Blood glucose monitoring computer intelligent analysis system and method based on AI drive

The AI-driven intelligent analysis system for blood glucose monitoring solves the problem of insufficient blood glucose data update cycle by integrating dynamic trend decomposition and interval fluctuation characteristics. It enables real-time identification of blood glucose changes and optimization of behavioral intervention, thereby improving the effectiveness of chronic disease health management.

CN121460125APending Publication Date: 2026-02-03乐清市人民医院
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511582191.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-31
Publication Date
2026-02-03

AI Technical Summary

Technical Problem

Existing technologies acquire blood glucose data in a static manner, with limited update cycles, making it difficult to track the continuity of blood glucose fluctuations and high-risk zones. They also lack multi-parameter fusion judgment mechanisms, making it impossible to identify the interaction between behavioral events and blood glucose changes in real time. Consequently, risk information stratification and detailed intervention of behavioral responses are difficult to achieve.

Method used

It provides an AI-driven intelligent computer analysis system for blood glucose monitoring, including a fluctuation trend recognition module, a clustering window screening module, a food intake response discrimination module, a behavioral deviation analysis module, and an intervention level determination module. Through multi-parameter interactive verification and dynamic trend decomposition, it realizes trend clustering of blood glucose changes and identification of high-risk intervals, and dynamically optimizes the intervention priority of behavioral events.

Benefits of technology

It enables the linkage and discrimination of blood glucose changes with multiple behavioral scenarios, enhances the real-time quantification and stratification of blood glucose changes and behavioral responses, promotes individualized and graded responses and proactive adjustments, and improves the effectiveness of chronic disease health management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121460125A_ABST
    Figure CN121460125A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of health management, in particular to a blood glucose monitoring computer intelligent analysis system and method based on AI drive, and the system comprises a fluctuation trend recognition module, an aggregation window screening module, a feeding response judgment module, a behavior deviation analysis module and an intervention level judgment module. According to the method, through dynamic trend decomposition and interval fluctuation feature fusion of blood glucose monitoring data, trend clustering and high-risk interval active identification of blood glucose changes in a continuous time period are achieved, multi-source behavior information collaboratively analyzes blood glucose responses of events such as feeding, exercise and medication, tracing judgment is carried out on abnormal signals through multi-parameter interaction verification, and the accuracy of blood glucose monitoring is improved. The behavior migration factors and risk influences are quantified and layered in real time, the linkage judgment ability of blood glucose changes and various behavior situations is enhanced, the intervention priority of behavior events is dynamically optimized, and individualized and graded response and active adjustment in a chronic disease health management scene are promoted.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of health management technology, and in particular to an AI-driven intelligent computer analysis system and method for blood glucose monitoring. Background Technology

[0002] Health management involves the long-term dynamic monitoring, assessment, and intervention of the health status of individuals or groups. It mainly includes the collection and analysis of physiological parameters, health risk prediction, chronic disease management, generation of personalized intervention recommendations, and guidance on healthy behaviors. Traditional blood glucose monitoring involves periodically acquiring a user's blood glucose level using methods such as enzyme electrode glucometers or finger-prick blood sampling devices. This data is then manually recorded and interpreted by the user or healthcare professionals. It primarily relies on the electrochemical reaction of glucose oxidase to generate an electrical signal, which is then used to calculate the blood glucose concentration.

[0003] Existing technologies acquire blood glucose data in a static manner, resulting in limited data update cycles, insufficient information coverage, and difficulty in fully reflecting periodic changes. It is also difficult to track the continuity of blood glucose fluctuations and high-risk intervals, identify the interaction between behavioral events and blood glucose changes, lack a multi-parameter fusion judgment mechanism, and make it difficult to complete the risk information stratification and behavioral tracing triggered by abnormal blood glucose in real time. Furthermore, it is difficult to fully realize the detailed intervention of risk aggregation and behavioral response in health management. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide an AI-driven intelligent computer analysis system and method for blood glucose monitoring. The technical solution is as follows:

[0005] On the one hand, an AI-driven intelligent computer-based blood glucose monitoring and analysis system is provided, the system comprising:

[0006] The fluctuation trend recognition module, based on blood glucose monitoring data, determines the direction of blood glucose changes every hour. By comparing the trends of high and low points in each time period, it identifies the fluctuation characteristics of each time period and obtains cross-segment fluctuation characteristic groups.

[0007] The clustered window screening module compares the current fluctuation pattern with the historical window based on the cross-segment fluctuation feature group, determines whether there is a concentration trend in multiple consecutive windows, identifies windows that meet the features, and obtains the fluctuation clustering interval.

[0008] The feeding response discrimination module analyzes the feeding behavior data within the fluctuation clustering interval based on the fluctuation clustering interval, combines dietary structure, body temperature and exercise status, compares the blood glucose change process after eating, identifies key nodes of blood glucose fluctuation and decline, and obtains the dynamic stable interval of behavior.

[0009] The behavioral deviation analysis module compares the blood glucose performance after newly collected food intake items based on the dynamic stability interval of the behavior, analyzes the differences between the new behavior and the previous pattern, judges the relationship between the behavior and the abnormal response, and obtains the dynamic feature quantity of the deviation.

[0010] The intervention level determination module determines whether the behavior time interval is within the fluctuation cluster interval based on the offset dynamic feature quantity, screens eating behavior, and compares the impact of different behaviors in combination with diet, exercise and blood glucose decline to obtain the behavior intervention priority level.

[0011] On the other hand, the cross-segment fluctuation feature group includes fluctuation morphology, interval persistence structure, and change frequency distribution; the fluctuation clustering interval includes high-density period, risk classification number, and cluster affiliation attribute; the behavioral dynamic stability interval includes response stability index, fluctuation recovery parameter, and duration attribute; the offset dynamic feature quantity includes offset pattern type, response anomaly factor, and behavioral feature label; and the behavioral intervention priority level includes level order parameter, intervention implementation level, and priority grouping.

[0012] On the other hand, the fluctuation trend identification module includes:

[0013] The data stream receiving submodule analyzes the blood glucose monitoring data collected continuously for hours, compares the monitoring data information within each hour, determines the fluctuation boundary of blood glucose level in each time period, summarizes the monitoring time of the maximum and minimum points in each time period, and obtains the blood glucose boundary reference factor.

