Blood glucose control system and control method based on closed-loop technology

Abstract

The invention discloses a blood glucose control system and control method based on a closed-loop technology, the blood glucose control system comprises a dynamic blood glucose monitoring system, an insulin pump and a central algorithm control system, the central algorithm control system comprises a storage recording module, a database access module and an algorithm module, the database access module outputs an insulin injection unit A1 or a carbon water intake amount B1, and the database access module outputs the insulin injection unit A1 or the carbon water intake amount B1; the algorithm module outputs an insulin injection unit A2 or a carbon water intake amount B2, and the database access module and the algorithm module output are combined to carry out optimization calculation on the algorithm module to obtain a final insulin injection unit A3 or a final carbon water intake amount B3. According to the blood glucose control system and control method based on the closed-loop technology, through optimization processing of the insulin injection unit A1 or the carbon water intake amount B1 output by the database access module and the insulin injection unit A2 or the carbon water intake amount B2 output by the algorithm module, the output result is more accurate and reliable, so that the blood glucose of a user is efficiently adjusted, and the user experience is improved. And user life quality is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial pancreas, and in particular to a blood sugar control system and a control method based on closed-loop technology. Background Art

[0002] The artificial pancreas closed-loop system is an automatic insulin delivery device (AID) that can automatically infuse basal insulin to make the blood sugar of diabetic users more stable at night. The system consists of three elements, including a dynamic blood glucose monitoring system, an insulin pump, and a central algorithm control system, so that the system can determine the insulin infusion dose based on real-time dynamic blood glucose monitoring results.

[0003] The normal human pancreas secretes a small amount of insulin about every 4 minutes, and secretes a large amount of insulin after a meal to control the fluctuation of blood sugar in the human body within the physiological range throughout the day. However, injection of insulin: a large dose of insulin is injected once before each meal, which has poor absorption effect and is difficult to stably control the blood sugar level of the human body, which is easy to cause high blood sugar at night.

[0004] Insulin pumps simulate the normal pancreatic secretion mode to infuse insulin. There are two methods: basal rate and bolus before meals. The basal rate is a 24-hour continuous micro-infusion of insulin to control blood sugar stability between meals and at night; the bolus before meals is an infusion of insulin for meal food to control blood sugar stability after meals. The bolus can be flexibly set according to meal time, meal size and food type.

[0005] In the artificial pancreas system, how to achieve accurate dynamic blood glucose monitoring is the core problem that needs to be solved at present. The control of the infusion volume of the insulin pump directly affects the diabetes control and quality of life of the artificial pancreas users. Summary of the invention

[0006] To solve the above problems, the present invention provides a blood sugar control system and control method based on closed-loop technology, which combines the oref1 algorithm with the big data of the database, optimizes the output of the oref1 algorithm, and thus efficiently regulates the user's blood sugar and ensures the user's quality of life.

[0007] To achieve the above object, the present invention provides a blood glucose control system based on closed-loop technology, including a dynamic blood glucose monitoring system, an insulin pump and a central algorithm control system.

[0008] A dynamic glucose monitoring system that continuously monitors glucose levels in the subcutaneous interstitial fluid and transmits glucose levels to a central algorithm control system;

[0009] The central algorithm control system includes a storage and recording module, a database access module and an algorithm module. The database access module outputs the injected insulin unit A1 or the carbohydrate intake B1, and the algorithm module outputs the injected insulin unit A2 or the carbohydrate intake B2. The database access module and the algorithm module output are combined to optimize the algorithm module to obtain the final injected insulin unit A3 or the carbohydrate intake B3.

[0010] The insulin pump is connected to the central algorithm control system and performs insulin infusion according to the final injected insulin unit A3 issued by the central algorithm control system.

[0011] Preferably, the algorithm module of the central algorithm control system includes oref1, and oref1 includes SMB, UAM, and ZTpredBG.

[0012] Preferably, the database access module can access a database including a database of food carbohydrate content, and a database of blood sugar variation factors in the human body under sports events and time.

[0013] Preferably, there is at least one central algorithm control system, which is installed on two or more devices, so that the user and his or her guardian can instantly know the user's blood sugar information; through password verification, the guardian can remotely control the start, pause, dosage and change of the basal rate of the insulin pump.

