be determined based on the first degree of difference (also referred to as the error term of the loss function) between the second concentration (that is, the true value) of each detection point and the first concentration (that is, the estimated value) obtained by processing the sensor signal, and the second degree of difference (also referred to as the regularization term of the loss function) between the target value of the target parameter and the reference value of the target parameter. In this case, the estimated value can be made as close as possible to the true value without deviating from the reference value, thereby obtaining a more appropriate target value of the target parameter, which can further reduce the impact of the finger blood error. In other examples, the above-mentioned degree of difference can also be determined based only on the first degree of difference.

[0099] In some examples, the parameter calibration unit 33 may include a loss calculation unit 331 (described later). The loss calculation unit 331 may obtain an error term of a loss function based on the second concentrations of the plurality of detection points and the sensing signals of the plurality of detection points. In some examples, the sensing signal is a current signal. In some examples, a regularization term of a loss function may be obtained based on a reference value of the target parameter. In some examples, a loss function may be obtained based on an error term and a regularization term. In this case, the estimated value of the target parameter can be made as close to the true value as possible and does not deviate from the reference value, thereby obtaining a more suitable target parameter for calibration.

[0100] In some examples, in step S300, the error term may be obtained based on the second concentration and the sensor signal of the detection point in the optimization data. In some examples, the optimization may be performed using maximum likelihood estimation based only on the difference degree of the error term.

[0101] In some examples, the error term of the loss function corresponding to the parameter calibration unit 33 may be expressed as formula (1):

[0102]

[0103] Where E represents the error term of the loss function corresponding to the parameter calibration unit 33, BG1(t i ) represents the tth i The first concentration at time, BG2(t i ) represents the tth i The second concentration at time, N represents the number of detection points in the optimization data, i represents the index of the detection point in the optimization data, t i represents the detection time of the i-th detection point. As described above, the first concentration is obtained by processing the sensor signal using the target value of the target parameter. Thus, the error term for the target parameter can be obtained.

[0104] In some examples, the regularization term can be obtained based on a reference value of the target parameter. In some examples, the reference value of the target parameter includes at least one of a factory parameter of the monitoring system 1 or a historical target parameter. In this case, existing parameter values ​​with a high confidence level can be effectively utilized. In some examples, the historical target parameter can be a target parameter obtained by a previous calibration.

[0105] In some examples, the regularization term may be formula (2):

[0106] N=λ(p-p0) 2 ,

[0107] Among them, N represents the regularization term, λ represents the regularization coefficient, p represents the value of the target parameter obtained after calibration (i.e., the target value), and p0 represents the reference value of the target parameter. Since the target parameter will not fluctuate too much in a short period of time, the introduction of the regularization term can make the loss function more reasonable and further improve the accuracy of the target value of the target parameter obtained after iteration. In this case, by introducing the regularization coefficient, the preset values ​​of different target parameters can have different weight coefficients, thereby increasing the weight of the preset value with high confidence and reducing the weight of the preset value with low confidence.

[0108] In some examples, in step S300, the target value of the target parameter can be found by optimizing the loss function, and the target value of the target parameter can maximize the posterior probability that the monitoring system 1 measures the first concentration that is the same as the second concentration. In this case, the target value of the target parameter obtained by using the maximum a posteriori probability estimation can not only make the likelihood function larger, but also the prior probability of the target value of the target parameter itself appearing is larger, thereby reducing the impact of the error of the second concentration. In general, the calibration data (such as the second concentration) used for optimization may have errors, and considering the prior information in the optimization process can reduce the risk of large errors caused by calibration data with errors. In some examples, the loss function can be determined using a Bayesian algorithm.

[0109] In some examples, the parameter calibration unit 33 can optimize the loss function through an iterative formula to obtain the target value of the target parameter. In some examples, the loss function of the target parameter can be obtained based on the calibration data, the iterative formula of the target parameter can be obtained based on the loss function, and the target value of the target parameter can be obtained based on the iterative formula. In some examples, multiple iterations can be performed based on the iterative formula until a condition for stopping iteration is reached. In some examples, the condition for stopping iteration can be that the number of iterations reaches a preset number or the loss function reaches a preset value.

