An optimized calibration-free method compatible with multiple subcutaneous glucose sensor outputs

Abstract

The application relates to an optimized calibration-free method compatible with outputs of various subcutaneous glucose sensors, and belongs to the technical field of data processing. The method comprises the following steps: obtaining normal conversation message codes, calibration message codes and background calculation message codes according to message codes; inputting the normal conversation message codes into a first data channel to obtain a target noise model, and obtaining effective data according to the normal conversation message codes and the target noise model; inputting the calibration message codes into a second data channel and inputting the background calculation message codes into a third data channel to obtain a target attenuation model and a target matching model; and inputting the effective data into the target attenuation model and / or the target matching model to obtain a measurement result. The application has the effect of improving the measurement accuracy of subcutaneous glucose.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an optimized calibration-free method compatible with multiple subcutaneous glucose sensor outputs. BACKGROUND

[0002] Continuous glucose monitors (CGM) are through the electrode on the glucose oxidase and interstitial fluid glucose in the electrochemical reaction, and then the CGM sensor transmits the current value, through the processing of the current value to realize the conversion of the current value to blood glucose value. In actual use, the actual blood glucose value (glucose value from blood, GB) and the interstitial fluid glucose value (glucose value from interstitial fluid, GIF) measured by the CGM sensor have the problems of deviation, lag, nonlinearity and dynamics, mainly manifested in:

[0003] 1. Time dimension: GIF changes lag behind GB changes, and GIF changes delay or advance GB, which is called drag phenomenon. In the process of blood glucose rising, the delay of GIF increases due to the transfer of glucose from blood to tissue; in the process of blood glucose falling, GIF decreases before GB, which may be due to the influence of insulin on the transfer of glucose from interstitial fluid to surrounding cells.

[0004] 2. Absolute value: there is always an absolute numerical difference between GIF and GB, and the specific difference between GIF and GB may depend on insulin sensitivity and vascularization degree of CGM sensor site, as well as regional blood flow, capillary permeability, adjacent cell metabolic rate and other factors that affect glucose transport.

[0005] 3. CGM sensor error. Due to the differences in device preparation process, enzyme quality, blood glucose calibration technology, etc., the effects are uneven.

[0006] In general, there are errors between actual blood glucose value GB and GIF, and the errors of the sensor, which leads to the decline of the measurement accuracy of CGM sensor, so the measurement and calibration of CGM sensor are particularly important.

[0007] Currently, there are ways to correct errors by finger blood calibration. The disadvantages of finger blood calibration are that it needs to be tested frequently to calibrate data, which is not convenient to use and has poor experience. In addition, there is also a least square method for calibrating CGM sensor, the calibration process is: first, design a nonlinear compensation model according to the characteristics of CGM sensor, then perform temperature compensation, then calibrate the data according to the compensation data, and calculate the relative error: The CGM sensor is calibrated again using the relative error e.

[0008] However, the above error model requires the user to wear the CGM sensor for a long time to obtain a rich training set to ensure the accuracy of the error model. However, under long-term wear, the current value tends to be periodic over time, the irregularity decreases, the sensor sensitivity decreases, and other factors cause the accuracy of the error model to decrease, resulting in inaccurate calibration of the CGM sensor.

[0009] Therefore, the current calibration method cannot meet the calibration requirements of the CGM sensor, and the measurement accuracy of the CGM sensor cannot be guaranteed. SUMMARY

[0010] In order to solve the problem that the measurement accuracy of the subcutaneous glucose sensor cannot be guaranteed, the application provides an optimized calibration-free method compatible with multiple subcutaneous glucose sensor outputs.

[0011] The application provides an optimized calibration-free method compatible with multiple subcutaneous glucose sensor outputs. The method comprises:

[0012] According to the message code, normal conversation message codes, calibration message codes, and background calculation message codes are obtained;

[0013] The normal conversation message codes are input into a first data channel to obtain a target noise model, and effective data is obtained according to the normal conversation message codes and the target noise model;

[0014] The calibration message codes are input into a second data channel, and the background calculation message codes are input into a third data channel to obtain a target attenuation model and a target matching model;

[0015] The effective data is input into the target attenuation model and / or the target matching model to obtain a measurement result.

