An artificial pancreas long-term adaptive parameter adjustment system based on batch optimization

Abstract

The application discloses a long-term adaptive parameter adjustment system of an artificial pancreas based on batch optimization. The application comprises: a data acquisition and preprocessing module, which is used for acquiring historical data of a patient in an adaptive cycle; a blood glucose condition quantitative evaluation module, which is used for calculating multi-dimensional blood glucose evaluation indexes according to each time period divided by a set insulin basal rate based on the historical data; an adaptive parameter optimization module, which is used for receiving blood glucose evaluation indexes of a last adaptive cycle and calculating insulin basal rate candidate values for each time period according to a preset batch optimization adjustment law; and a safety constraint application module, which is used for applying multi-level safety constraints to the insulin basal rate candidate values to generate a final safe and effective insulin basal rate. The application realizes the robust and individualized updating of key parameters of a bottom controller by quantitatively evaluating the blood glucose control effect in a historical cycle and combining a batch optimization algorithm and multi-level safety constraints.

Description

Technical Field

[0001] This invention belongs to the field of artificial pancreas control technology, and in particular relates to a batch-optimized long-term adaptive parameter adjustment system for artificial pancreas, which is used to optimize key parameters of the underlying controller using historical data to achieve long-term safe closed-loop management of blood glucose. Background Technology

[0002] In the human body, blood glucose levels are strictly controlled by insulin and glucagon secreted by the pancreas. Type 1 diabetes patients suffer from insulin secretion due to the destruction of pancreatic beta cells by their own immune system, requiring lifelong insulin replacement therapy. Artificial pancreas (AP), by integrating a glucose sensor, insulin infusion pump, and controller, mimics the blood glucose regulation function of a healthy pancreas and is currently an effective means of controlling blood glucose in type 1 diabetes.

[0003] Current active insulin control (AP) systems primarily rely on underlying closed-loop control algorithms (such as proportional-integral-derivative control, model predictive control, and active disturbance rejection control). The performance of these algorithms is highly dependent on a set of pre-set key parameters—the basal insulin rate (BR). The basal insulin rate determines the baseline insulin infusion volume at different time periods (such as nighttime, pre-meal, and post-meal). However, the initial setting of the basal insulin rate currently relies mainly on the physician's experience and the patient's repeated trial and error, resulting in limited accuracy. More challenging is the fact that a patient's insulin requirements drift slowly and over a long period depending on their physiological state (such as weight changes, hormone levels, exercise habits, and stress). When the basal insulin rate mismatches with the patient's current actual needs, even if the underlying controller has some real-time adjustment capabilities, it cannot completely correct the resulting deterioration in glycemic control, thereby increasing the risk of hyperglycemia or hypoglycemia.

[0004] To address the aforementioned issues, existing technologies have been explored. For example, patent application CN103418053A proposes a method for optimization by establishing a simple linear relationship between blood glucose levels and basal infusion rate, but it ignores the complex nonlinear characteristics of human blood glucose metabolism, resulting in limited optimization accuracy. Patent application CN109791807A proposes a method for adjusting basal insulin rate based on real-time glucose status risk measurement, but it mainly focuses on real-time control and fails to systematically utilize long-term blood glucose and medication history data to drive a long-term personalized adaptation process of basal insulin rate to cope with slow changes in physiological state. Patent application CN202210425970.0 proposes a Bayesian optimization-based method for optimizing key parameters of artificial pancreas, but this patent's data-driven black-box optimization framework has low clinical interpretability and high computational complexity.

[0005] Therefore, how to fully utilize the historical data (including blood glucose monitoring data and insulin infusion data) generated by patients during daily AP use, and establish a system that can safely and robustly drive the long-term self-adaptation of key parameters (basal insulin rate) of the AP's underlying controller, so as to achieve sustained and safe blood glucose control with minimal human intervention, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the shortcomings of existing technologies, such as long-term mismatch of key parameters (basal insulin rate) in AP systems, lack of systematic safety guarantees and interpretability in the adjustment process, and failure to fully utilize historical data, this invention provides a batch-optimized long-term adaptive parameter adjustment system for an artificial pancreas, where the control target of the artificial pancreas is the blood glucose range. This system achieves robust and individualized updates to key parameters of the underlying controller by quantitatively evaluating blood glucose control performance over historical periods, combined with batch optimization algorithms and multi-level safety constraints.

[0007] A batch-optimized long-term adaptive parameter adjustment system for an artificial pancreas includes:

[0008] Data acquisition and preprocessing module: used to acquire patient data in an adaptive cycle. The system stores historical data, including time-synchronized continuous blood glucose monitoring data, insulin pump infusion data, and patient meal time records. This module also removes data segments interfered with by meal events according to preset rules to ensure the accuracy of subsequent assessments. The historical data is obtained directly from the existing AP system.