[0014] The trend inflection point extraction submodule, based on the blood glucose boundary reference factor, compares the temporal changes of the boundaries of each time period, analyzes the switching of blood glucose trends between each two adjacent time periods, determines the time when the blood glucose rising and falling trends change, and obtains the trend inflection sequence.

[0015] The cross-segment feature construction submodule analyzes the blood glucose fluctuation performance in each consecutive time period based on the trend turning sequence, determines the pattern type of trend change, classifies the fluctuation structure characteristics of each time period, and obtains the cross-segment fluctuation feature group.

[0016] On the other hand, the aggregation window screening module includes:

[0017] The window structure construction submodule, based on the cross-segment fluctuation feature group, compares the blood glucose fluctuation structure in each time period, judges the continuity of fluctuation performance between consecutive time periods, identifies time segments with consistent fluctuation patterns, optimizes the changes in blood glucose fluctuation curves within the time window, and obtains continuous fluctuation groups.

[0018] The trend pattern comparison submodule compares the trend characteristics of fluctuation changes in each window based on the continuous fluctuation group, analyzes the dominant direction of blood glucose level changes within the window, judges the correlation of trend paths between windows, identifies the distribution sequence of trend change points, and calculates the continuity index of trend patterns to obtain a trend consistency sequence.

[0019] The clustering interval identification submodule, based on the trend consistency sequence, determines the segments in the continuous window sequence that exhibit synchronous fluctuation characteristics, identifies time intervals with consistent trends, optimizes the time range marking of the window sequence, and obtains the fluctuation clustering interval.

[0020] On the other hand, the feeding response discrimination module includes:

[0021] The behavioral data extraction submodule compares the dietary structure, eating time and exercise performance recorded in the eating behavior data based on the fluctuation cluster interval, determines the time correspondence between eating events and blood glucose fluctuation cycles, identifies the synchronization of dietary components, body temperature and exercise records in each time period, and obtains a set of eating behavior features.

[0022] The dynamic curve comparison submodule compares the eating behavior feature set with the eating event and the monitored blood glucose curve, analyzes the dynamic changes in the continuous blood glucose monitoring data after eating, determines the blood glucose level rise and fall segments, filters the change trend of each curve segment, and obtains the blood glucose response curve index.

[0023] The stable interval identification submodule, based on the blood glucose response curve index, determines the time interval in which blood glucose fluctuations gradually tend to stabilize, filters out segments where the fluctuation range continuously converges, sorts out the food event numbers and the time sequence of each interval, and obtains the dynamic stable interval of behavior.

[0024] On the other hand, the behavioral deviation analysis module includes:

[0025] The new behavior comparison submodule analyzes the blood glucose monitoring sequence corresponding to the newly collected eating behavior based on the dynamic stable interval of the behavior, compares the similarities and differences between the blood glucose change trend under the new behavior and the stable interval, judges the blood glucose response change caused by the new behavior, identifies the monitoring parameters that differ in different time periods, and obtains the new behavior difference feature quantity.

[0026] The cooperation relationship determination submodule, based on the new behavioral difference feature quantity, judges the medication records and exercise interruption information within the corresponding time period, analyzes the synergy of food, drug and exercise factors at each time node, identifies the interaction performance of behavioral factors under the combination of different time sequences, and obtains a set of behavioral synergy indicators.

[0027] The abnormal feature identification submodule analyzes the relationship between each behavior and blood glucose response based on the behavioral coordination index set, judges the blood glucose deviation performance under each interaction combination, identifies the intervals and features exhibiting abnormal fluctuations, and obtains the dynamic feature quantity of the deviation.

[0028] On the other hand, the intervention level determination module includes:

[0029] The time interval discrimination submodule determines whether the time period of each behavioral event coincides with the fluctuation aggregation interval based on the offset dynamic feature quantity, filters the feeding behaviors associated within the time interval, and obtains the behavioral event interval index.

[0030] The factor correlation analysis submodule, based on the behavioral event interval index, compares the dietary categories, exercise performance, and blood glucose decline process involved in eating behavior, analyzes the correlation between changes in dietary structure, exercise intervention, and blood glucose fluctuations, determines the strength of the synergistic effect of each behavioral factor, and obtains the correlation degree of behavioral effects.

[0031] The priority output submodule determines the driving strength of multiple behavioral interventions on blood glucose dynamics based on the correlation of the behavioral effects, compares the differences in the performance of various intervention measures under different combinations of behavioral factors, identifies the intervention content with the dominant influence, and obtains the priority level of the behavioral intervention.

[0032] On the other hand, the direction of blood glucose change refers to the point in continuous data where blood glucose changes from rising to falling or from falling to rising, i.e., the inflection point of blood glucose trend. The fluctuation pattern refers to the trend and structural characteristics of blood glucose level changes over time within the current time window.

[0033] On the other hand, the central tendency refers to the consistent fluctuations of multiple time windows within a certain period of time; the eating behavior data refers to the eating information entered by the user in the recording platform or device; and the blood glucose fluctuation refers to the situation where the blood glucose value rises or falls significantly in a short period of time after eating.

[0034] On the other hand, an AI-driven computer-based intelligent analysis method for blood glucose monitoring is provided. This method is applied to an AI-driven computer-based intelligent analysis system for blood glucose monitoring and includes the following steps:

[0035] S1: Based on blood glucose monitoring data, determine the direction of blood glucose changes every hour, identify the fluctuation characteristics of each time period by comparing the trends of high and low points in each time period, and obtain the cross-segment fluctuation characteristic group.

[0036] S2: Based on the cross-segment fluctuation feature group, compare the fluctuation patterns of the current window with those of the same period in history, determine whether there is a concentrated trend in multiple consecutive windows, identify windows that meet the features, and obtain the fluctuation clustering interval.

[0037] S3: Based on the fluctuation clustering interval, analyze the eating behavior data within the interval, combine dietary structure, body temperature and exercise, compare the blood glucose change process after eating, identify the key nodes of blood glucose fluctuation and decline, and obtain the dynamic stable interval of behavior.