[0016] By using the above technical solution, first, the application divides the input data into normal conversation message codes, calibration message codes, and background calculation message codes, and then uses different data channels to process the message codes corresponding to the data channels, such as using the first data channel to remove noise in the normal conversation message codes, and using the second data channel to process the calibration message codes and the third data channel to process the background calculation message codes to update and obtain the target attenuation model and the target matching model. Finally, the effective data after removing noise is input into the target attenuation model and / or the target matching model to obtain a measurement result. Therefore, the application covers multiple data processing models, which can optimize the measurement data of different types of CGM sensors, so as to output accurate blood glucose values without any calibration, thereby achieving the purpose of improving the measurement accuracy of the subcutaneous glucose sensor.

[0017] In a possible implementation, the normal conversation message code comprises a current value.

[0018] The inputting of the normal conversation message code into the first data channel to obtain a target noise model comprises: updating an initial noise model by using the current value to obtain the target noise model.

[0019] In a possible implementation, the obtaining of the valid data according to the normal conversation message code and the target noise model comprises:

[0020] The evaluation of the current value by using the target noise model obtains an evaluation result.

[0021] The judging of whether the evaluation result is normal comprises:

[0022] If yes, the target noise model is used to remove noise contained in the current value to obtain the valid data.

[0023] Otherwise, when available data is monitored in a preset monitoring time period, the target noise model is used to remove noise contained in the available data to obtain the valid data, the available data being current value containing noise lower than a noise threshold.

[0024] In a possible implementation, the judging of whether the evaluation result is normal comprises: when the current value continuously exceeds a current threshold for more than n times or the current value sharply increases or sharply decreases, it is determined that the evaluation result is abnormal.

[0025] In a possible implementation, the second data channel and the third data channel have a common end, the common end being configured with an initial deviation model, an initial matching model and an initial attenuation model, the second data channel being configured with a calibration queue on a data segment independent of the third data channel, and the third data channel being configured with a database interface on a data segment independent of the second data channel, the database interface being used to trigger calling data in a database, and the database interface being used to call the database to establish the initial attenuation model when triggered by a background calculation message code.

[0026] The inputting of the calibration message code into the second data channel and the inputting of the background calculation message code into the third data channel to obtain a target attenuation model and a target matching model comprises:

[0027] The calibration queue is used to update the initial deviation model to obtain a target deviation model when the calibration message code is received.

[0028] The initial matching model is updated according to the calibration queue and the target deviation model to obtain a target matching model.

[0029] According to the calibration queue, the target deviation model, and the target matching model, the initial attenuation model is updated to obtain a target attenuation model.

[0030] In a possible implementation, the calibration queue is used to update the initial deviation model to obtain a target deviation model when the calibration message code is received, including:

[0031] The parameters of the initial deviation model are updated by using the data uploaded by the CGM sensor to obtain an initial target deviation model.

[0032] The parameters of the initial target deviation model are further updated by calling the historical data uploaded by the CGM sensor to obtain a final target deviation model.

[0033] In a possible implementation, the initial matching model is updated to obtain a target matching model according to the calibration queue and the target deviation model, including:

[0034] The initial matching model generates a deviation data queue according to the historical data uploaded by the CGM sensor.

[0035] The average value of the deviation data queue is calculated.

[0036] The adjustment parameters of the data uploaded by the CGM sensor are calculated by using the target deviation model.

[0037] The initial matching model is updated by using the average value and the adjustment parameters to obtain a target matching model.

[0038] In a possible implementation, before the effective data is input into the target attenuation model and / or the target matching model to obtain a measurement result, the method further includes:

[0039] It is judged whether the effective data satisfies: input time < 2h, where the input time refers to the time when the effective data is input into the target noise model.

[0040] If yes, the effective data is input into the target attenuation model.

[0041] If no, the effective data is sequentially input into the target matching model and the target attenuation model.

