Intelligent automatic insulin injection system based on blood glucose measurement

By adopting multi-level calibration and adaptive filtering algorithms in the intelligent insulin automatic injection system, combining adaptive learning and dynamic metabolic models, combined with intelligent microfluidic control and environmental perception modules, the problems of blood sugar data error and individual differences are solved, and the problem of blood sugar data error and individual differences is realized. Accurate insulin injection and personalized diabetes management.

CN118949198AInactive Publication Date: 2024-11-15TONGJI HOSPITAL ATTACHED TO TONGJI MEDICAL COLLEGE HUAZHONG SCI TECH
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411269363.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-11
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure CN118949198A_ABST
    Figure CN118949198A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent automatic insulin injection system based on blood glucose measurement. The intelligent automatic insulin injection system based on blood glucose measurement comprises a real-time blood glucose monitoring sensor which is used for being attached to the skin of a patient and continuously monitoring the blood glucose level of the patient, and an insulin pump which is in communication connection with the real-time blood glucose monitoring sensor and is used for accurately injecting insulin according to a received instruction. According to the intelligent automatic insulin injection system based on blood glucose measurement, accurate calculation is conducted through control software, insulin is automatically injected, a data processing module adopts a multi-level calibration and self-adaptive filtering algorithm, errors caused by the service life of a sensor, the skin condition of a patient and external environment factors are effectively corrected, the accuracy of blood glucose data is ensured, and the accuracy of the blood glucose data is improved. The algorithm model module combines adaptive learning, multivariate prediction, reinforcement learning, a self-optimization model and a dynamic metabolism model, can dynamically adjust the insulin demand based on individual parameters of a patient, and ensures that the injection dosage is more accurate and personalized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of automatic insulin injection, and in particular to an intelligent automatic insulin injection system based on blood sugar measurement. Background Art

[0002] An intelligent automatic insulin injection system based on blood sugar measurement is mainly composed of the following key parts: real-time blood sugar monitoring sensor, insulin pump, control software and communication module. The real-time blood sugar monitoring sensor is attached to the patient's skin surface, usually placed on the arm or abdomen, and continuously monitors the patient's blood sugar level in a minimally invasive way, and transmits the data to the control software in real time. The control software calculates the required insulin dose through a specific algorithm based on the received blood sugar data and the patient's individual parameters such as age, gender, and weight. Then, the control software sends the calculation results to the insulin pump, and the insulin pump automatically injects the corresponding dose of insulin according to the instructions to maintain the patient's blood sugar within the target range. The system can operate in automatic mode, or manually set the insulin dose corresponding to a specific blood sugar value, so that patients can flexibly respond to different blood sugar fluctuations.

[0003] Although the system is intelligent and automated, it still has some potential flaws. The accuracy and stability of blood glucose monitoring sensors may be limited by the life of the sensor, the patient's skin condition, and external environmental factors, resulting in the accumulation of errors in blood glucose data, thereby affecting the accuracy of insulin injection. When dealing with complex blood glucose fluctuations, the automatic control algorithm of the insulin pump may not fully consider the individual differences of patients, such as diet, exercise, stress and other external factors, which may lead to deviations in insulin dosage and even the risk of hypoglycemia or hyperglycemia. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an intelligent automatic insulin injection system based on blood sugar measurement, which solves the problem that the life of the sensor, the patient's skin condition and external environmental factors lead to the accumulation of errors in blood sugar data, thereby affecting the accuracy of insulin injection; the automatic control algorithm of the insulin pump may not fully take into account the individual differences of patients, such as diet, exercise, and stress when dealing with complex blood sugar fluctuations.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent automatic insulin injection system based on blood sugar measurement, comprising:

[0006] Real-time blood glucose monitoring sensors are used to attach to the patient's skin to continuously monitor the patient's blood glucose level;

[0007] An insulin pump, in communication with the real-time blood glucose monitoring sensor, for accurately injecting insulin according to received instructions;

[0008] Control software, including data processing module, algorithm model module and control module, among which:

[0009] A data processing module is used to receive and process blood glucose data from a real-time blood glucose monitoring sensor, and to correct errors caused by sensor life, patient skin condition, and external environmental factors through multi-level calibration and adaptive filtering algorithms;

[0010] The algorithm model module dynamically adjusts the insulin demand calculation formula based on individual patient parameters, including but not limited to age, gender, weight, eating habits, exercise patterns and stress levels, using adaptive learning algorithms, multivariate prediction models, self-optimization models based on reinforcement learning and dynamic metabolic models;

[0011] The control module sends precise injection instructions through the communication interface with the insulin pump according to the insulin demand calculated by the algorithm model module.