[0038] S4: Based on the dynamic stable interval of the behavior, compare the blood glucose performance after the newly collected food intake item by item, analyze the difference between the new behavior and the previous pattern, determine the relationship between the behavior and the abnormal response, and obtain the offset dynamic feature quantity;

[0039] S5: Based on the offset dynamic feature, determine whether the behavior time interval is within the fluctuation cluster interval, screen eating behavior, and compare the impact of different behaviors in combination with diet, exercise and blood glucose decline to obtain the behavior intervention priority level.

[0040] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0041] By dynamically decomposing blood glucose monitoring data and fusing interval fluctuation characteristics, the system enables trend clustering of blood glucose changes over continuous periods and proactive identification of high-risk intervals. It also facilitates collaborative analysis of blood glucose responses to events such as eating, exercise, and medication using multi-source behavioral information. Abnormal signals are traced and determined through multi-parameter interactive verification. Behavioral deviation factors and risk impacts are quantified and stratified in real time, enhancing the ability to correlate blood glucose changes with various behavioral scenarios. This dynamically optimizes the intervention priority of behavioral events and promotes individualized, tiered responses and proactive adjustments in chronic disease health management scenarios. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the system of the present invention;

[0044] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0045] Figure 3 This is a flowchart of the fluctuation trend recognition module of the present invention;

[0046] Figure 4 This is a flowchart of the aggregation window screening module of the present invention;

[0047] Figure 5 This is a flowchart of the feeding response discrimination module of the present invention;

[0048] Figure 6 This is a flowchart of the behavior deviation analysis module of the present invention;

[0049] Figure 7 This is a flowchart of the intervention level determination module of the present invention;

[0050] Figure 8 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0053] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0054] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0055] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0056] This invention provides an AI-driven intelligent computer-aided blood glucose monitoring and analysis system, such as... Figure 1 As shown, the system includes:

[0057] The fluctuation trend identification module analyzes blood glucose monitoring data collected continuously for hours, determines the change in the direction of blood glucose change within each hour, compares the changes in the high and low points of each time period segment by segment, compares the blood glucose fluctuation performance of each hour, identifies the main characteristics of the fluctuation performance of continuous time periods, and sequentially judges the change amplitude of the fluctuation performance of each time period to obtain the cross-segment fluctuation feature group.

[0058] The clustered window screening module is based on cross-segment fluctuation feature groups. It compares the fluctuation change patterns of the current time window with those of the same period in history, analyzes their respective change processes, determines whether the fluctuation trends of multiple consecutive time windows show concentrated characteristics, screens windows that meet the judgment conditions, and obtains the fluctuation clustering interval.

[0059] The food intake response discrimination module analyzes the food intake behavior data corresponding to the interval marker based on the fluctuation cluster interval. By judging the user's diet structure, body temperature and exercise performance, it compares the dynamic change curve of blood glucose after eating, identifies the key time points of rapid blood glucose fluctuation and decline process, and obtains the dynamic stable interval of behavior.

[0060] The behavioral deviation analysis module compares the dynamic blood glucose performance after new eating behaviors based on the dynamic stability interval of behavior, analyzes the differences between the new behaviors and previous patterns, and judges the coordination relationship between eating behavior, drug effects and exercise intervention by combining medication timing and exercise interruption information, identifies abnormal response types, and obtains deviation dynamic feature quantities.

[0061] The intervention level determination module uses the offset dynamic feature quantity to determine whether the time interval corresponding to the behavior is in the fluctuation cluster interval, screens the eating behaviors belonging to the corresponding time interval, analyzes the food category, exercise performance and blood glucose decline process in sequence, compares the correlation degree of the difference behavioral factors, and obtains the behavioral intervention priority level.

[0062] The cross-segment fluctuation characteristic group includes fluctuation morphology, interval persistence structure, and frequency distribution of changes; the fluctuation cluster interval includes high-density periods, risk classification number, and cluster affiliation attribute; the behavioral dynamic stability interval includes response stability index, fluctuation recovery parameter, and duration attribute; the offset dynamic characteristic quantity includes offset pattern type, response anomaly factor, and behavioral characteristic label; and the behavioral intervention priority level includes level order parameter, intervention implementation level, and priority grouping.

[0063] In the fluctuation trend recognition module, blood glucose monitoring data refers to the raw collection sequence of blood glucose values ​​obtained hourly by a continuous blood glucose monitoring instrument (such as CGM / finger-prick blood glucose meter), representing the real-time blood glucose level record of the user within a certain period. The change in the direction of blood glucose change refers to the point in the continuous data where blood glucose changes from rising to falling or from falling to rising, i.e., the inflection point of the blood glucose trend, used to describe the rhythm of blood glucose curve changes. Segmental comparison refers to dividing the monitoring period into several continuous time periods and comparing the start and end levels and change patterns of blood glucose changes in each time period to form a horizontal comparison between time periods. Blood glucose fluctuation performance refers to the fluctuation, trend and intensity of blood glucose levels in each hour or each short period of time, which is a quantitative description of the characteristics of the raw blood glucose data. The main features refer to the core attributes reflecting the characteristics of blood glucose fluctuations extracted after analyzing the blood glucose change performance of multiple continuous time periods, such as fluctuation frequency, amplitude, and rate of change. The amplitude of change refers to the difference between the maximum and minimum blood glucose values ​​in a certain period, reflecting the intensity of blood glucose in that period.

[0064] In the clustering window screening module, the fluctuation pattern refers to the overall trend and structural characteristics of blood glucose level changes over time within the current time window (e.g., a three-hour period) (e.g., continuous rise, continuous fall, rapid oscillation, etc.). The individual change process refers to the complete dynamic change process of blood glucose fluctuation from start to finish within the current analysis window and the historical same period window. The fluctuation trend refers to the overall trend of blood glucose within a certain time window (e.g., upward trend, downward trend, stable fluctuation or violent fluctuation, etc.). The clustering characteristic refers to the phenomenon that the fluctuation changes of multiple time windows tend to be consistent within a certain period of time, and a large-scale synchronous change occurs. It is usually used to identify the risk of abnormal blood glucose clustering. The window that meets the judgment criteria refers to a set of continuous time windows that are judged to meet the clustering characteristic criteria (e.g., fluctuation amplitude, synchronous trend, etc.) during the comparative analysis process.