[0042] In summary, the present application includes the following beneficial technical effects:

[0043] Firstly, the input data is divided into normal conversation message code, calibration message code and background calculation message code, and then different data channels are used to process the corresponding message codes, such as using the first data channel to remove noise in the normal conversation message code, using the second data channel to process the calibration message code, and using the third data channel to process the background calculation message code to update and obtain the target attenuation model and the target matching model. Finally, the effective data after removing noise is input into the target attenuation model and / or the target matching model to obtain the measurement result. Therefore, the present application covers multiple data processing models, which can optimize the measurement data of different types of CGM sensors, so as to output accurate blood glucose values without any calibration, thereby achieving the purpose of improving the measurement accuracy of the subcutaneous glucose sensor. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is a flowchart of the compatible multi-subcutaneous glucose sensor output optimization calibration-free method of the embodiment of the present application.

[0045] Figure 2 is a flowchart of the first data channel processing normal conversation message code in the method embodiment of the present application.

[0046] Figure 3 is a flowchart of the initial bias model updating to obtain the target bias model in the method embodiment of the present application.

[0047] Figure 4 is a flowchart of the initial attenuation model updating to obtain the target attenuation model in the method embodiment of the present application.

[0048] Figure 5 is a broken line graph of the relative error in the method embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0050] In order to improve the measurement accuracy of the subcutaneous glucose sensor, the present application constructs a complete processing method of the current value output by the CGM sensor, which covers multiple data processing models and can optimize the measurement data of different types of CGM sensors, so as to output accurate blood glucose values without any calibration.

[0051] Figure 1A flow chart of an optimized calibration-free method compatible with outputs of multiple subcutaneous glucose sensors is shown, and the main flow of the method is described as follows.

[0052] In step S101, normal conversation message codes, calibration message codes and background calculation message codes are obtained according to message codes.

[0053] Data obtained by a user wearing a CGM sensor for a period of time is received, which includes but is not limited to the time at which the CGM sensor starts to measure blood glucose values and the number of consecutive measurement days, the current value measured by the CGM sensor, and the temperature of the measurement environment of the CGM sensor. In this example, the time at which the CGM sensor starts to measure blood glucose values is taken as the message code, and TD is used to represent it.

[0054] After obtaining the data uploaded by the CGM sensor, it is first determined whether the message code satisfies TD<now and now-TD<day threshold, where now refers to the current time, and the day threshold is a preset value. In one specific example, the day threshold is 15 days. It should be noted that when determining whether the message code satisfies the condition, the difference between now and TD is compared with the day threshold, rather than directly comparing the number of consecutive measurement days with the day threshold, because it is necessary to ensure that the data uploaded by the CGM sensor is recent data, and that it is recent and continuous for many days, so as to ensure that the uploaded data can accurately reflect the current blood glucose value of the user.

[0055] When the message code satisfies the above condition, the data uploaded by the CGM sensor is divided into normal conversation message codes, calibration message codes and background calculation message codes. The normal conversation message codes include the time at which the CGM sensor starts to measure blood glucose values, the number of consecutive measurement days, and the current value measured by the CGM sensor. In this application, data channels corresponding to the normal conversation message codes, the calibration message codes and the background calculation message codes are respectively set, and the data channel corresponding to the normal conversation message codes is referred to as the first data channel, the data channel corresponding to the calibration message codes is referred to as the second data channel, and the data channel corresponding to the background calculation message codes is referred to as the third data channel.

[0056] In addition, when the message code does not satisfy the above condition, the measurement is stopped, and the system is exited. The system refers to a computer system for processing the data uploaded by the CGM sensor.

[0057] In step S102, the normal conversation message codes are input into the first data channel to obtain a target noise model, and effective data is obtained according to the normal conversation message codes and the target noise model.

[0058] The initial noise model is configured in the first data channel, and the initial noise model refers to an untrained noise model or an un-updated noise model. In a specific example, the noise model is one of a moving average model (MAM), a median filtering model (MFM), a weighted average model (WAM), a kalman filtering model (KFM), or a combination of multiple models.