[0014] A control method of the blood sugar control system based on the closed-loop technology as described above, which is controlled by a central algorithm control system, comprises:

[0015] S1. Set the basal rate mode and micro-bolus mode;

[0016] S2. Before the user takes in or consumes carbohydrates, he / she inputs the corresponding items into the central algorithm control system. The database access module outputs the injected insulin unit A1 or the amount of carbohydrates ingested B1 by accessing the database. The oref1 algorithm outputs the injected insulin unit A2 or the amount of carbohydrates ingested B2. The central algorithm control system performs optimization calculation based on A1, A2 or B1, B2 to obtain the final insulin infusion amount A3 or the amount of carbohydrates ingested B3.

[0017] S3: Before the user consumes carbohydrates, the micro-high-dose mode is activated;

[0018] S4. After the user consumes carbohydrates, the basal rate mode is automatically turned on by detecting changes in insulin in the user's body.

[0019] Preferably, in step S1, the basic rate mode operation process is:

[0020] Set a safety threshold, safety threshold = blood sugar target lower limit - 0.5 * blood sugar lower limit - 0.5 * 40, where the blood sugar target unit is mmol / L;

[0021] Predict the user's blood sugar and obtain predicted blood sugar at multiple time points;

[0022] Initiate temporary basal rate until minimum predicted blood glucose is above the safety threshold;

[0023] Blood sugar is regulated by infusing insulin at a temporary basal rate.

[0024] Preferably, in step S1, the micro-high-dose mode operation process is:

[0025] Predict the user's blood sugar and get the current additional insulin;

[0026] Insulin is infused as a bolus using a basal rate insulin dose for a future time, where the time period for the future time is set to 0.

[0027] Preferably, the method for the central algorithm control system to perform optimization calculation in combination with A1 and A2 includes:

[0028] (1) When A1=A2, directly output A3=A1=A2;

[0029] (2) When A1≠A2, the central algorithm control system takes the arithmetic mean of A1 and A2

[0030] Will Substitute into oref1 algorithm for parameter optimization and output the injected insulin unit A2;

[0031] (3) If Then output

[0032] like Then repeat step (2) until the arithmetic mean of the input is Equal to the output injected insulin unit A2.

[0033] Preferably, the method for the central algorithm control system to perform optimization calculation in combination with B1 and B2 includes:

[0034] S1, when B1=B2, directly output B3=B1=B2;

[0035] S2. When B1≠B2, the central algorithm control system takes the arithmetic mean of B1 and B2. Will Substitute into oref1 algorithm for parameter optimization and output the carbohydrate intake B2;

[0036] S3. If then output

[0037] If then repeat step S2 again until the input arithmetic mean is equal to the output ingested carbohydrate amount B2.

[0038] Preferably, a user - personalized DNN model is established based on the user's carbohydrate intake, insulin injection amount, and blood glucose change situation to optimize the method of step S2, specifically including:

[0039] (1) Continuously add the user's new carbohydrate intake, insulin injection amount, and blood glucose change situation data to the training set for incremental learning of the DNN model;

[0040] (2) During the incremental learning process, fix the first - layer network, release the second - layer and third - layer networks for training to obtain a new DNN model;

[0041] (3) The database access module no longer outputs the injected insulin unit A1 or the ingested carbohydrate amount B1. The database access module only outputs the theoretical carbohydrate absorption amount or carbohydrate consumption amount. The new DNN model outputs the user - personalized carbohydrate absorption amount or carbohydrate consumption amount, and further outputs an insulin injection amount or insulin intake amount. At the same time, the new DNN model outputs a predicted blood glucose value C1;

[0042] (4) Compare the predicted blood glucose reduction value C1 output by the new DNN model with the target blood glucose C2 set by the oref1 algorithm:

[0043] If C1 = C2, output the insulin injection amount or carbohydrate intake amount output by the new DNN model;

[0044] If C1 > C2, output the insulin injection amount or carbohydrate intake amount output by the new DNN model, and add the user's blood glucose change situation to the training set to continue training the model;

[0045] If C1 < C2, output the insulin injection amount or carbohydrate intake amount output by the oref1 algorithm, and add the user's blood glucose change situation to the training set to continue training the model.

[0046] A blood glucose control system and control method based on closed - loop technology of the present invention have the following technical effects:

[0047] (1) Through the optimization process of the database access module outputting the injected insulin unit A1 or the ingested carbohydrate amount B1, and the algorithm module outputting the injected insulin unit A2 or the ingested carbohydrate amount B2, the output results are more accurate and reliable;

[0048] (2) A DNN model is trained using the user's carbohydrate intake, insulin injection volume, and blood sugar changes. By continuously inputting new user data, the DNN model is trained on a personalized basis, making the closed-loop system more consistent with the user's physical changes and reflecting the user's diabetes development to a certain extent. The personalized model design further ensures the user's quality of life.