[0110] In some examples, in step S300, the iterative formula can be obtained based on the loss function and using the Hessian matrix. Thus, the iterative formula can be simplified by using the Hessian matrix.

[0111] In addition, there can be multiple iterative formulas. Preferably, the iterative formula can be based on the Newton-Raphson iteration method. In this case, the nonlinear equations are gradually reduced to linear equations through the Newton-Raphson iteration method, and then the loss function is solved, which can simplify the iteration process. Specifically, the iterative formula can be obtained by accumulating the product of the inverse matrix of the Hesser matrix and the gradient of the loss function relative to the target parameter. To this end, the present disclosure also provides an iterative formula based on the Hesser matrix and the Newton-Raphson iteration method. It should be noted that it does not represent a limitation to the present disclosure, and other iterative formulas can be selected according to actual conditions.

[0112] Specifically, the iterative formula for optimizing the loss function is:

[0113]

[0114] Among them, p k+1 represents the target value of the target parameter obtained in the k+1th iteration, p k represents the target value of the target parameter obtained at the kth iteration, represents the gradient of the loss function with respect to the target parameter, f represents the loss function, H represents the Heather matrix, H-1 represents the inverse matrix of H, and k represents the index of the number of iterations. In this case, by iterating and optimizing the target parameters using the iterative formula, the target parameters that meet the conditions defined by the loss function can be obtained for calibration. In addition, based on the iterative formula of the Heather matrix, the convergence speed can reach the second order, the convergence is fast, and the optimization efficiency is high, which is conducive to the layout of the optimization system in some low-performance hardware environments, integration in the monitoring system 1, and setting a higher optimization frequency.

[0115] In addition, in step S300, and H can be determined by the target parameters that need to be calibrated.

[0116] In some examples, It can be a column matrix composed of the signs of the first-order partial derivatives of f with respect to the target parameters to be calibrated. For example, in the case of calibrating the delay coefficient, sensitivity coefficient and deviation coefficient at the same time, in, represents the sign of partial derivative, τ represents the delay coefficient, s represents the sensitivity coefficient, and b represents the deviation coefficient. For another example, when calibrating the sensitivity coefficient and the deviation coefficient at the same time,

[0117] As described above, H can be determined by the target parameter that needs to be calibrated. In some examples, Among them, x1, x2, ..., xj represent multiple target parameters that need to be calibrated, and j represents the index of the target parameter that needs to be calibrated.

[0118] For example, in the case of calibrating the delay coefficient, sensitivity coefficient and deviation coefficient at the same time,

[0119] Figure 7 is a block diagram showing a parameter calibration unit 33 according to an example of the present disclosure.

[0120] Figure 8 is a block diagram showing another example of the parameter calibration unit 33 involved in the example of the present disclosure.

[0121] like Figure 7 and Figure 8 As shown, in some examples, the parameter calibration unit 33 may include a loss calculation unit 331. The loss calculation unit 331 may perform the above process of obtaining the loss function to obtain the loss function.