[0059] The first data channel is used to process normal session message codes, and the processing process is as shown in Figure 2 First, the initial noise model is updated by using the current value to obtain a target noise model, then the target noise model is used to evaluate the current value to obtain an evaluation result, and then it is determined whether the evaluation result is normal. In a specific example, the abnormal evaluation result includes the following cases: the current value exceeds the current threshold n times continuously, or the current value increases or decreases sharply, for example, the adjacent current values change by more than 5 times, and n can be set according to the actual application scenario. In this example, n = 9.

[0060] When the determination result is that the evaluation result is normal, the target noise model is used to remove the noise contained in the current value to obtain valid data; otherwise, when the determination result is that the evaluation result is abnormal, it is determined whether there is available data in the preset monitoring time period. The available data refers to the current value in which the noise is lower than the noise threshold. If the result of the second determination is that there is available data, the target noise model is used to remove the noise in the available data to obtain valid data, otherwise the session is abnormal, and the measurement optimization is stopped.

[0061] In step S103, the calibration message code is input into the second data channel and the background calculation message code is input into the third data channel to obtain a target attenuation model and a target matching model.

[0062] The second data channel and the third data channel have a common section in which an initial deviation model, an initial matching model and an initial attenuation model are configured. The second data channel is configured with a calibration queue on a data section independent of the third data channel, and the calibration queue is used to store calibration message codes. After the calibration message codes enter the calibration queue, the calibration queue automatically corrects the calibration message codes, for example, by removing the maximum value or the minimum value in the calibration message codes. Then, the data in the calibration queue that remains after the correction is used to update the initial deviation model, the initial matching model and the initial attenuation model. Meanwhile, the third data channel is configured with a database interface on a data section independent of the second data channel, and the database interface can be triggered by a background calculation message code to call data in a database. The database is a set of data that is set in advance to support the establishment of an attenuation model. When the third data channel receives the background calculation message code, the data in the database is called through the database interface to construct an initial attenuation model, and the constructed initial attenuation model is input into the common section of the second data channel and the third data channel.

[0063] The initial deviation model is mainly used to automatically update the target deviation model according to the calibration queue, and the target deviation model is used to process the data uploaded by the CGM sensor. The processing result is further used in the updating process of the initial matching model and the initial attenuation model. As shown in FIG. 4, the updating process of the initial deviation model is as follows: first, the data uploaded by the CGM sensor is calculated and processed to update the parameters of the initial deviation model to obtain a target deviation model, and then the target deviation model is used to participate in the subsequent process of updating the initial matching model. At the same time, when updating the initial deviation model, the historical data uploaded by the CGM sensor also needs to be added to further update the parameters of the target deviation model, so as to further correct the obtained target deviation model, and finally output the parameters of the target deviation model. Figure 3

[0064] The initial matching model generates a deviation data queue according to the historical data uploaded by the CGM sensor, and in each updating process, the average value of the deviation data queue is calculated, which is used as the parameter for updating the initial matching model. At the same time, the target deviation model is used to calculate the data uploaded by the CGM sensor to obtain an adjustment parameter, and then the adjustment parameter is used to further update the initial matching model to obtain a target matching model. Finally, the target matching model is used to participate in the calculation of the upper and lower bounds of the blood glucose value. The upper bound of the blood glucose value refers to the maximum value of the blood glucose value, and the lower bound of the blood glucose value refers to the minimum value of the blood glucose value. The range between the minimum value and the maximum value is the ideal range of the blood glucose value. In addition, the target matching model is used to calculate the data uploaded by the CGM sensor, and the calculation result is fed back to the target deviation model for further updating of the target deviation model. At the same time, the calculation result is also transmitted to the initial attenuation model to support the updating of the initial attenuation model.