[0065] In the food intake response discrimination module, food intake behavior data refers to the eating-related information entered by the user in the recording platform or device, including the time of eating, food category, nutritional components (such as the proportion of carbohydrates, fats, and proteins), and physiological and behavioral information related to the eating behavior. The blood glucose dynamic change curve refers to the curve formed by the change of blood glucose level monitored after eating over time, which is used to reflect the real-time impact of eating on blood glucose. Rapid blood glucose fluctuation refers to the situation where blood glucose value rises or falls significantly in a short period of time after eating, which is common in the rapid rise of blood glucose after high carbohydrate intake. The decline process refers to the process in which blood glucose level gradually decreases and tends to stabilize after reaching a high point, reflecting the role of insulin secretion and metabolic regulation. Key time points refer to representative nodes in the blood glucose change curve, such as the start of blood glucose rising, reaching the peak, and the start of decline, which are important time coordinates used to discern the blood glucose response process.

[0066] In the behavioral deviation analysis module, newly collected eating behavior refers to the latest recorded data on a user's eating behavior, used for comparative analysis with previous eating patterns. Past patterns refer to the statistical regularities, trends, and stable intervals of a user's multiple eating behaviors and blood glucose responses throughout history. Medication timing refers to data related to the actual time sequence of medication (such as oral hypoglycemic agents or insulin) taken before and after each eating session. Exercise interruption information refers to records of whether the user has any unplanned exercise pauses or terminations during eating or within the blood glucose response cycle. Coordination relationship refers to the degree of interaction and synergy between eating behavior, medication administration, and exercise intervention in terms of time and behavior. Abnormal response type refers to atypical blood glucose changes caused by behavioral deviations, imbalances in medication or exercise intervention, such as abnormally high blood glucose levels or insufficient blood glucose reduction.

[0067] In the intervention level determination module, the time interval corresponding to the behavior refers to the time correspondence between the time of a specific eating behavior and the risk period of blood glucose accumulation, which is used to locate the blood glucose risk level when the behavior occurs. The corresponding time period refers to the time range in which risk fluctuations are determined. Eating behaviors falling into this time period are considered as key analysis objects. Differentiated behavioral factors refer to various behavioral variables that lead to different blood glucose responses, such as dietary structure, exercise performance, and lifestyle interventions. The degree of correlation refers to the strength of the correlation between each behavioral factor and blood glucose changes, which is usually obtained through statistical analysis or logical judgment.

[0068] like Figure 2 and Figure 3 As shown, the fluctuation trend identification module includes:

[0069] The data stream receiving submodule analyzes the blood glucose monitoring data collected continuously for hours, compares the monitoring data information within each hour, determines the fluctuation boundary of blood glucose level in each time period, summarizes the monitoring time of the maximum and minimum points in each time period, and obtains the blood glucose boundary reference factor.

[0070] The raw blood glucose values ​​were divided into multiple independent time periods by hour. Within each time period, all recorded blood glucose values ​​and their corresponding time information were extracted. Then, the blood glucose values ​​within each hour were compared one by one to identify the maximum and minimum blood glucose values ​​within that hour, and the specific time position of the extreme values ​​was recorded. For example, if the blood glucose values ​​recorded between 08:00 and 09:00 were 5.2, 5.9, 6.7, 7.4, 6.1, and 5.6, the maximum value for that hour was 7.4, corresponding to 08:30, and the minimum value was 5.2, corresponding to 08:00. After extracting the maximum and minimum values ​​and their times, the time interval between these two time points was used to determine the difference. This is used to reflect whether the time location of fluctuations is concentrated. If the time difference between two time points is 30 minutes or more, it is marked as a cross-segment fluctuation; otherwise, it is marked as a concentrated fluctuation. Next, the difference between the maximum and minimum blood glucose values ​​within each hour is numerically determined. If the difference is equal to or greater than 2.5, it is classified as a severe fluctuation; if it is between 1.0 and 2.5, it is classified as a moderate fluctuation; and if it is less than 1.0, it is classified as a slight fluctuation. This determination standard is based on empirical extraction and setting of the daily fluctuation range of continuous blood glucose monitoring results. If it is found that there are fewer than 4 valid monitoring points in a certain hourly segment during the determination process, the hourly segment will not be included in the analysis to avoid identification bias due to insufficient data. This yields the blood glucose boundary reference factor.

[0071] The trend inflection point extraction submodule is based on the blood glucose boundary reference factor. It compares the temporal changes of the boundary of each time period, analyzes the switching of blood glucose trends between each two adjacent time periods, determines the time when the blood glucose rising and falling trends change, and obtains the trend inflection sequence.

[0072] After receiving the blood glucose boundary reference factors for each hourly segment, the times of maximum and minimum values ​​in any two adjacent hourly segments are extracted sequentially. By comparing these two time points, if the maximum value of the previous hourly segment occurs in the latter half of the segment, while the minimum value of the subsequent hourly segment occurs in the first half, it indicates a shift in blood glucose from an upward trend to a downward trend. After identifying this shift, the midpoint between the maximum and minimum values ​​is recorded as the trend inflection point and added to the trend inflection point sequence. If the opposite occurs, i.e., the minimum value of the previous hourly segment occurs in the latter half, and the maximum value of the subsequent hourly segment occurs in the first half, then... When a clear trend changes from downward to upward, the midpoint of this time frame is also recorded as the turning point. During the judgment process, only when the difference in blood glucose fluctuations between two hours exceeds 1.5 will the time point be judged as a valid turning point, filtering out trend switching errors caused by slight fluctuations. At the same time, a trend consistency constraint is set, that is, only when the blood glucose trend is consistent for two consecutive hours will subsequent hours be allowed to be included in the trend switching judgment. For example, if two consecutive hours are in an upward trend, subsequent hours are allowed to be included in the turning point judgment range; otherwise, the hour is ignored. This method ensures the continuity and rationality of the trend turning sequence, forming a set of time points.

[0073] The cross-segment feature construction submodule analyzes blood glucose fluctuations in each consecutive time period based on trend reversal sequences, determines the pattern type of trend change, classifies the fluctuation structure characteristics of each time period, and obtains cross-segment fluctuation feature groups.