[0122] In some examples, in step S300, the error term may include a first type of error term; the first type of error term may be used to calibrate the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameter at the same time. Or the error term may include a second type of error term; the second type of error term may be used to calibrate the sensitivity coefficient and deviation coefficient in the target parameter at the same time, and calibrate the delay coefficient in the target parameter based on the calibrated sensitivity coefficient and deviation coefficient. In this case, the delay coefficient, sensitivity coefficient and deviation coefficient are optimized at the same time, and the mutual influence between the three coefficients is comprehensively considered, which is more in line with the actual situation, and the target value obtained is more in line with the physiological characteristics of the target object. In addition, for optimizing the sensitivity coefficient and the deviation coefficient at the same time, but the delay coefficient does not participate in the optimization, the mutual influence between the sensitivity coefficient and the deviation coefficient is comprehensively considered, and the situation of the instability of the delay coefficient is considered at the same time, which can support the situation of the instability of the delay coefficient, which is conducive to improving the stability of the performance of the monitoring system 1. In addition, since multiple coefficients are taken into account, compared with the current solution that only considers the sensitivity coefficient, the accuracy of the first concentration can be effectively improved. At the same time, since the second concentration and the first concentration may not be exactly the same, the changes may not be completely synchronized. Considering the deviation coefficient and the delay coefficient can be more in line with the physiological characteristics of the target object, and thus a more accurate first concentration can be obtained.

[0123] refer to Figure 7 , the parameter calibration unit 33 may include a first type optimization unit 332. The first type optimization unit 332 may simultaneously obtain target values ​​of the delay coefficient, sensitivity coefficient, and deviation coefficient in the target parameter based on the first type error term. That is, assuming that the delay coefficient, sensitivity coefficient, and deviation coefficient in the target parameter are relatively stable, the delay coefficient, sensitivity coefficient, and deviation coefficient may be used as unknown quantities (also referred to as variables) in the loss function.

[0124] In some examples, the first type of optimization unit 332 corresponds to the tth i The first concentration at the moment can be expressed as formula (3):

[0125]

[0126] Among them, I(t i +τ) represents the tth i +τ is the sensing signal at time, τ represents the delay coefficient, s represents the sensitivity coefficient, and b represents the deviation coefficient.

[0127] Based on equations (1) and (3), the first type of error term can be expressed as equation (4):

[0128]

[0129] Among them, E1 represents the first type of error term, I(t i+τ) represents the tth of multiple detection points i +τ time sensor signal, BG(t i ) represents the tth i The second concentration at the moment, N represents the number of multiple detection points, i represents the index of multiple detection points, t i represents the detection time of the i-th detection point, τ represents the delay coefficient, s represents the sensitivity coefficient, and b represents the deviation coefficient. In this case, the delay coefficient, sensitivity coefficient, and offset coefficient obtained by the loss function can match the current-glucose concentration model, and the delay coefficient, sensitivity coefficient, and offset coefficient obtained in this way can make the loss function smaller.

[0130] As described above, the first type of optimization unit 332 can be used to simultaneously obtain the target values ​​of the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameters. Accordingly, for the regularization term of the loss function corresponding to the first type of optimization unit 332, p can be the target value of the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameters, and p0 can be the reference value of the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameters.

[0131] refer to Figure 8 , the parameter calibration unit 33 may include a second type optimization unit 333. The second type optimization unit 333 may simultaneously obtain the target values ​​of the sensitivity coefficient and the deviation coefficient in the target parameter based on the second type error, and obtain the target value of the delay coefficient in the target parameter based on the target values ​​of the sensitivity coefficient and the deviation coefficient that have been obtained. That is, in the case where the delay coefficient is not very stable, assuming that the sensitivity coefficient and the deviation coefficient in the target parameter are relatively stable, the delay coefficient may not be estimated temporarily, and the sensitivity coefficient and the deviation coefficient may be used as unknown quantities in the loss function. Specifically, the second type optimization unit 333 may obtain the loss function based on the calibration data, and simultaneously obtain the target values ​​of the sensitivity coefficient and the deviation coefficient in the target parameter by optimizing the loss function, and obtain the target value of the delay coefficient in the target parameter based on the target values ​​of the sensitivity coefficient and the deviation coefficient that have been obtained.

[0132] In some examples, the second type of optimization unit 333 corresponds to the tth i The first concentration at the moment can be expressed as formula (5):

[0133]

[0134] Among them, I(t i +τ i ) represents the tth i +τ i The sensor signal at time τ i represents the delay coefficient corresponding to the i-th detection point (that is, τ iIt is no longer a parameter that needs to be optimized), s represents the sensitivity coefficient, and b represents the deviation coefficient. In this case, the delay coefficient is no longer a parameter that needs to be optimized, and it can support the situation where the delay coefficient is unstable, which is conducive to improving the stability of the performance of the monitoring system 1.