[0065] ​The initial attenuation model receives the calibration queue, the parameters of the target deviation model and the parameters of the target matching model to realize automatic updating to obtain a target attenuation model, which is mainly used to calculate an iterative attenuation value. In the process of calculating the iterative attenuation value, not only are the parameters of the target attenuation model updated, but also the target deviation algorithm and the target matching model are further called to update and calculate the result of calculating the data uploaded by the CGM sensor. In the process of continuous iterative calculation, trend disturbance counting is also performed, and different coefficients are used to further update the parameters of the target attenuation model. The updating process of the initial attenuation model is as shown in Figure 4

[0066] It should be noted that when the data uploaded by the CGM sensor is input, the iterative attenuation value is calculated, and it is judged whether the iterative attenuation value exceeds a critical value. If yes, the system enters a state conversation abnormal mode. The system here also refers to a computer system for processing the data uploaded by the CGM sensor. Otherwise, the system will enter the next step.

[0067] In step S104, the effective data is input into the target attenuation model and / or the target matching model to obtain a measurement result.

[0068] For the effective data obtained in step S102, it is first judged whether the effective data satisfies: input time < 2h, the input time refers to the time when the effective data is input into the computer system of the present application, specifically the time when the effective data is input into the target noise model. When the effective data satisfies the above condition, the effective data is input into the target attenuation model, and the corrected blood glucose value, the blood glucose trend value, the algorithm state, the noise evaluation, the blood glucose upper and lower limit, the predicted blood glucose value and other information are calculated by the target attenuation model according to the already trained parameters. When the effective data does not satisfy the above condition, the effective data is input into the target matching model, the target matching model is updated using the effective data, and then the target attenuation model is updated again by the target matching model, and then the effective data is processed by the target attenuation model to obtain the corrected blood glucose value, the blood glucose trend value, the algorithm state, the noise evaluation, the blood glucose upper and lower limit, the predicted blood glucose value and other information.

[0069] The present application takes the corrected blood glucose value, the blood glucose trend value, the algorithm state, the noise evaluation, the blood glucose upper and lower limit, the predicted blood glucose value and other information calculated as the measurement result, which is output to the display for display on one hand. The display interface of the measurement result is as follows:

[0070] Struct LastStepRes {

[0071] Uint16 Ls1;

[0072] Uint32 Lf1;

[0073] Uint8 Lb1;​

[0074] Uint8 Lb2;

[0075] Uint32 Time;

[0076] Uint16 Iv1;

[0077] Uint16 lv2;

[0078] Uint16 lv3;

[0079] Uint16 Iv4;

[0080] Uint8 Ib3;

[0081] Uint16 pre;

[0082] }

[0083] The initialization value of Uint16 Ls1 is 1, indicating a corrected blood glucose value, the initialization value of Uint32 Lf1 is 127, indicating a blood glucose trend value, and the absolute value of the trend value is less than 8 under normal circumstances, the initialization value of Uint8 Lb1 is 1, indicating the running state of the algorithm model for processing the blood glucose value, the initialization value of Uint8 Lb2 is 5, indicating the evaluation state of the noise model, Uint32 Time represents the current time, the initialization value of Uint16 lv1 is 600, indicating the calculated upper limit of blood glucose, the initialization value of Uint16 lv2 is 20, indicating the calculated lower limit of blood glucose, the initialization value of Uint16 lv3 is 20, which is undefined, the initialization value of Uint16 lv4 is 600, which is undefined, the initialization value of Uint8 lb3 is 0, indicating the measurement result state identifier, and the initialization value of Uint16 pre is 13, indicating the predicted blood glucose value.

[0084] Therefore, by displaying the measurement result, the user can conveniently view the data measured by the CGM sensor.

[0085] On the other hand, the computer system also saves the measurement result of this time, so as to accumulate the measurement result for subsequent prediction of the blood glucose value.

[0086] In order to verify that the accuracy of the measurement result obtained by the above method is improved, the relative error and the average relative error are calculated according to the measurement result, and the specific calculation formula is:

[0087] The calculation formula of the relative error is shown in formula 1:

[0088] (1)

[0089] The calculation formula of the average relative error is shown as formula 2:

[0090] (2)

[0091] Wherein, GBn represents the actual blood glucose value at the nth measurement, GIFn represents the blood glucose value contained in the measurement result at the nth measurement.