[0074] The entire monitoring time is divided into several sub-segments based on the intervals between inflection points. Each sub-segment consists of two inflection point times. By reading complete blood glucose monitoring records within each segment, the blood glucose values ​​within that segment are compared, and the maximum, minimum, average, initial, and final values ​​are extracted. Then, the number of blood glucose fluctuations within that segment is determined, i.e., the total number of times the blood glucose trend changes from rising to falling within that segment. The fluctuation frequency density is then calculated based on the duration of the segment. When the fluctuation frequency density exceeds a certain standard, the segment is marked as a high-frequency fluctuation segment; if it is within the intermediate standard range, it is marked as a medium-frequency fluctuation segment; and so on. The lowest standard is considered a low-frequency fluctuation segment. The higher the frequency density, the more frequent the blood glucose fluctuations within that segment. In addition, it is necessary to determine the overall trend within the segment, that is, the direction of blood glucose value change from the start to the end of the segment. If the overall value continues to decrease, it is marked as a decreasing type. If it rises in the middle and then falls, it is marked as a peak type. If the overall value is close to stable, it is considered as a plateau type. At the same time, compare the blood glucose difference between the start and end of the segment. If the change exceeds 2.0, it is marked as a severe mode. If it is between 1.0 and 2.0, it is a moderate mode. If it is below 1.0, it is a moderate mode. Each segment is processed according to the above steps once to form a cross-segment fluctuation characteristic group.

[0075] like Figure 2 and Figure 4 As shown, the clustered window screening module includes:

[0076] The window structure construction submodule is based on the cross-segment fluctuation feature group. It compares the blood glucose fluctuation structure in each time period, judges the continuity of fluctuation performance between consecutive time periods, identifies time segments with consistent fluctuation patterns, optimizes the changes in blood glucose fluctuation curves within the time window, and obtains continuous fluctuation blocks.

[0077] The system reads the fluctuation structure type information for each time period in the feature group, retrieves the start and end times, fluctuation density, trend pattern, and frequency index recorded for each segment, arranges the structural parameters in chronological order, and then compares the fluctuation structure segment by segment between adjacent time periods. The system calculates the numerical difference between the fluctuation density values ​​of any two adjacent segments. If the difference does not exceed 0.01 and the trend patterns of the two segments are identical, they are considered structurally similar segments. If structural similarity occurs in three or more consecutive time periods, these segments are grouped into a single window unit. When constructing the window, the continuous start and end times of the time periods must be read and merged into a unified time segment. For example, if the three time periods 08:00–08:30, 08:30–09:00, and 09:00–09:30 have consecutive frequency densities of 0.0 and 0.0 respectively... 56, 0.055, and 0.057 all show an increasing trend pattern, so they are considered to meet the merging criteria and are uniformly defined as a fluctuation window of 08:00–09:30. Subsequently, the same merging judgment is performed on all subsequent time periods. The blood glucose curves in each merged time segment window are smoothed again. The average value of the blood glucose value sequence in each group of windows is taken in 5-minute increments to form a more stable window blood glucose curve. If the frequency of fluctuation of blood glucose value in the window does not exceed one after merging, the window is marked as structurally stable. If the frequency of fluctuation exceeds two, it is marked as structurally unstable. At the same time, the number of fluctuation extreme points in each window is counted. If the window duration exceeds 60 minutes and there are no less than 4 extreme points, the window is considered to be highly volatile, forming a continuous fluctuation block.

[0078] The trend pattern comparison submodule is based on continuous fluctuation blocks. It compares the trend characteristics of fluctuation changes in each window, analyzes the dominant direction of blood glucose level changes within the window, judges the correlation of trend paths between windows, identifies the distribution sequence of trend change points, and calculates the trend pattern continuity index to obtain the trend consistency sequence.

[0079] Read the blood glucose curve trend for each window, extract the blood glucose value sequence at 5-minute intervals from the start time to the end time, calculate the total duration of the rising and falling segments in sequence and accumulate them separately, compare the rising duration with the falling duration, if the rising time is longer than the falling time by more than 15 minutes, the dominant direction of that window is defined as rising, otherwise it is defined as falling, if the difference is less than 5 minutes, it is defined as stable. After determining the dominant direction of each window, select the dominant direction between two adjacent windows for matching, if the directions are the same, it is considered a continuous path, if they are different, it is marked as a path switching point, and the switching point is recorded at time. The specific location in the time sequence is determined by performing this process sequentially on all window pairs, forming a distribution sequence of trend change points. In this sequence, the time interval of each directional change is calculated. If the time interval of three consecutive changes is less than 30 minutes, it is determined to be a high-frequency change zone. If the interval is more than 60 minutes, it is a low-frequency change zone. The intermediate range is defined as a medium-frequency change zone. Then, the number of each type of frequency zone in the entire sequence is counted. If the proportion of high-frequency change zones exceeds 60% of the total number of window pairs, the trend consistency sequence is marked as unstable. If the proportion of low-frequency change zones exceeds 70%, it is stable. The rest are considered as transitional. The trend consistency sequence is then output.

[0080] The clustering interval identification submodule is based on the trend consistency sequence to determine the segments in the continuous window sequence that exhibit synchronous fluctuation characteristics, identify the time intervals with consistent trends, optimize the time range labeling of the window sequence, and obtain the fluctuation clustering intervals.

[0081] The switching point time and change frequency type in the trend consistency sequence are read one by one. Each group of windows that continuously show the same trend path and are marked as stable are grouped into a candidate synchronization segment. If the candidate segment contains more than three consecutive windows with completely consistent dominant directions, it is judged as a segment with synchronous fluctuation characteristics. In the judgment process, the start time difference and end time difference of the windows are also compared. If the difference between the start time of any two windows does not exceed 15 minutes and the difference between the end time does not exceed 20 minutes, it is considered to have continuity in time structure and meet the time interval requirements of trend consistency. At this time, the windows are merged into a cluster. The time interval is divided into units, and the start and end times of the units are re-recorded as the boundaries of the time interval. At the same time, the order of the windows in the original sequence is compared. If all window numbers are consecutive integers, they are marked as sequential clusters; otherwise, they are marked as skip clusters. Each cluster interval needs to be checked again to see if there is a reversal in the direction of blood glucose fluctuations. If there is a reversal, the window segment is split from the reversal point, the candidate segment is reconstructed, and the above screening process is repeated to select all window segments that meet the requirements of stability, direction consistency, time structure continuity, and sequential number continuity. Each segment is assigned a new interval number, and the fluctuation cluster interval is output.