[0135] In some examples, τ i It can be determined based on the sensitivity coefficient and deviation coefficient corresponding to the i-th detection point. That is, τ i It can be determined based on the current sensitivity coefficient and deviation coefficient of the i-th detection point. Specifically, for each detection point, a target delay coefficient can be found as the delay coefficient of the detection point, and the target delay coefficient satisfies the first concentration obtained by processing the sensor signal of the detection point based on the current sensitivity coefficient and deviation coefficient, and the target delay coefficient, and the corresponding second concentration is closest. In other words, you can try not to optimize the delay coefficient, and for each detection point, you can find an optimal delay so that the first concentration is closest to the second concentration under the current sensitivity coefficient and deviation coefficient.

[0136] Therefore, in some examples, based on equations (1) and (5), the second type error term can be expressed as equation (6):

[0137]

[0138] Among them, E2 represents the second type of error term, I(t i +τ i ) represents the tth of multiple detection points i +τ i The sensor signal at the moment, BG2(t i ) represents the tth i The second concentration at the moment, N represents the number of multiple detection points, i represents the index of multiple detection points, t i represents the detection time of the i-th detection point, τ i represents the delay coefficient corresponding to the i-th detection point, s represents the sensitivity coefficient, and b represents the deviation coefficient. In this case, different detection points can be allowed to have different delays, so as to further fit the actual situation and thus improve the accuracy of calibration.

[0139] Fig. 9 is a block diagram showing a second type of optimization unit 333 involved in the example of the present disclosure.

[0140] like Fig. 9 As shown, in some examples, the second type optimization unit 333 may include a delay unit 3331. In some examples, the delay unit 3331 may determine the target value of the delay coefficient in the target parameter based on the target value of the sensitivity coefficient and the deviation coefficient in the target parameter acquired by the second type optimization unit 333.

[0141] In some examples, in step S300, the delay unit 3331 can determine the target value of the delay coefficient in the target parameter in combination with the delay calculation data. In addition, the delay calculation data can include the second concentration and the sensing signal corresponding to a plurality of delayed detection points whose detection time is before the signal to be processed.

[0142] In some examples, the delay calculation data may be the calibration data described above. In this case, the acquired calibration data can be used. In addition, the detection point in the delay calculation data may be referred to as a delay detection point.

[0143] Specifically, the delay unit 3331 can obtain the target value of the delay coefficient in the target parameter based on the delay calculation data, the target values ​​of the sensitivity coefficient and the deviation coefficient in the acquired target parameter. In this case, the delay coefficient that does not participate in the optimization can be obtained by the target values ​​of the sensitivity coefficient and the deviation coefficient that have been acquired, which can reduce the risk that the delay coefficient caused by the instability of the delay coefficient may affect the system performance when participating in the optimization.

[0144] In some examples, the delay unit 3331 can obtain the delay coefficient corresponding to any delay detection point by minimizing the difference between the second concentration corresponding to any delay detection point among the multiple delay detection points and the first concentration corresponding to the sensor signal of the any delay detection point and use it as the target value of the delay coefficient in the target parameter. Thus, the target value of the delay coefficient can be obtained according to any delay detection point.

[0145] In other examples, the delay unit 3331 can search for a parameter value that minimizes the sum of the differences between the second concentration of all the delay detection points in the plurality of delay detection points and the first concentration corresponding to the sensor signal and use the parameter value as the target value of the delay coefficient in the target parameter. Thus, the target value of the delay coefficient can be obtained by synthesizing the data of the plurality of delay detection points.