[0092] It should be noted that the above relative error and average relative error are both used as an index for judging the measurement accuracy of the CGM sensor, and the values of MARD1 and MARD2 both represent the error between the actual blood glucose value GB and the blood glucose value GIF measured by the CGM sensor, which is used to measure the accuracy of the measurement result measured by the CGM sensor, and the smaller the values of MARD1 and MARD2 are, the higher the accuracy is.

[0093] The following is a specific example of calculating the actual blood glucose value GB and the blood glucose value GIF measured by the CGM sensor:

[0094] First, the above measurement optimization method is applied to a CGM sensor on the market, and the relative error of the blood glucose value at different time points obtained in 4 days is shown in the following table. The abscissa is the number of times of actual blood glucose test using finger blood calibration, and the ordinate is the relative error. Data A is the relative error between the blood glucose value obtained by using the measurement optimization method of the present application and the actual blood glucose value, wherein the average relative error MARD(A): 11.78%; Data B is the relative error between the blood glucose value measured without using the measurement optimization method of the present application and the actual blood glucose value, and the average relative error MARD(B): 20.24%, as shown in the following table. Figure 5

[0095] As can be seen from the above table, the measurement optimization method of the present application can output more accurate blood glucose values without any calibration by the user when the ordinary sensor needs to be calibrated by finger blood, and the accuracy is not lower than that of the CGM sensor without using the calibration algorithm. Figure 5 Therefore, the optimization calibration-free method of the present application compatible with multiple subcutaneous glucose sensor outputs can be applied to any CGM sensor as needed. In the present example, the calibration-free algorithm of the present application is compatible with at least more than 3 different types of CGM sensors, which can also be considered as compatible with multiple types of CGM sensors, and the current frequency output by the multiple types of CGM sensors can be different, for example, the current output frequency can be 3 min, 5 min, 15 min, and the CGM sensor with a current output frequency of 15 min can output a blood glucose value not lower than 3 min after being calibrated by the calibration-free algorithm of the present application.

[0096]

[0097] ​​In addition, after the electrode of the CGM sensor collects the current value of the human tissue, the calibration-free algorithm of the present application is used for optimization, specifically, outputting one current value per minute, or outputting more closely current values, so as to facilitate continuous analysis and improve the accuracy of the subsequent measurement results. Moreover, due to the characteristics that the current values of different human tissues are different, the output blood glucose values are maximized after the calibration-free algorithm of the present application, which facilitates the analysis of users with similar current values but large differences in actual blood glucose values, so as to timely find the population with high blood glucose values, and further dynamically update the calibration-free algorithm of the present application, improve the subsequent optimization accuracy, so that the CGM sensor originally requiring fingertip blood calibration can also abandon fingertip blood calibration and rely on the calibration-free algorithm of the present application for optimization, to realize more accurate blood glucose value output under the condition of calibration-free.

[0098] In order to better implement the above method, the present application further provides an optimization calibration-free device compatible with multiple subcutaneous glucose sensor outputs, which comprises a memory and a processor.

[0099] The memory can be used to store instructions, programs, codes, code sets or instruction sets. The memory can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function, and instructions for implementing the above-mentioned optimization calibration-free method compatible with multiple subcutaneous glucose sensor outputs, etc.; the data storage area can store data involved in the above-mentioned optimization calibration-free method compatible with multiple subcutaneous glucose sensor outputs, etc.

[0100] The processor can include one or more processing cores. The processor calls the data stored in the memory by running or executing the instructions, program, code set or instruction set stored in the memory, and performs various functions and processes data of the present application. The processor can be at least one of an application specific integrated circuit, a digital signal processor, a digital signal processing device, a programmable logic device, a field programmable gate array, a central processing unit, a controller, a microcontroller and a microprocessor. It can be understood that for different devices, the electronic devices used to implement the functions of the above-mentioned processor can also be other, and the embodiments of the present application are not limited specifically.

[0101] The present application further provides a computer readable storage medium, for example, various media that can store program codes such as U disk, mobile hard disk, read only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc. The computer readable storage medium stores a computer program capable of being loaded by the processor and executing the above-mentioned optimization calibration-free method compatible with multiple subcutaneous glucose sensor outputs.