[0049] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0051] Figure 1 This is a schematic diagram of the connection relationship of a blood sugar control system based on closed-loop technology of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical scheme and advantages disclosed in the embodiments of the present invention more clearly understood, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0053] A blood glucose control system based on closed-loop technology, such as Figure 1 As shown, it includes a dynamic blood glucose monitoring system, an insulin pump and a central algorithm control system. Among them, the sensor probe of the dynamic blood glucose monitoring system is implanted on the body surface to monitor the glucose level of the interstitial fluid of the subcutaneous tissue and transmit the glucose level to the central algorithm control system to achieve real-time blood glucose monitoring.

[0054] The insulin pump contains the necessary mechanical structure for insulin infusion and is controlled by the central algorithm control system. According to the current insulin infusion volume data sent by the central algorithm control system, the insulin pump infuses the current required amount of insulin into the user's body. At the same time, the infusion status of the insulin pump can also be fed back to the central algorithm control system in real time to monitor whether the insulin pump is in normal working condition.

[0055] The central algorithm control system has the functions of storing, recording and accessing the database. Storage and recording means that the user's real-time blood sugar data and insulin infusion data can be stored, which is convenient for intuitively analyzing the changes in the user's blood sugar level and insulin infusion volume within a certain period of time, so as to more intuitively observe the user's condition.

[0056] Accessing a database means being able to access a database of food carbohydrate content or a database of factors that affect blood sugar changes in the human body, such as exercise intensity and exercise time.

[0057] These databases are often models built after training with a large amount of data. They are universal and the output value of the model can be obtained after inputting the corresponding parameters.

[0058] In one embodiment of the present invention, the user inputs information such as food type and weight into the central algorithm control system before eating, and the central algorithm control system accesses the food library big data to perform auxiliary calculations on the carbohydrates and insulin of the user's food.

[0059] In one embodiment of the present invention, the user inputs information such as exercise items and exercise time into the central algorithm control system before exercising. The central algorithm control system accesses the exercise database big data to assist in calculating the user's carbohydrate consumption. When the central algorithm control system prompts the intake of carbohydrates, it directly prompts the customer through the food library big data which foods the customer can eat and the quality of the food to prevent hypoglycemia.

[0060] The calculation formula is as follows:

[0061]

[0062] ISF is insulin sensitivity (mmol / L / U), which refers to the amount of blood sugar drop caused by injecting 1U of insulin.

[0063] IC is the ratio of insulin to carbohydrates (carbohydrate index) (g / U), that is, the ratio of each unit of insulin to cover each gram of carbohydrates.

[0064] The algorithm of the central algorithm control system is oref1, the core of which includes SMB and UAM.

[0065] SMB stands for super micro bolus, which means borrowing the future basal rate insulin to achieve immediate drug administration through the bolus function.

[0066] In one embodiment of the present invention, by setting the basal rate mode and the micro-bolus mode of SMB, the micro-bolus mode can be selected to infuse insulin during the blood sugar increase stage (for example, after a meal), thereby maintaining the blood sugar of diabetic patients within a normal range.

[0067] When SMB is enabled, UAM, or UnAnnounced Meals, is enabled, which is a way for the algorithm to identify carbohydrate absorption and effectively handle blood sugar spikes caused by less carbohydrate input than actually absorbed, or blood sugar spikes caused by hormonal changes in the body. UAW can provide automatic correction boluses to help prevent high blood sugar.

[0068] UAM can limit the amount of SMB injected during meals to reduce the possibility and severity of hypoglycemia. When the amount of carbohydrates ingested exceeds the actual amount absorbed, or other unexpected events occur, it only allows SMB injection when blood sugar rises to a certain level, thereby preventing the full impact on blood sugar.

[0069] When the UAM feature is enabled (recommended by default), any rise in blood glucose relative to the prediction for insulin alone triggers the creation of a blood glucose dataset based on the UAM predictions. This dataset projects the UAM deviations into the future for a full DIA cycle (Duration of Insulin Action, set in the preferences).

[0070] If the current deviation is higher than any deviation seen in the past approximately 45 minutes, then the future deviation is forecast to slowly decrease from the current level at a rate proportional to the rate of increase.

[0071] Once the current deviation begins to fall below the deviation from 15-30 minutes (or 30-60 minutes ago), future deviations are projected to fall at the same rate of decline since the most recent peak until they reach zero, or up to DIA hours if they are not predicted to reach zero. If this brings the predicted blood sugar value within range, the algorithm deems that no additional insulin injection is needed.