[0009] Blood glucose status quantitative assessment module: Based on historical data output from the data acquisition and preprocessing module, it divides the data into various time periods according to the set basal insulin rate. Calculate multi-dimensional blood glucose assessment indicators; these indicators include at least: the percentage of time with low blood glucose (blood glucose <70 mg / dl), the percentage of time with high blood glucose (blood glucose above the upper limit of the target control range), the percentage of time with low blood glucose risk (blood glucose below the lower limit of the target control range), and the time period in which the blood glucose is at risk. The average blood glucose level within the area.

[0010] Adaptive Parameter Optimization Module: The core of this module is a run-to-run-based adjustment law. This module receives the previous adaptive cycle output from the blood glucose status quantification assessment module. The blood glucose evaluation index is used, and candidate values ​​for basal insulin rate for each time period are calculated based on the preset batch optimization adjustment law.

[0011] Safety constraint application module: This module applies multi-level safety constraints, including controller response constraints, smoothing constraints, and step size constraints, to the candidate insulin basal rate values ​​to generate the final safe and effective insulin basal rate. Wherein:

[0012] (1) Controller response constraints: Based on the underlying artificial pancreas controller in the corresponding time period The historical average infusion volume is used to limit the upper limit of the candidate value of insulin basal rate, ensuring that the adjusted insulin basal rate does not exceed the physiological range in which the controller can effectively respond.

[0013] (2) Step size constraint and boundary constraint: Limit the change range of insulin basal rate in a single adaptive adjustment to ensure the stability and safety of the adjustment process.

[0014] (3) Smoothing constraint: Using candidate values ​​of insulin basal rate or historical insulin basal rate from two adjacent time periods, the current time period is subjected to a smoothing constraint. The candidate values ​​of insulin basal rate are smoothed to avoid drastic jumps in the candidate values ​​of insulin basal rate between any adjacent time periods.

[0015] The historical insulin basal rate refers to the previous adaptive cycle. The basal rate of insulin used.

[0016] Furthermore, the specific method for the data acquisition and preprocessing module to eliminate the influence of dining is as follows:

[0017] Identify the patient's meal times and exclude all blood glucose monitoring data and insulin infusion data within a fixed time period (e.g., 2.5 hours) after the start of the meal.

[0018] Furthermore, the specific form of the batch-optimized adjustment law is as follows:

[0019] For the Sub-adaptive cycle , No. Time period Candidate values ​​of insulin basal rate Its calculation method depends on the blood glucose evaluation index corresponding to that time period, as follows:

[0020] If the first Time period Percentage of time spent with hypoglycemia Then, a reduction adjustment will be implemented: ;in, Indicates historical basal insulin rate, This indicates the preset adjustment gain. Indicates the first The time period in the sub-adaptive cycle The percentage of time spent with hypoglycemia.

[0021] If the first Time period Percentage of time with high blood sugar Percentage of time greater than the risk of hypoglycemia Adding a threshold will then perform incremental adjustments: ;in, , This indicates the preset adjustment gain. express The time period in the sub-adaptive cycle The percentage of time with high blood sugar; This represents the average blood glucose level during that time period. This is the preset average blood glucose target. Otherwise, perform a reduction adjustment: . express The time period in the sub-adaptive cycle The percentage of time at risk of hypoglycemia.

[0022] Furthermore, the controller response constraint takes the following form: ;

[0023] in This represents the historical average insulin infusion volume within the corresponding time period, after excluding the impact of meals. The controller control cycle (minutes). To respond to the constraint gain.

[0024] Furthermore, the step size constraint and boundary constraint are in the following forms: .

[0025] in This is the step size constraint gain.

[0026] Furthermore, the smoothing constraint takes the form of: .

[0027] in, Indicates the final basal insulin rate. To smooth out the constraint gain.

[0028] Furthermore, the system also includes an adaptive termination condition judgment module, which terminates the adaptive adjustment process when the preset maximum number of adjustments or the blood glucose control target (such as the average blood glucose reaching the target range and the proportion of normal blood glucose time being greater than the threshold) is reached.

[0029] Compared with the prior art, the present invention has the following significant advantages:

[0030] 1. Systematic use of historical data: This invention systematically extracts multidimensional blood glucose control indicators (such as the proportion of hypoglycemia and hyperglycemia time) from the previous cycle through a quantitative assessment module of blood glucose status. It makes full use of the continuous monitoring data generated by patients in their daily lives, providing a richer and more robust decision-making basis for parameter adjustment than single-point blood glucose values.