[0102] The above description is merely exemplary of the application and the principles thereof. It is to be understood that those skilled in the art will be able to devise various arrangements which, although not explicitly described or shown herein, embody the principles of the application and are included within its spirit and scope. Furthermore, there are several variations to the application described herein which have not been described but will be understood by those skilled in the art. For example, the features of the application described and shown can be combined with other features of the application described and shown (but not limited to) in the patent specification and drawings.

Claims

1. An optimized calibration-free method compatible with multiple subcutaneous glucose sensor outputs, comprising: The method comprises the following steps: Obtaining normal conversation message code, calibration message code and background calculation message code according to message code; Inputting the normal conversation message code into a first data channel to obtain a target noise model, and obtaining effective data according to the normal conversation message code and the target noise model; Inputting the calibration message code into a second data channel and inputting the background calculation message code into a third data channel to obtain a target attenuation model and a target matching model; Inputting the effective data into the target attenuation model and / or the target matching model to obtain a measurement result; The normal conversation message code comprises a current value; The step of inputting the normal conversation message code into a first data channel to obtain a target noise model comprises the following steps: Updating an initial noise model by using the current value to obtain a target noise model; The step of obtaining effective data according to the normal conversation message code and the target noise model comprises the following steps: Evaluating the current value by using the target noise model to obtain an evaluation result; Judging whether the evaluation result is normal; If yes, removing the noise contained in the current value by using the target noise model to obtain effective data; Otherwise, when available data is monitored within a preset monitoring time period, removing the noise contained in the available data by using the target noise model to obtain effective data, wherein the available data refers to the current value whose noise is lower than a noise threshold; The step of judging whether the evaluation result is normal comprises the following steps: When the current value exceeds a current threshold for n times or more or the current value increases or decreases sharply, it is judged that the evaluation result is abnormal; The second data channel and the third data channel have a common end, and the common end is configured with an initial bias model, an initial matching model and an initial attenuation model, the second data channel is configured with a calibration queue on a data segment independent of the third data channel, and the third data channel is configured with a database interface on a data segment independent of the second data channel, the database interface is used to trigger the data in the database, and the database interface is used to call the data in the database to establish the initial attenuation model when triggered by the background calculation message code; The step of inputting the calibration message code into a second data channel and inputting the background calculation message code into a third data channel to obtain a target attenuation model and a target matching model comprises the following steps: The calibration queue is used to update the initial bias model to obtain a target bias model when the calibration message code is received; 2. The optimized calibration-free method compatible with outputs of multiple subcutaneous glucose sensors of claim 1, wherein, The initial matching model is updated according to the calibration queue and the target bias model to obtain a target matching model; The initial attenuation model is updated according to the calibration queue, the target bias model and the target matching model to obtain a target attenuation model. The step of updating the initial bias model to obtain a target bias model when the calibration message code is received comprises the following steps: The parameters of the initial bias model are updated by using the data uploaded by the CGM sensor to obtain an initial target bias model; The parameters of the initial target bias model are further updated by calling the historical data uploaded by the CGM sensor to obtain a final target bias model.

3. The optimized calibration-free method compatible with outputs of multiple subcutaneous glucose sensors of claim 1, wherein, The updating of the initial matching model according to the calibration queue and the target bias model comprises: The initial matching model generates a bias data queue according to historical data uploaded by a CGM sensor; An average value of the bias data queue is calculated; An adjustment parameter is calculated by using the target bias model and data uploaded by the CGM sensor; The initial matching model is updated by using the average value and the adjustment parameter to obtain a target matching model.

4. The optimized calibration-free method compatible with outputs of multiple subcutaneous glucose sensors of claim 1, wherein, Before the effective data is input into the target attenuation model and / or the target matching model to obtain a measurement result, the method further comprises: It is judged whether the effective data satisfies: input time < 2h, wherein the input time refers to the time when the effective data is input into the target noise model; If yes, the effective data is input into the target attenuation model; If no, the effective data is input into the target matching model and the target attenuation model in sequence.

Images