[0012] Preferably, the spectral analysis function of the real-time blood glucose monitoring sensor monitors the hydration and blood flow changes of the subcutaneous tissue through near-infrared spectroscopy technology, providing additional physiological data to improve the accuracy of blood glucose measurement. The real-time blood glucose monitoring sensor includes a multi-sensor array and integrates nanoscale sensing technology and spectral analysis functions to detect multiple biomarkers such as lactate and ketone bodies under the patient's skin to assist in correcting blood glucose data.

[0013] Preferably, the data processing module includes a multi-level feedback mechanism, which adjusts the parameters of data processing in real time by combining machine learning and biofeedback technology to cope with individual differences among patients and environmental changes.

[0014] Preferably, the algorithm model module uses the following formula to calculate the insulin requirement:

[0015]

[0016] Where: I(t) represents the insulin dose required to be injected at time t, BG(t) is the currently measured blood glucose value, BG target is the target blood glucose value, S is the patient's insulin sensitivity coefficient, is the blood glucose change rate, C(t) is the correction factor related to the patient's individualized parameters, including dietary intake, exercise intensity, and stress level, RL(t) is the self-optimization item based on the reinforcement learning model, which automatically adjusts the insulin injection strategy by learning the patient's long-term behavior pattern, DM(t) is the correction item of the dynamic metabolic model, which takes into account the patient's basal metabolic rate (BMR), postprandial metabolic response, and insulin resistance, α, β, γ, δ, ∈ are adjustment coefficients, and personalized adjustments are made based on the patient's historical data.

[0017] Preferably, the dynamic metabolic model is based on the patient's daily diet, exercise, metabolic rate and sleep pattern, and uses a time series analysis method to predict blood sugar changes under different circumstances. The dynamic metabolic model also integrates the effects of fluctuations in glucagon and cortisol levels in the body to more accurately predict blood sugar response and insulin demand.

[0018] Preferably, the control software includes a set of intelligent threshold adjustment mechanisms. When the calculated insulin demand exceeds the safety threshold, the system will automatically recalibrate the threshold based on the patient's historical data, reduce the alarm frequency, and improve the system's safety and user experience. The reinforcement learning model uses a deep reinforcement learning algorithm combined with a convolutional neural network and a long short-term memory network to better capture and predict complex blood sugar change patterns.

[0019] Preferably, the insulin pump includes an intelligent microfluidic system, which realizes precise control and microinjection of insulin through microfluidic technology, thereby reducing the risk of overinjection. The intelligent threshold adjustment mechanism adopts a fuzzy logic control system, which can flexibly adjust the threshold range and response strategy according to different clinical scenarios and patient historical data. The intelligent microfluidic system integrates a real-time feedback loop to instantly adjust the insulin injection rate according to the blood sugar level feedback from the sensor, thereby realizing more precise dosage control.

[0020] Preferably, the intelligent automatic insulin injection system also includes an environmental sensing module, which monitors the environmental factors around the patient in real time through sensors, and inputs these data into the algorithm model module to further improve the accuracy of insulin demand prediction. The environmental factors include temperature, humidity, and air pressure. The environmental sensing module combines geographic location data with correlation analysis to automatically identify the patient's life and activity patterns and make more detailed predictions of insulin needs.

[0021] Preferably, the intelligent automatic insulin injection system also includes an intelligent diet recognition module, which uses computer vision technology to automatically identify the type and amount of food ingested by the patient, and uses this information as one of the input variables to directly affect the calculation of insulin requirements.

[0022] Preferably, the intelligent diet recognition module integrates an image analysis algorithm based on deep learning, which can identify and quantify the nutritional components of carbohydrates, fats, and proteins, and provide real-time feedback to the algorithm model module for calculation. The detailed model of the real-time feedback correction term F(t)

[0023] F(t)=θ1·Env (t)+θ2·Food (t)+θ3·Exercise (t)

[0024] Among them, Env(t): the impact of environmental factors monitored in real time, including temperature, humidity, and air pressure. This item is calculated in real time through the data provided by the sensor module. Food(t): the impact of the nutritional components of the current food intake identified and quantified by the intelligent diet recognition module. Exercise(t): the impact of the current amount of exercise on insulin demand, usually calculated based on heart rate, number of steps, and exercise intensity data. θ1, θ2, θ3: weight coefficients used to adjust the relative impact of different real-time feedback factors, which depend on individualized patient parameters and historical data.

[0025] The present invention provides an intelligent automatic insulin injection system based on blood sugar measurement.