[0031] 2. Intelligent Batch Optimization Adjustment: This invention employs a batch optimization framework, treating parameter adjustment as a periodic optimization problem based on historical batches. Its adjustment law designs differentiated adjustment directions and magnitudes for different blood glucose states (hypoglycemia, hyperglycemia, suboptimal blood glucose), with clear logic and high efficiency, effectively tracking long-term, slow changes in the patient's physiological state.

[0032] 3. Multi-layered safety assurance mechanism: This invention innovatively integrates controller response constraints, smoothing constraints, and step size constraints. Controller response constraints ensure that the adjusted basal insulin rate matches the actual performance of the controller; smoothing constraints avoid drastic fluctuations in the basal insulin rate between adjacent time periods, which is more in line with physiological laws; step size constraints limit the aggressiveness of a single adjustment. These three constraints work together to fundamentally guarantee the safety of the long-term adaptive adjustment process and significantly reduce the risk of hypoglycemia or hyperglycemia caused by parameter misadjustment. Attached Figure Description

[0033] Figure 1 This is a framework diagram of the AP long-term adaptation system based on batch optimization provided by the present invention.

[0034] Figure 2 A flowchart illustrating the implementation steps of this invention.

[0035] Table 1 shows the changes in blood glucose management performance indicators under long-term adaptation of the virtual patient AP provided by this invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0037] like Figure 1 As shown, this invention provides a batch-optimized long-term adaptive parameter adjustment system for an artificial pancreas, comprising a data acquisition and preprocessing module, a blood glucose status quantitative assessment module, an adaptive parameter optimization module, and a safety constraint application module. This system serves the underlying artificial pancreas closed-loop controller (e.g., but not limited to, an active disturbance rejection controller (ADRC), a model predictive controller (MPC), etc.), periodically updating its key parameter—basal insulin rate.

[0038] In this embodiment, the adaptive period is 1 week. The 24 hours of the day are divided into [a specific number of hours] based on clinical practice or patient experience. A time period, denoted as Then, the corresponding time period for each day within a week corresponds to a basal insulin rate. (Unit: U / h), and the sum of all time periods is 24 hours. This system aims to adjust various... The numerical value, not the division of time periods.

[0039] The following is combined Figure 2 The flowchart shown illustrates in detail the specific implementation steps of this embodiment.

[0040] Step 1: Data Acquisition and Preprocessing

[0041] The system is triggered at the end of each adaptive cycle. The data acquisition and preprocessing module first collects historical data for the current cycle, including: continuous glucose monitoring (CGM) data every 5 minutes, closed-loop controlled insulin pump infusion data (U) every 5 minutes, and the patient's recorded meal start time.

[0042] To eliminate the impact of meal disturbances on glycemic assessment and controller behavior analysis, this module removes all CGM and insulin infusion data within 2.5 hours after each meal. If the adaptive period is multiple days, the average of valid data from those multiple days is used for each indicator. Furthermore, a data integrity check is performed: if the amount of valid CGM or insulin infusion data on a given day is less than 70% of the total data to be collected (e.g., fewer than 202 out of 288 CGM points), all data for that day is discarded; if the final number of valid days is less than the preset minimum number of days (default is 2 days), this adaptive adjustment is not executed.

[0043] Step Two: Quantitative Assessment of Blood Glucose Status

[0044] The blood glucose status quantitative assessment module receives preprocessed data and analyzes it for each time period. Calculate the following four key indicators:

[0045] • Percentage of time spent with hypoglycemia During the time period Within the corresponding effective time window, the cumulative time with blood glucose concentration below 70 mg / dl constitutes the proportion of the total effective time (time period). The percentage of effective time after deducting meal time.

[0046] Percentage of time with high blood sugar During the time period Within the corresponding effective time window, the cumulative time during which the blood glucose concentration is higher than the upper limit of the control target interval (e.g., if the default target interval is [120, 160] mg / dl, then the upper limit is 160 mg / dl) represents the percentage of the total effective time (time period). The percentage of effective time after deducting meal time.

[0047] • Percentage of time at risk of hypoglycemia During the time period Within the corresponding effective time window, the cumulative time during which the blood glucose concentration is below the lower limit of the control target interval (i.e., 120 mg / dl) accounts for a portion of the total effective time (time period). The percentage of effective time after deducting meal time.

[0048] Average blood glucose level During the time period The average value of all blood glucose monitoring data within the corresponding valid time window.