[0072] If the minimum predicted blood glucose level during that time frame is above the maximum target blood glucose, the algorithm deems insulin necessary and safe, and delivers it based on the final predicted blood glucose. However, if the minimum predicted blood glucose level drops too low, the temporary basal rate is set to 0, and no more insulin is delivered until the minimum predicted blood glucose level returns to a safe range.

[0073] The algorithm uses UAM to generate a series of predicted blood glucose value data sets, ZTpredBG is the "other" prediction seen in NightScout. It shows how long it would take for blood glucose to reach / above target if the deviation suddenly stopped and ran 0 temporary basal rate before that. This value is used to regulate UAM and prevent the algorithm from injecting too much insulin when running under most UAM-driven predictions.

[0074] The application of ZTpredBG can effectively reduce the occurrence of hypoglycemia caused by excessive insulin delivery, exercise or drinking. When the algorithm determines that hypoglycemia cannot be prevented by reducing insulin delivery, it will prompt the user to consume carbohydrates.

[0075] In one embodiment of the present invention, the central algorithm control system is connected to multiple devices. In addition to the controller carried by the user, the guardian can remotely obtain the user's blood sugar information. Through password verification, the guardian can remotely control the insulin pump to start, pause, administer large doses, and change the basal rate.

[0076] In one embodiment of the present invention, the control method of the central algorithm control system of the present invention includes:

[0077] S1: Set SMB basal rate mode and micro-bolus mode.

[0078] (1) The operation process of the basic rate model is as follows:

[0079] Set a safety threshold, safety threshold = blood sugar target lower limit - 0.5 * blood sugar lower limit - 0.5 * 40, where the blood sugar target unit is mmol / L;

[0080] Predict the user's blood sugar and obtain predicted blood sugar at multiple time points;

[0081] Initiate temporary basal rate until minimum predicted blood glucose is above the safety threshold;

[0082] Blood sugar is regulated by infusing insulin at a temporary basal rate.

[0083] (2) The operation process of micro-high-dose mode is as follows:

[0084] Predict the user's blood sugar and get the current additional insulin;

[0085] Insulin is infused as a bolus using a future time basal rate insulin dose, where the future time period is set to 0.

[0086] S2. Before the user takes in / consumes carbohydrates, he / she inputs the corresponding items into the central algorithm control system. The central algorithm control system outputs the injected insulin unit A1 or the carbohydrate intake amount B1 by accessing the database. Meanwhile, the central algorithm control system oref1 algorithm outputs the injected insulin unit A2 or the carbohydrate intake amount B2. The central algorithm control system performs optimization calculation based on A1 / A2 or B1 / B2 to obtain the final insulin infusion amount A3 or carbohydrate intake amount B3.

[0087] In one embodiment of the present invention, before the user takes in carbohydrates, he inputs parameters such as food type and food intake into the central algorithm control system, and the central algorithm control system accesses the food carbohydrate content database to obtain the corresponding carbohydrate intake, and then calculates the insulin unit A1 that the user needs to inject. At this time, the SMB, UAM and other modules in the oref1 algorithm of the central algorithm control system output an injected insulin unit A2.

[0088] When A1=A2, directly output A3=A1=A2;

[0089] When A1≠A2, the central algorithm control system takes the arithmetic mean of A1 and A2 Will Substitute into oref1 algorithm for parameter optimization and output the injected insulin unit A2;

[0090] like Then output

[0091] like or Then take it again The arithmetic mean of A2 repeats the above steps of oref1 algorithm parameter optimization until the arithmetic mean of the input is equal to the output injected insulin units.

[0092] In one embodiment of the present invention, before exercising, the user inputs parameters such as exercise items and exercise time into the central algorithm control system. The central algorithm control system accesses the database of exercise carbohydrate consumption and outputs an intake of carbohydrate amount B1. Due to differences in physical fitness and exercise intensity among different people, the output value may also have deviations.

[0093] At this time, the SMB, UAM and other modules in the central algorithm control system oref1 algorithm output a carbohydrate intake amount B2.

[0094] When B1=B2, directly output B3=B1=B2;

[0095] When B1≠B2, the central algorithm control system takes the arithmetic mean of B1 and B2. Will Substitute into oref1 algorithm for parameter optimization and output the carbohydrate intake B2;

[0096] like Then output

[0097] like Then take the arithmetic mean of B1 and B2 again and repeat the above steps of oref1 algorithm parameter optimization until the input arithmetic mean is equal to the output carbohydrate intake.