[0026] Beneficial effects:

[0027] This intelligent automatic insulin injection system based on blood glucose measurement accurately calculates and automatically injects insulin through control software. The data processing module adopts multi-level calibration and adaptive filtering algorithms to effectively correct errors caused by sensor life, patient skin condition and external environmental factors to ensure the accuracy of blood glucose data. The algorithm model module combines adaptive learning, multivariate prediction, reinforcement learning, self-optimization model and dynamic metabolic model, which can dynamically adjust insulin requirements based on the patient's individualized parameters to ensure that the injection dose is more accurate and personalized.

[0028] The system also integrates a variety of innovative technologies to enhance user experience and safety. The sensor uses nanoscale technology and spectral analysis capabilities to monitor biomarkers under the skin in real time, further improving the accuracy of blood glucose measurement. The intelligent threshold adjustment mechanism in the control software uses fuzzy logic and reinforcement learning algorithms to automatically optimize injection strategies and reduce unnecessary alarms when facing complex blood glucose changes. It also adjusts the insulin injection rate in real time through an intelligent microfluidic system to prevent overinjection. In addition, the system also includes environmental perception and intelligent diet recognition modules, which monitor environmental factors and automatically identify food intake, and provide real-time feedback to the algorithm model for calculation, making insulin demand prediction more refined, and ultimately providing patients with a safer and more effective diabetes management solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the internal structure of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] like Figure 1 As shown, an embodiment of the present invention provides an intelligent automatic insulin injection system based on blood glucose measurement, including a real-time blood glucose monitoring sensor for attaching to the patient's skin to continuously monitor the patient's blood glucose level. The spectral analysis function of the real-time blood glucose monitoring sensor monitors the hydration and blood flow changes of the subcutaneous tissue through near-infrared spectroscopy technology, and provides additional physiological data to improve the accuracy of blood glucose measurement. The real-time blood glucose monitoring sensor includes a multi-sensor array and integrates nano-scale sensing technology and spectral analysis function to detect multiple biomarkers lactate and ketone bodies under the patient's skin to assist in correcting blood glucose data.

[0032] The insulin pump is connected to the real-time blood glucose monitoring sensor for accurately injecting insulin according to the received instructions. The insulin pump contains an intelligent microfluidic system, which uses microfluidic technology to achieve precise control and microinjection of insulin, reducing the risk of overdose. The intelligent threshold adjustment mechanism adopts a fuzzy logic control system, which can flexibly adjust the threshold range and response strategy according to different clinical scenarios and patient historical data. The intelligent microfluidic system integrates a real-time feedback loop to instantly adjust the insulin injection rate according to the blood glucose level fed back by the sensor, achieving more accurate dosage control. The intelligent automatic insulin injection system also includes an environmental perception module, which monitors the environmental factors around the patient in real time through sensors, and inputs these data into the algorithm model module The module is used to further improve the accuracy of insulin demand prediction. Environmental factors include temperature, humidity, and air pressure. The environmental perception module combines geographic location data and automatically identifies the patient's life and activity patterns through association analysis to make more precise predictions of insulin demand. The intelligent insulin automatic injection system also includes an intelligent diet recognition module, which uses computer vision technology to automatically identify the type and amount of food consumed by the patient, and uses this information as one of the input variables, which directly affects the calculation of insulin demand. The intelligent diet recognition module integrates an image analysis algorithm based on deep learning, which can identify and quantify the nutritional components of carbohydrates, fats, and proteins, and provide real-time feedback to the algorithm model module for calculation. The detailed model of the real-time feedback correction term F(t)

[0033] F(t)=θ1·Env (t)+θ2·Food (t)+θ3·Exercise (t)

[0034] Among them, Env(t): the impact of environmental factors monitored in real time, including temperature, humidity, and air pressure. This item is calculated in real time through the data provided by the sensor module. Food(t): the impact of the nutritional components of the current food intake identified and quantified by the intelligent diet recognition module. Exercise(t): the impact of the current amount of exercise on insulin demand, usually calculated based on heart rate, number of steps, and exercise intensity data. θ1, θ2, θ3: weight coefficients used to adjust the relative impact of different real-time feedback factors, which depend on individualized patient parameters and historical data.