[0049] Step 3: Adaptive Parameter Optimization (Batch Optimization)

[0050] The adaptive parameter optimization module receives the above And the basal insulin rate of the previous cycle Candidate values ​​for insulin basal rate are calculated based on the batch optimization adjustment law. .

[0051] If hypoglycemia occurred during that time period ( If so, the system's primary objective is to reduce the candidate value of basal insulin rate.

[0052] If hypoglycemia does not occur, but hyperglycemia is a prominent issue, that is... In this case, the system needs to increase the candidate value of basal insulin rate to control hyperglycemia.

[0053] In other cases (where blood glucose control is basically within the target range or slightly low), the system cautiously and slightly reduces the candidate value of the basal insulin rate:

[0054] In this embodiment, the gain value is adjusted to... , , target mean blood glucose Set as the target interval mean of the underlying controller.

[0055] Step 4: Applying safety constraints

[0056] Safety constraint application module for insulin basal rate candidate values Three constraints are applied sequentially to generate the final basal insulin rate. .

[0057] 1. Controller response constraints: Calculate the relationship between the previous period and the current time period. Correspondingly, the historical average insulin infusion volume after excluding data from 2.5 hours postprandial. Candidate values ​​for the basal insulin rate must not exceed the upper limit of the actual controller response capability reflected by this mean. in, minutes (CGM sampling period) Set the value to 1.5. If a candidate value exceeds this upper limit, truncate it to the upper limit value.

[0058] 2. Step size constraint and boundary constraint: Limit the adjustment range and absolute value range of candidate values ​​for insulin basal rate.

[0059] Among them, step size constraint gain The value is 0.3, meaning that the single adjustment range shall not exceed 30%. and The values ​​are 0.25 U / h and 3.0 U / h, respectively. If the calculated value exceeds the boundary, the boundary value is taken.

[0060] 3. Smoothing constraint: Utilize two adjacent time periods (the first time period) and The candidate values ​​of insulin basal rate (segment) and the historical insulin basal rate of the previous cycle are used to smooth the current candidate values. Among them, smoothing constraint gain The value is set to 0.8. This constraint prevents unreasonable steep changes in the basal insulin rate between adjacent time periods.

[0061] Step 5: Update and Iterate the Low-Level Controller Parameters

[0062] The final determination The basal insulin rate, used for the next adaptive cycle, is updated in the underlying artificial pancreas controller. Simultaneously, the underlying reference control gain of the control system is... It can also be updated based on the mean basal insulin rate: ,in The system then waits for the next cycle to end and repeats steps one through five to achieve long-term parameter adaptability.

[0063] Experimental results

[0064] To verify the effectiveness of this invention, simulation experiments were conducted on the FDA-approved UVA / Padova T1DM blood glucose metabolism simulator. The underlying controller employed an Active Disturbance Rejection Control (ADRC) algorithm. For virtual patients with excessively high basal insulin rates, their basal insulin rates were divided into four segments, and the system of this invention was applied for long-term adaptive adjustment. A total of 10 virtual adult patients were simulated. The results of changes in blood glucose management performance indicators are shown in Table 1, with data in the interquartile range (interquartile range). As shown in Table 1, before optimization, the overall blood glucose levels of the patients fluctuated drastically, with frequent occurrences of hypoglycemia (<70 mg / dl) and a low percentage of time within the normal blood glucose range. After 5-6 adaptive cycles (5-6 weeks), the patients' blood glucose levels were stably controlled within the normal blood glucose range (70-180 mg / dl), the average blood glucose level of the patients significantly improved, and the percentage of time within the normal blood glucose range (TIR) ​​increased from an initial overall average of less than 80% to more than 90%, with no severe hypoglycemic events caused by parameter adjustments occurring throughout the process.

[0065] Table 1. Results of changes in blood glucose management performance indicators under long-term adaptation of the virtual patient AP provided by this invention.

[0066]

[0067] In summary, the present invention provides a batch-optimized long-term adaptive parameter adjustment system for artificial pancreas. Through a data-driven "evaluation-optimization-constraint-update" mechanism, it can safely and effectively automatically calibrate the key parameters of the AP underlying controller, significantly improve long-term blood glucose control, reduce manual intervention, and has high industrial application value.

[0068] This application also provides an electronic device, including a processor and a memory.

[0069] The memory is used to store computer programs.

[0070] When the processor executes a program stored in the memory, it implements any of the methods described in this application.

[0071] In one possible implementation, the electronic device of this application embodiment further includes a communication interface and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.

[0072] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0073] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0074] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0075] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0076] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, implements any of the methods described in this application.

[0077] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform any of the methods described in this application.