[0146] Based on the above methods and modules, the blood glucose meter 4 can be used to calibrate the target parameters of the monitoring system 1 at a reasonable time, and then the first concentration obtained by the monitoring system 1 can be calibrated.

[0147] In some examples, the automatic calibration system 3 may transmit the calibrated target parameter to the processing system 2 , and the processing system 2 acquires the first concentration based on the calibrated target parameter and the sensor signal of the monitoring system 1 .

[0148] In summary, the present disclosure calibrates the target parameters based on abnormal situation detection. In this case, the target parameters can be calibrated based on reasonable timing, thereby improving the accuracy and availability of the continuous glucose monitoring system 1. At the same time, because when a system deviation occurs, the first concentration will be continuously greater than or continuously less than the second concentration, therefore, when an abnormal situation is detected, the system deviation can be effectively measured and the continuous glucose monitoring system 1 can be calibrated based on the size relationship of the difference between the first concentration and the second concentration, and the influence of random deviation on the calibration timing can be reduced. In addition, it can be performed automatically based on abnormal situation detection, thereby realizing automatic calibration of the continuous glucose monitoring system 1. In addition, optimizing the loss function based on data from multiple detection points can keep the loss function at a small value during the optimization process, reducing the influence of the error of the blood glucose meter 4.

[0149] According to the present disclosure, an automatic calibration system 3 and an automatic calibration method of a continuous glucose monitoring system 1 can be provided, which can automatically determine a calibration timing and thus improve the accuracy of the glucose concentration reading of the continuous glucose monitoring system 1 .

[0150] Although the present disclosure is specifically described above in conjunction with the accompanying drawings and examples, it is to be understood that the above description does not limit the present disclosure in any form. Those skilled in the art may modify and change the present disclosure as needed without departing from the essential spirit and scope of the present disclosure, and these modifications and changes all fall within the scope of the present disclosure.

Claims

1. An automatic calibration system for a continuous glucose monitoring system, characterized in that: The automatic calibration system is used to calibrate the target parameter of the continuous glucose monitoring system, and the target parameter is used to convert the sensor signal of the target object obtained by the continuous glucose monitoring system into a corresponding glucose concentration as the first concentration; the automatic calibration system includes a blood glucose meter, a detection unit and a parameter calibration unit, the blood glucose meter is used to measure the second concentration of the target object, the second concentration is the glucose concentration of the target object's blood, the detection unit detects whether the continuous glucose monitoring system has an abnormal situation based on the second concentrations of multiple detection points and the first concentrations of the multiple detection points, the detection points are time points for obtaining the second concentrations, the abnormal situation includes that there are no less than a preset number of the detection points among the multiple detection points that meet a preset deviation condition, the preset deviation condition includes that the difference between the first concentration and the second concentration of the detection point is not less than a first preset value or not greater than a second preset value, and if the detection unit detects that the abnormal situation exists, the parameter calibration unit obtains a loss function related to the target parameter based on the second concentrations of the multiple detection points, the sensor signals of the multiple detection points and the reference value of the target parameter, and calibrates the target parameter by optimizing the loss function.

2. The automatic calibration system according to claim 1, characterized in that: The detection points having a number not less than a preset number are arranged in a front-to-back manner among the plurality of detection points.

3. The automatic calibration system according to claim 1, characterized in that: It also includes a data receiving and storage unit, which is used to receive and store the second concentrations of the multiple detection points, the first concentrations of the multiple detection points, and the sensing signals of the multiple detection points.

4. The automatic calibration system according to claim 3, characterized in that: The data receiving and storage unit is also used to detect whether there are abnormal data points in the second concentrations of the multiple detection points, and the abnormal data points include abnormal second concentrations. If the abnormal data points are detected, the abnormal data points are removed from the second concentrations of the multiple detection points. 5 . The automatic calibration system according to claim 1 , wherein the first preset value is not less than 0, and the second preset value is the opposite of the first preset value.