[0082] like Figure 2 and Figure 5 As shown, the feeding response discrimination module includes:

[0083] The behavioral data extraction submodule compares the dietary structure, eating time and exercise performance recorded in the eating behavior data based on the fluctuation cluster interval, determines the temporal correspondence between eating events and blood glucose fluctuation cycles, identifies the synchronization of dietary components, body temperature and exercise records in each time period, and obtains a set of eating behavior features.

[0084] The time boundary information is extracted segment by segment from the cluster intervals. The start and end times of each interval are compared with the user's recorded eating behavior data. The eating time, food name, nutritional composition, and exercise performance records marked in the eating records are retrieved and matched using time as an index. During the matching process, it is determined whether the time of each eating event falls within the start and end range of the blood glucose fluctuation cluster interval. If the time overlap is no more than 5 minutes, a direct correspondence is determined between the eating event and the cluster interval; if it exceeds 5 minutes but is less than 15 minutes, an indirect correspondence is determined. After confirming the correspondence, the dietary structure parameters of the eating event are read, and the proportions of carbohydrates, fats, and proteins are compared. If the carbohydrate proportion exceeds 60%, the event is marked as high-carbohydrate intake; if it is between 40% and 60%, it is moderate intake; and if it is below 40%, it is low-carbohydrate intake. Simultaneously, body temperature records during the time period of the eating event are read. A body temperature above 37.2℃ is marked as an elevated body temperature state; a temperature between 36.5℃ and 37.2℃ is considered a stable state; and a temperature below 36.5℃ is considered a low body temperature state. Then, the exercise data within that time period is retrieved, and the duration of exercise 30 minutes before and after eating is calculated. If the exercise time before eating exceeds 20 minutes and after eating is less than 10 minutes, it is recorded as "pre-exercise dominant"; otherwise, it is recorded as "post-exercise dominant." If the time difference before and after is less than 5 minutes, it is marked as "balanced exercise." The three labels—diet type, body temperature state, and exercise performance—are integrated into a complete record according to the eating event number. For example, if eating occurs at 08:00, with a carbohydrate ratio of 65%, a body temperature of 37.3℃, and exercise lasts for 25 minutes before and 10 minutes after, the event is labeled as high carbohydrate, elevated body temperature, and pre-exercise dominant. This judgment logic is executed one by one for each eating event within each cluster interval and output as a set, forming a set of eating behavior characteristics.

[0085] The dynamic curve comparison submodule is based on the set of eating behavior characteristics. It compares eating events with blood glucose monitoring curves, analyzes the dynamic changes in continuous blood glucose monitoring data after eating, determines the blood glucose level rise and fall segments, filters the change trends of each curve segment, and obtains blood glucose response curve indicators.

[0086] Read the time tag of each feeding event in the dataset and retrieve the corresponding blood glucose curve segment from the continuous blood glucose monitoring data. Set the observation period as 3 hours after the start of eating. Arrange the blood glucose values ​​every 5 minutes within this period in chronological order, and calculate the difference between adjacent data points to determine the direction of blood glucose change. If three consecutive differences are positive, it is defined as an upward segment; if three consecutive differences are negative, it is defined as a downward segment; otherwise, it is a fluctuating segment. Identify the upward and downward segments sequentially throughout the observation period and record their start and end times. When the duration of the upward segment is greater than 30 minutes and the blood glucose change amplitude is greater than 2, the segment is marked as a significant upward segment. If the duration of the downward segment is greater than 40 minutes and the amplitude is greater than 1.5, it is marked as a significant downward segment. Perform analysis on the upward and downward segments of each feeding event. Matching is performed such that if the time interval between two events does not exceed 1 hour, they are considered to be the same response process; otherwise, they are recorded as independent response curves. Within the same response curve, the time interval from the start of the rising segment to the highest point is calculated as the peak interval, and the time interval from the highest point to the end of the decline is calculated as the decline interval. For example, if blood glucose starts to rise 15 minutes after eating, reaches its peak at 45 minutes, and returns to near baseline at 110 minutes, then the peak interval is 30 minutes and the decline interval is 65 minutes. Subsequently, the number and duration of each curve segment are counted. Events with peak intervals shorter than 20 minutes are classified as rapid response type, those between 20 and 50 minutes as medium response type, and those exceeding 50 minutes as slow response type. The blood glucose response curve index is generated by comparing the corresponding diet type and exercise state in the set of eating behavior characteristics.

[0087] The stable interval identification submodule uses the blood glucose response curve index to determine the time interval in which blood glucose fluctuations gradually tend to stabilize, filters out segments where the fluctuation range continuously converges, sorts out the food event numbers and the time sequence of each interval, and obtains the dynamic stable interval of behavior.

[0088] Read the blood glucose change sequence from the response curve of each eating event, and determine the direction of change in the difference between adjacent measurements point by point. If the change amplitude is less than 0.3 for four consecutive times, it is determined to have entered a stable phase. Record the first time point that meets the condition as the start time of stability. Then, detect the change in the difference between subsequent adjacent values. If the change amplitude exceeds 0.5 for three consecutive times, it is considered the end time of the stable phase. Define the interval between the start and end times as the blood glucose fluctuation stabilization segment. Arrange multiple stable segments within the same eating event in chronological order and assign numbers. For example, if the blood glucose change remains within the range of 6 to 6.3 between 90 minutes and 130 minutes after eating, then... This interval is marked as stable interval 1. If the change is between 5.8 and 6.1 between 180 and 210 minutes, it is stable interval 2. Then, each feeding event number and its corresponding stable intervals are sorted in chronological order. If the interval between two stable intervals is less than 30 minutes, they are merged into a continuous segment; otherwise, they are kept as independent records. The duration of the stable segment of each feeding event is then calculated. If the duration exceeds 40 minutes, it is recorded as long-term stable; between 20 and 40 minutes, it is recorded as medium-term stable; and less than 20 minutes, it is recorded as short-term stable. The numbers, start and end times, and duration levels of all feeding events are sorted in sequence to form the behavioral dynamic stable intervals.