[0078] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0079] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0080] The various embodiments in this specification are described in a related manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other.

[0081] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application are included within the scope of protection of this application.

Claims

1. A batch-optimized long-term adaptive parameter adjustment system for artificial pancreas, characterized in that, include: Data acquisition and preprocessing module: used to acquire patient data in an adaptive cycle. Historical data, including time-synchronized continuous blood glucose monitoring data, insulin pump infusion data, and patient meal time records; Blood glucose status quantitative assessment module: Based on historical data output from the data acquisition and preprocessing module, it divides the data into various time periods according to the set basal insulin rate. Calculate multi-dimensional blood glucose evaluation indicators; Adaptive Parameter Optimization Module: The core of this module is a batch-optimized adjustment law. This module receives the previous adaptive cycle output by the blood glucose status quantification assessment module. The blood glucose evaluation index is used, and candidate values ​​of basal insulin rate for each time period are calculated based on the preset batch optimization adjustment law. Safety constraint application module: used to apply multi-level safety constraints, including controller response constraints, step size constraints and boundary constraints, and smoothing constraints, to the candidate value of insulin basal rate, so as to generate the final safe and effective insulin basal rate.

2. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 1, characterized in that, The data acquisition and preprocessing module is also used to remove data segments that are interfered with by meal events according to preset rules. Specifically, it identifies the patient's meal time and removes all blood glucose monitoring data and insulin infusion data within a set fixed time period after the start of the meal.

3. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 1, characterized in that, The adaptive parameter optimization module's blood glucose evaluation indicators include at least: the percentage of time with low blood glucose, the percentage of time with high blood glucose, the percentage of time with low blood glucose risk, and the time period itself. The average blood glucose level within the area.

4. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 3, characterized in that, The safety constraint application module is implemented as follows: Controller response constraints: Based on the underlying artificial pancreas controller within the corresponding time period The historical average infusion volume is used to limit the upper limit of the candidate value of insulin basal rate, ensuring that the adjusted candidate value of insulin basal rate does not exceed the physiological range in which the controller can effectively respond. Step size constraint and boundary constraint: Limit the range of change of candidate values ​​of insulin basal rate in a single adaptive adjustment; Smoothing constraint: Using candidate values ​​of insulin basal rate from two adjacent time periods or historical insulin basal rate, the current time period is subjected to smoothing constraint. Candidate values ​​for insulin basal rate are smoothed and limited.

5. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 3, characterized in that, The historical insulin basal rate refers to the previous adaptive cycle. The basal rate of insulin used.

6. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 4, characterized in that, The specific form of the adjustment law based on batch optimization in the adaptive parameter optimization module is as follows: For the Sub-adaptive cycle , No. Time period Candidate values ​​of insulin basal rate Its calculation method depends on the blood glucose evaluation index corresponding to that time period, as follows: If the first Time period Percentage of time spent with hypoglycemia Then, a reduction adjustment will be implemented: ;in, Indicates historical basal insulin rate, This indicates the preset adjustment gain. Indicates the first The time period in the sub-adaptive cycle The percentage of time spent with hypoglycemia; If the first Time period Percentage of time with high blood sugar Percentage of time greater than the risk of hypoglycemia Adding a threshold will then perform incremental adjustments: ;in, , This indicates the preset adjustment gain. express The time period in the sub-adaptive cycle The percentage of time with high blood sugar; This represents the average blood glucose level during that time period. The target average blood glucose level is set; otherwise, a reduction adjustment is performed. ; express The time period in the sub-adaptive cycle The percentage of time at which hypoglycemia is at risk.

7. The long-term adaptive parameter adjustment system for artificial pancreas based on batch optimization according to claim 4, characterized in that, The form of the controller response constraint is: ； in This represents the historical average insulin infusion volume within the corresponding time period, after excluding the impact of meals. For the controller control cycle, To respond to the constraint gain.

8. A batch-optimized long-term adaptive parameter adjustment system for artificial pancreas according to claim 4, characterized in that, The step size constraint and boundary constraint are in the following forms: ； in, Step-size constrained gain; and These are the lower and upper bounds of the boundary constraint, respectively.

9. A batch-optimized long-term adaptive parameter adjustment system for an artificial pancreas according to claim 4, characterized in that, The smoothing constraint takes the form of: ； in, Indicates the final basal insulin rate. To smooth out the constraint gain.

10. A batch-optimized long-term adaptive parameter adjustment system for an artificial pancreas according to claim 4, characterized in that, The system also includes an adaptive termination condition judgment module, which terminates the adaptive adjustment process when the preset maximum number of adjustments or the blood glucose control target is reached.

Images