[0035] The control software includes a data processing module, an algorithm model module and a control module. The control software includes a set of intelligent threshold adjustment mechanisms. When the calculated insulin demand exceeds the safety threshold, the system will automatically recalibrate the threshold based on the patient's historical data to reduce the alarm frequency, while improving the system's safety and user experience. The reinforcement learning model uses a deep reinforcement learning algorithm, combined with a convolutional neural network and a long short-term memory network, to better capture and predict complex blood sugar change patterns, including:

[0036] The data processing module is used to receive and process blood glucose data from the real-time blood glucose monitoring sensor. It corrects errors caused by sensor life, patient skin condition and external environmental factors through multi-level calibration and adaptive filtering algorithms. The data processing module contains a multi-level feedback mechanism. By combining machine learning and biofeedback technology, it adjusts data processing parameters in real time to cope with individual differences among patients and environmental changes.

[0037] The algorithm model module uses an adaptive learning algorithm, a multivariate prediction model, a self-optimization model based on reinforcement learning, and a dynamic metabolic model to dynamically adjust the insulin requirement calculation formula based on individual patient parameters, including but not limited to age, gender, weight, eating habits, exercise patterns, and stress levels. The algorithm model module uses the following formula to calculate insulin requirements:

[0038]

[0039] Where: I(t) represents the insulin dose required to be injected at time t, BG(t) is the currently measured blood glucose value, BG target is the target blood glucose value, S is the patient's insulin sensitivity coefficient, is the blood glucose change rate, C(t) is the correction factor related to the patient's individualized parameters, including dietary intake, exercise intensity, and stress level. RL(t) is a self-optimization item based on the reinforcement learning model. By learning the patient's long-term behavior pattern, the insulin injection strategy is automatically adjusted. DM(t) is a correction item of the dynamic metabolic model, which takes into account the patient's basal metabolic rate (BMR), postprandial metabolic response, and insulin resistance. α, β, γ, δ, ∈ are adjustment coefficients, which are personalized according to the patient's historical data. The dynamic metabolic model is based on the patient's daily diet, exercise, metabolic rate, and sleep pattern, and uses a time series analysis method to predict blood glucose changes under different circumstances. The dynamic metabolic model also integrates the effects of fluctuations in glucagon and cortisol levels in the body to more accurately predict blood glucose response and insulin demand.

[0040] The control module sends precise injection instructions through the communication interface with the insulin pump according to the insulin demand calculated by the algorithm model module.

[0041] Intelligent insulin automatic injection system based on blood glucose measurement. The patient's current blood glucose level is BG(t) = 180 mg / dL, and the target blood glucose value is BG target is 120mg / dL, the patient's insulin sensitivity coefficient S is 50mg / dL, and the blood sugar change rate measured by the system in real time The patient's current exercise status is equivalent to moderate-intensity exercise, and the diet recognition module detects that the patient has just consumed food containing 60 grams of carbohydrates.

[0042] Using the algorithm model module in the system, the calculation formula for insulin demand is as follows:

[0043]

[0044] It is assumed that the initial values ​​of the parameters are: α = 1.2, β = 0.8, γ = 1.0, δ = 0.9, ∈ = 1.1, ζ = 1.0, η = 1.2

[0045] In this case, the correction factor C(t) includes the effect of dietary intake C food = 1.5 Calculated based on 60g carbohydrate intake, exercise correction factor C exercise =0.8, the correction terms based on the assumed reinforcement learning model and dynamic metabolic model are RL(t)=1.0 and DM(t)=0.9 respectively, the correction term of the pharmacokinetic / pharmacodynamic model is PKPD(t)=1.0, and the correction term of the environmental perception module feedback is F(t)=1.2.

[0046] Substitute the above values ​​into the formula for calculation: the system will calculate that under the current conditions, the patient needs to inject approximately 9.67 units of insulin to adjust the blood sugar level to the target level. This calculation result will be sent to the insulin spring through the control module to ensure that the patient can receive the appropriate amount of insulin injection in a timely and accurate manner, minimizing the impact of blood sugar fluctuations on the body. This calculation example shows how the intelligent automatic insulin injection system can dynamically adjust the insulin injection dose by integrating multiple real-time feedback mechanisms and algorithm models, thereby achieving personalized and precise diabetes management.

[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent automatic insulin injection system based on blood sugar measurement, characterized in that: include: Real-time blood glucose monitoring sensors are used to attach to the patient's skin to continuously monitor the patient's blood glucose level; An insulin pump, in communication with the real-time blood glucose monitoring sensor, for accurately injecting insulin according to received instructions; Control software, including data processing module, algorithm model module and control module, among which: A data processing module is used to receive and process blood glucose data from a real-time blood glucose monitoring sensor, and to correct errors caused by sensor life, patient skin condition, and external environmental factors through multi-level calibration and adaptive filtering algorithms; The algorithm model module dynamically adjusts the insulin demand calculation formula based on individual patient parameters, including but not limited to age, gender, weight, eating habits, exercise patterns and stress levels, using adaptive learning algorithms, multivariate prediction models, self-optimization models based on reinforcement learning and dynamic metabolic models; The control module sends precise injection instructions through the communication interface with the insulin pump according to the insulin demand calculated by the algorithm model module.

2. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The spectral analysis function of the real-time blood glucose monitoring sensor monitors the hydration and blood flow changes of the subcutaneous tissue through near-infrared spectroscopy technology. The real-time blood glucose monitoring sensor includes a multi-sensor array and integrates nano-scale sensing technology and spectral analysis function.

3. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The data processing module includes a multi-level feedback mechanism that combines machine learning and biofeedback technology to adjust data processing parameters in real time to cope with individual differences in patients and environmental changes.

4. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The algorithm model module uses the following formula to calculate insulin requirement: Where: I(t) represents the insulin dose required to be injected at time t, BG(t) is the currently measured blood glucose value, BG target is the target blood glucose value, S is the patient's insulin sensitivity coefficient, is the blood glucose change rate, C(t) is the correction factor related to the patient's individualized parameters, including dietary intake, exercise intensity, and stress level, RL(t) is the self-optimization item based on the reinforcement learning model, which automatically adjusts the insulin injection strategy by learning the patient's long-term behavior pattern, DM(t) is the correction item of the dynamic metabolic model, which takes into account the patient's basal metabolic rate (BMR), postprandial metabolic response, and insulin resistance, α, β, γ, δ, ∈ are adjustment coefficients, and personalized adjustments are made based on the patient's historical data.

5. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The dynamic metabolic model is based on the patient's daily diet, exercise, metabolic rate and sleep pattern, and uses a time series analysis method to predict blood sugar changes under different circumstances. The dynamic metabolic model also integrates the effects of fluctuations in glucagon and cortisol levels in the body to more accurately predict blood sugar responses and insulin requirements.

6. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The control software includes a set of intelligent threshold adjustment mechanisms. When the calculated insulin demand exceeds the safety threshold, the system will automatically recalibrate the threshold based on the patient's historical data, reduce the alarm frequency, and improve the system's safety and user experience. The reinforcement learning model uses a deep reinforcement learning algorithm combined with a convolutional neural network and a long short-term memory network.

7. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The insulin pump includes an intelligent microfluidic system, which realizes precise control and microinjection of insulin through microfluidic technology. The intelligent threshold adjustment mechanism adopts a fuzzy logic control system. The intelligent microfluidic system integrates a real-time feedback loop to instantly adjust the insulin injection rate according to the blood sugar level feedback from the sensor, thereby realizing more precise dosage control.

8. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 7, characterized in that: The intelligent automatic insulin injection system also includes an environmental perception module, which monitors the environmental factors around the patient in real time through sensors, and inputs these data into the algorithm model module to further improve the accuracy of insulin demand prediction. The environmental factors include temperature, humidity, and air pressure. The environmental perception module combines geographic location data and automatically identifies the patient's life and activity patterns through correlation analysis to make more precise predictions of insulin demand.

9. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The intelligent insulin automatic injection system also includes an intelligent diet recognition module, which uses computer vision technology to automatically identify the type and amount of food consumed by the patient, and uses this information as one of the input variables to directly affect the calculation of insulin requirements.

10. The intelligent automatic insulin injection system based on blood sugar measurement according to claim 1, characterized in that: The intelligent diet recognition module integrates an image analysis algorithm based on deep learning, which can identify and quantify the nutritional components of carbohydrates, fats, and proteins, and provide real-time feedback to the algorithm model module for calculation. The detailed model of the real-time feedback correction term F(t) F(t)=θ1·Env (t)+θ2·Food (t)+θ3·Exercise (t) Among them, Env(t): the impact of environmental factors monitored in real time, including temperature, humidity, and air pressure. This item is calculated in real time through the data provided by the sensor module. Food(t): the impact of the nutritional components of the current food intake identified and quantified by the intelligent diet recognition module. Exercise(t): the impact of the current amount of exercise on insulin demand, usually calculated based on heart rate, number of steps, and exercise intensity data. θ1, θ2, θ3: weight coefficients used to adjust the relative impact of different real-time feedback factors, which depend on individualized patient parameters and historical data.

Citation Information

Cited By

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

    CN119633202A

  • Insulin infusion decision-making system and method combining reinforcement learning and metabolism simulation

    CN120565097A

  • Blood glucose information management system for dynamically monitoring blood glucose

    CN120938427A