[0089] like Figure 2 and Figure 6 As shown, the behavioral deviation analysis module includes:

[0090] The new behavior comparison submodule analyzes the blood glucose monitoring sequence corresponding to the newly collected eating behavior based on the dynamic stable interval of the behavior, compares the similarities and differences between the blood glucose change trend under the new behavior and the stable interval, judges the blood glucose response change caused by the new behavior, identifies the monitoring parameters that differ in different time periods, and obtains the differential feature quantity of the new behavior.

[0091] The system reads the latest collected eating behavior data and extracts features such as eating time, food type, carbohydrate content, fat ratio, and protein ratio. It then matches these features with corresponding blood glucose monitoring sequences in chronological order, comparing the blood glucose trend of this new behavior with the typical curve of the historical stable range. By comparing the starting point, peak time, and end time of the decline in the two sets of data, the direction and rate of blood glucose change caused by the new behavior are determined. During execution, the average fluctuation range recorded in the historical stable range is set as the benchmark range. A blood glucose rise exceeding 20% ​​of the benchmark is defined as a high response, and a rise below 20% is defined as a low response. A rise within this range is considered a normal response. For example, if the average rise in the historical stable range is 2, a rise of 2.6 in blood glucose during the new behavior is considered a high response. The comparison process also requires detecting the time distribution of the blood glucose curve during the rising and falling phases of the new behavior. If the rising duration is less than 30% of the historical average, it is identified as a rapid rise type; if it exceeds 50%, it is identified as a slow rise type. At the same time, the rate of decline after the peak is calculated. If the rate of decline is less than 0.8 times the historical average rate, it is considered a delayed decline. The blood glucose values ​​at each detection point are compared, and the monitoring time with a difference of more than 0.5 is extracted as the difference node. The nodes are marked on the time axis to form a time difference sequence. Then, it is matched with the food energy density parameters in the food intake record. If a sustained high response occurs under high energy intake events, it is marked as a high-intensity behavioral difference type; if a low response occurs under low energy intake events, it is marked as a low-intensity difference type. The difference items are arranged in chronological order to form a set of new behavioral difference characteristic quantities.

[0092] The coordination relationship determination submodule uses new behavioral difference features to determine medication records and exercise interruption information within the corresponding time period, analyzes the coordination of food, drug and exercise factors at each time node, identifies the interaction performance of behavioral factors under the combination of different time sequences, and obtains a set of behavioral coordination indicators.

[0093] Extract the time intervals corresponding to the differential feature values, and read the medication records and exercise information within these time intervals. Compare the order of food intake time, medication administration time, and exercise interruption time one by one to determine the degree of overlap among the three in the time dimension. If the medication administration time is 10 to 30 minutes earlier than the food intake time, and the exercise interruption event occurs within 20 minutes after the food intake, it is recorded as "pre-medication exercise interruption type". If the time difference between medication administration and food intake is less than 5 minutes and no exercise interruption occurs, it is recorded as "synchronous behavior type". If the food intake time is more than 15 minutes earlier than the medication administration time, it is recorded as "delayed medication type". In each type, the time interval values ​​between food intake, medication, and exercise are counted separately, and the interval length is divided into closely coordinated (interval) types. The study categorized participants into three groups: those with close cooperation (less than 10 minutes), those with moderate cooperation (intervals between 10 and 30 minutes), and those with loose cooperation (intervals exceeding 30 minutes). It also determined whether the peak time corresponding to the blood glucose fluctuation occurred during medication or exercise. If the peak time coincided with the medication recording time, the medication was considered to have influenced the fluctuation. If the peak time was within 10 minutes before or after exercise interruption, it was marked as an exercise interruption-related influence. Furthermore, the percentage of each cooperation category was calculated based on its frequency. If close cooperation accounted for more than 60%, overall behavioral coordination was high; if loose cooperation accounted for more than 50%, behavioral coordination was low. Based on these criteria, the time correspondence between food intake, medication, and exercise was recorded in a node format to generate a behavioral coordination index.

[0094] The abnormal feature identification submodule analyzes the relationship between each behavior and blood glucose response based on the behavioral coordination index set, judges the blood glucose deviation performance under each interaction combination, identifies the intervals and features that show abnormal fluctuations, and obtains the dynamic feature quantity of the deviation.

[0095] Read the behavioral combination data from each synergistic indicator, extract the correspondence between the time nodes of food intake, medication, and exercise and the blood glucose response interval for each combination, align the blood glucose monitoring sequence with the behavioral time nodes, and take 30 minutes of monitoring data before and after each node to calculate the range of blood glucose changes and determine whether it exceeds the normal fluctuation range. When the fluctuation amplitude of two consecutive nodes is greater than 3, it is recorded as a strong abnormal segment; when only one node exceeds 2, it is recorded as a moderate abnormal segment; and when it is less than 1, it is a slight fluctuation. Then, detect the fluctuation direction within the abnormal segment. If blood glucose continues to rise for more than 60 minutes without a downward trend, it is defined as a persistently high abnormality. If it fails to stabilize within two hours after decreasing, it is defined as... For delayed stabilization anomalies, and by comparing the type of food intake with the timing of drug action, if a high-carbohydrate intake pattern accompanied by delayed medication results in an abnormally high level, the segment is labeled as "delayed drug effect type." If a low-carbohydrate intake is accompanied by exercise interruption and a short-term increase, it is labeled as "exercise interruption triggered type." All abnormal segments are arranged in chronological order of occurrence. The duration and maximum fluctuation amplitude of each abnormal segment are calculated and assigned a number. For example, if blood glucose rises from 5.5 to 8.8 between 10:00 and 11:30 and does not fall back, it is labeled as abnormal segment A1, lasting 90 minutes with an amplitude of 3.3. Finally, all numbered segments, their corresponding behavioral combinations, abnormal types, and durations are summarized to form the offset dynamic characteristic quantity.

[0096] like Figure 2 and Figure 7 As shown, the intervention level determination module includes:

[0097] The time interval discrimination submodule uses offset dynamic features to determine whether the time period of each behavioral event coincides with the fluctuation clustering interval, filters the feeding behaviors associated within the time interval, and obtains the behavioral event interval index.