[0098] S3: Before the user consumes carbohydrates, the micro-high-dose mode is activated;

[0099] S4. After the user consumes carbohydrates, the basal rate mode is automatically turned on by detecting the changes in insulin in the user's body. By setting the basal rate mode, the user's nighttime blood sugar can be effectively controlled to avoid nighttime hypoglycemia.

[0100] In one embodiment of the present invention, the control method of the central algorithm control system of the present invention also includes establishing a user-personalized DNN model: the central algorithm control system learns the carbon conversion amount of different users based on the dynamic blood glucose monitoring system, carbon water intake and insulin pump injection amount. By analyzing the actual carbon water absorption amount and absorption time of the user after eating, and mastering the carbon water and absorption rate of the food through continuous practice, a related model based on carbon water intake-carbon water absorption (blood sugar increase)-insulin injection-blood sugar reduction for different users is established to help make better meal decisions next time.

[0101] During the model building process, new data on users' carbohydrate intake, insulin injections, and blood sugar changes are continuously added to the training set, and different weights are reassigned to the data to perform incremental learning on the model.

[0102] Reload the saved model structure and parameters, and retrain the new training set. In order to retain the generalization ability of the model, fix the first layer of the network with the largest number of parameters, release the second and third layers of the network for training, and obtain a new model.

[0103] The first layer model has the largest number of parameters, so retaining the first layer network can ensure that the model's fitting performance for the old training set data set does not weaken. At the same time, the second and third layer network parameters are small and can fit the user's new data well. This method can ensure that the model does not weaken the generalization performance of the overall model in the process of fitting new data.

[0104] After a complete user-personalized model is established, the food library data no longer outputs the injected insulin unit A1, but only outputs the corresponding carbohydrate content of the food. The user's carbohydrate absorption amount is output through the user-personalized model, and further outputs an insulin injection amount.

[0105] At the same time, the model outputs a predicted blood sugar value C1, and compares the model's prediction result with the target blood sugar C2 set by the oref1 algorithm:

[0106] If C1=C2, the amount of insulin injected output by the model is directly output;

[0107] If C1>C2, the insulin injection amount output by the user's personalized model is used as the final injection amount, and the user's blood sugar changes are added to the training set to continue training the model;

[0108] If C1 < C2, the insulin injection amount output by the oref1 algorithm is used as the final injection amount, and the user's blood glucose change situation is added to the training set to continue training the model.

[0109] Therefore, for a blood glucose control system and control method based on closed-loop technology according to the present invention, through the optimization process of the injection insulin unit A1 or the ingested carbohydrate amount B1 output by the database access module and the injection insulin unit A2 or the ingested carbohydrate amount B2 output by the algorithm module, the output result is made more accurate and reliable, so as to efficiently regulate the user's blood glucose. At the same time, a DNN model is trained using the user's carbohydrate intake, insulin injection amount, and blood glucose change situation. By continuously inputting new user data, personalized training of the user for the DNN model is carried out, so that the closed-loop system better conforms to the user's physical changes, can reflect the development of the user's diabetes to a certain extent, and the personalized model design further ensures the user's quality of life.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements do not make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A blood glucose control system based on closed-loop technology, including a dynamic blood glucose monitoring system, an insulin pump and a central algorithm control system, characterized in that: A dynamic glucose monitoring system that continuously monitors glucose levels in the subcutaneous interstitial fluid and transmits glucose levels to a central algorithm control system; The central algorithm control system includes a storage and recording module, a database access module and an algorithm module. The database access module outputs the injected insulin unit A1 or the carbohydrate intake B1, and the algorithm module outputs the injected insulin unit A2 or the carbohydrate intake B2. The database access module and the algorithm module output are combined to optimize the algorithm module to obtain the final injected insulin unit A3 or the carbohydrate intake B3. The insulin pump is connected to the central algorithm control system and performs insulin infusion according to the final injected insulin unit A3 issued by the central algorithm control system.

2. A blood sugar control system based on closed-loop technology according to claim 1, characterized in that: The algorithm modules of the central algorithm control system include oref1, and oref1 includes SMB, UAM, and ZTpredBG.

3. A blood sugar control system based on closed-loop technology according to claim 1, characterized in that: The database access module can access a database including the carbohydrate content of food, and a database of blood sugar variation factors in the human body under sports events and time.