6. The automatic calibration system according to claim 1, characterized in that: The iterative formula for optimizing the loss function is: Among them, p k+1 represents the target value of the target parameter obtained in the k+1th iteration, p k represents the target value of the target parameter obtained at the kth iteration, represents the gradient of the loss function with respect to the target parameter, f represents the loss function, H represents the Hesser matrix, H -1 represents the inverse matrix of H, and k represents the index of the number of iterations.

7. The automatic calibration system according to claim 1, characterized in that: The target parameters include one or more of a delay coefficient, a sensitivity coefficient and a deviation coefficient.

8. The automatic calibration system according to claim 1, characterized in that: The reference value of the target parameter includes at least one of a factory parameter or a historical target parameter of the continuous glucose monitoring system.

9. The automatic calibration system according to claim 1, characterized in that: The parameter calibration unit includes a loss calculation unit, which obtains an error term of the loss function based on the second concentrations of the multiple detection points and the sensor signals of the multiple detection points, wherein the sensor signals are current signals, obtains a regularization term of the loss function based on a reference value of the target parameter, and obtains the loss function based on the error term and the regularization term.

10. The automatic calibration system according to claim 7 or 9, characterized in that: The error terms include first-type error terms, which are used to simultaneously calibrate the delay coefficient, sensitivity coefficient and deviation coefficient in the target parameters; or the error terms include second-type error terms, which are used to simultaneously calibrate the sensitivity coefficient and deviation coefficient in the target parameters, and calibrate the delay coefficient in the target parameters based on the calibrated sensitivity coefficient and deviation coefficient.

11. The automatic calibration system according to claim 10, characterized in that: The first type of error term is expressed as: Among them, E1 represents the first type of error term, I(t i +τ) represents the tth of the multiple detection points i +τ time sensor signal, BG2(t i ) represents the tth i The second concentration at the time, N represents the number of the plurality of detection points, i represents the index of the plurality of detection points, t i represents the detection time of the i-th detection point, τ represents the delay coefficient, s represents the sensitivity coefficient, and b represents the deviation coefficient.

12. The automatic calibration system according to claim 10, characterized in that: The second type of error term is expressed as: Among them, E2 represents the second type of error term, I(t i +τ i ) represents the tth of the multiple detection points i +τ i The sensor signal at the moment, BG2(t i ) represents the tth i The second concentration at the time, N represents the number of the plurality of detection points, i represents the index of the plurality of detection points, t i represents the detection time of the i-th detection point, τ i represents the delay coefficient corresponding to the i-th detection point, s represents the sensitivity coefficient, and b represents the deviation coefficient.

13. The automatic calibration system according to claim 9, characterized in that: The regularization term is: N=λ(p-p0) 2 , Wherein, N represents the regularization term, λ represents the regularization coefficient, p represents the value of the target parameter obtained after calibration, and p0 represents the reference value of the target parameter.

14. A method for automatic calibration of a continuous glucose monitoring system, characterized in that: The automatic calibration method is used to calibrate the target parameters of the continuous glucose monitoring system, and the target parameters are used to convert the sensor signals of the target object obtained by the continuous glucose monitoring system into corresponding glucose concentrations as the first concentrations. The automatic calibration method includes: obtaining second concentrations of multiple detection points, where the second concentration is the glucose concentration of the target object's blood; obtaining first concentrations corresponding to the multiple detection points; detecting whether an abnormal situation occurs in the continuous glucose monitoring system based on the second concentrations of the multiple detection points and the first concentrations of the multiple detection points, wherein the abnormal situation includes that there are no less than a preset number of detection points among the multiple detection points that satisfy a preset deviation condition, and the preset deviation condition includes that the difference between the first concentration and the second concentration of the detection point is not less than a first preset value or not greater than a second preset value; and if the abnormal situation is detected, obtaining a loss function related to the target parameter based on the second concentrations of the multiple detection points, the sensor signals of the multiple detection points and the reference value of the target parameter, and calibrating the target parameter by optimizing the loss function.

Images