4. A blood sugar control system based on closed-loop technology according to claim 2, characterized in that: There is at least one central algorithm control system, which is installed on two or more devices, so that the user and his / her guardian can know the user's blood sugar information in real time; Through password verification, the guardian can remotely control the insulin pump to start, pause, administer doses, and change the basal rate.

5. A control method for a blood sugar control system based on a closed-loop technology as claimed in claim 1, characterized in that: Controlled by a central algorithm control system, including: S1. Set the basal rate mode and micro-bolus mode; S2. Before the user takes in or consumes carbohydrates, he / she inputs the corresponding items into the central algorithm control system. The database access module outputs the injected insulin unit A1 or the amount of carbohydrates ingested B1 by accessing the database. The oref1 algorithm outputs the injected insulin unit A2 or the amount of carbohydrates ingested B2. The central algorithm control system performs optimization calculation based on A1, A2 or B1, B2 to obtain the final insulin infusion amount A3 or the amount of carbohydrates ingested B3. S3: Before the user consumes carbohydrates, the micro-high-dose mode is activated; S4. After the user consumes carbohydrates, the basal rate mode is automatically turned on by detecting changes in insulin in the user's body.

6. The control method of a blood sugar control system based on closed-loop technology according to claim 5, characterized in that: In step S1, the basic rate mode operation process is: Set a safety threshold, safety threshold = blood sugar target lower limit - 0.5 * blood sugar lower limit - 0.5 * 40, where the blood sugar target unit is mmol / L; Predict the user's blood sugar and obtain predicted blood sugar at multiple time points; Initiate temporary basal rate until minimum predicted blood glucose is above the safety threshold; Blood sugar is regulated by infusing insulin at a temporary basal rate.

7. The control method of a blood sugar control system based on closed-loop technology according to claim 5, characterized in that: In step S1, the micro-high-dose mode operation process is: Predict the user's blood sugar and get the current additional insulin; Insulin is infused as a bolus using a future time basal rate insulin dose, where the future time period is set to 0.

8. The control method of a blood sugar control system based on closed-loop technology according to claim 5, characterized in that: The method of optimizing calculation by the central algorithm control system in combination with A1 and A2 includes: (1) When A1=A2, directly output A3=A1=A2; (2) When A1≠A2, the central algorithm control system takes the arithmetic mean of A1 and A2 Will Substitute into oref1 algorithm for parameter optimization and output the injected insulin unit A2; (3) If Then output like Then repeat step (2) until the arithmetic mean of the input is Equal to the output injected insulin unit A2.

9. The control method of a blood sugar control system based on closed-loop technology according to claim 5, characterized in that: The method for the central algorithm control system to perform optimization calculations in combination with B1 and B2 includes: S1. When B1 = B2, directly output B3 = B1 = B2; S2. When B1≠B2, the central algorithm control system takes the arithmetic mean of B1 and B2. Will Substitute into oref1 algorithm for parameter optimization and output the carbohydrate intake B2; S3, if Then output like Then repeat step S2 again until the arithmetic mean of the input Equal to the output carbohydrate intake B2.

10. The control method of a blood sugar control system based on closed-loop technology according to claim 5, characterized in that: Based on the user's carbohydrate intake, insulin injection amount, and blood glucose change situation, establish a user-specific DNN model to optimize the method in step S2, specifically including: (1) Continuously add the user's new carbohydrate intake, insulin injection amount, and blood glucose change situation data to the training set to perform incremental learning on the DNN model; (2) During the incremental learning process, fix the first-layer network, release the second-layer and third-layer networks for training, and obtain a new DNN model; (3) The database access module no longer outputs the insulin unit A1 injected or the carbohydrate intake amount B1. The database access module only outputs the theoretical carbohydrate absorption amount or carbohydrate consumption amount. The new DNN model outputs the user-specific carbohydrate absorption amount or carbohydrate consumption amount, and further outputs an insulin injection amount or insulin intake amount. At the same time, the new DNN model outputs a predicted blood glucose value C1; (4) Compare the predicted blood glucose value C1 output by the new DNN model with the target blood glucose C2 set by the oref1 algorithm: If C1 = C2, output the insulin injection amount or insulin intake amount output by the new DNN model; If C1 > C2, output the insulin injection amount or insulin intake amount output by the new DNN model, and add the user's blood glucose change situation to the training set to continue training the model; If C1 < C2, output the insulin injection amount or insulin intake amount output by the oref1 algorithm, and add the user's blood glucose change situation to the training set to continue training the model.

Images