[0098] Extract the start and end times of each offset dynamic feature to form an abnormal response time interval sequence. Then, read the start and end times of all segments from the generated fluctuation clustering intervals. Compare the two sets of intervals one by one according to the time axis. If the start time of an offset interval for a certain behavioral event is earlier than the end time of the fluctuation clustering interval and its end time is later than the start time of the clustering interval, it is considered that the time of the behavioral event overlaps with the clustering interval. Record the behavioral event number and its corresponding clustering segment number. If the offset interval overlaps with multiple clustering intervals, associate the behavioral event with each overlapping segment. At the same time, set the time overlap ratio as a filtering criterion. If the overlap between the offset interval and the clustering interval exceeds 50% of the total duration of the offset interval, then... The behavioral event is marked as a "primary overlapping event". If it is between 30% and 50%, it is a "secondary overlapping event". If it is less than 30%, no index relationship is established. During the processing, the feeding event time needs to be aligned with the corresponding offset interval to determine whether the feeding record occurred within the overlapping segment. If the feeding time point is between the start and end times of the overlapping segment and does not deviate by more than 10 minutes, then an index relationship is established between the feeding event and the clustering segment. For example, if feeding behavior A occurs at 09:30, the offset interval is 09:20–10:20, and the fluctuation clustering segment is 09:00–10:00, the overlap is 40 minutes, accounting for 67% of the offset segment. Then, feeding behavior A is identified as a primary overlapping event and an index is established with the clustering segment to form a behavioral event interval index.

[0099] The factor correlation analysis submodule, based on the behavioral event interval index, compares the dietary categories, exercise performance, and blood glucose decline process involved in eating behavior, analyzes the correlation between changes in dietary structure, exercise intervention, and blood glucose fluctuations, judges the strength of the synergistic effect of each behavioral factor, and obtains the correlation degree of behavioral effects.

[0100] Extract the corresponding eating behavior ID from each index record. Read the food category label, exercise performance record, and blood glucose data curve of the decline process for that eating behavior. Combine the behavior parameters by event ID. First, group the food categories according to carbohydrate ratio: if the carbohydrate ratio is greater than 60%, it is classified as high-carb; between 40% and 60%, it is classified as medium-carb; and below 40%, it is classified as low-carb. Then, read the exercise performance parameters under the same ID and calculate the total exercise time within 30 minutes before and after eating. If the total is less than 10 minutes, it is marked as "low exercise"; between 10 and 30 minutes, it is marked as "medium exercise"; and more than 30 minutes, it is marked as "high exercise". Then, analyze the trend of the blood glucose decline curve within 2 hours after eating. If the decline is less than 1 within 60 minutes after the peak, it is considered "weak decline"; a decline of 1 to 2 is considered "high decline". A decrease of 2 is considered a "moderate decline," while a decrease of more than 2 is considered a "strong decline." The dietary group, exercise level, and decline intensity of each behavior record are combined and coded. For example, high-carb + moderate exercise + moderate decline forms the code HCM. The frequency of each combination code within the same interval is then counted, and its correspondence with the abnormal types in the offset feature is recorded. If a combination occurs more than 40% of the total number of events and the corresponding abnormal type is mainly characterized by continuous increases, then the combination behavior is considered to have a strong positive synergistic effect. If a combination has a high frequency but the corresponding abnormal fluctuations are irregularly distributed, it is judged as a weak synergistic relationship. The correlation level is then calculated based on the distribution of each combination in different clustering intervals. A correlation level higher than 0.7 is judged as a strong synergistic behavior, and a correlation level lower than 0.3 is judged as a weak synergistic behavior. The correlation degree of the behavior effect is then output.

[0101] The priority output submodule determines the driving strength of various behavioral interventions on blood glucose dynamics based on the correlation of behavioral effects, compares the differences in the performance of various intervention measures under different combinations of behavioral factors, identifies the intervention content with the dominant influence, and obtains the priority level of behavioral interventions.

[0102] The correlation value and corresponding behavioral event number of each factor combination are read, and the proportion of the combined behavior in all events is extracted. The distribution proportion of each dietary structure in highly correlated behaviors is calculated separately. If high-carbohydrate intake accounts for more than 50% of highly correlated behaviors, it is determined to be an intervention-dominant dietary factor. Then, the dominant relationship between exercise behavior and blood glucose reduction is determined. If high exercise frequency accounts for more than 60% of strong reduction behaviors, exercise behavior is considered a major influencing factor. Scoring items are generated for the three behavioral interventions: diet, exercise, and blood glucose reduction. The scores are based on the frequency of factor occurrence and correlation value. The behavior is synthesized with dominant behaviors, and a scoring threshold range is set as follows: a score between 0.7 and 1.0 is high priority, 0.4 to 0.7 is medium priority, and below 0.4 is low priority. If a behavior factor appears in multiple behavior combinations and has a high score, it is identified as the global dominant intervention factor. For example, if protein intake is marked as a strong synergistic behavior in the HCM, HCH, and MCM combinations, then this dietary feature is output as the priority intervention target. All elements involved in behavior intervention are given priority numbers, and the numbering order is arranged from high priority to low priority to form a behavior intervention priority level.

[0103] like Figure 8 As shown, the AI-driven intelligent computer analysis method for blood glucose monitoring includes the following steps:

[0104] S1: Based on blood glucose monitoring data, determine the direction of blood glucose changes every hour, identify the fluctuation characteristics of each time period by comparing the trends of high and low points in each time period, and obtain the cross-segment fluctuation characteristic group.

[0105] S2: Based on the cross-segment fluctuation feature group, compare the fluctuation patterns of the current window with those of the same period in history, determine whether there is a concentrated trend in multiple consecutive windows, identify windows that meet the features, and obtain the fluctuation clustering interval;

[0106] S3: Based on the fluctuation clustering interval, analyze the eating behavior data within the interval, combine dietary structure, body temperature and exercise, compare the blood glucose change process after eating, identify the key nodes of blood glucose fluctuation and decline, and obtain the dynamic stable interval of behavior.

[0107] S4: Based on the dynamic stable interval of behavior, compare the blood glucose performance after newly collected food intake item by item, analyze the difference between the new behavior and the previous pattern, determine the relationship between behavior and abnormal response, and obtain the offset dynamic characteristic quantity;

[0108] S5: Based on the offset dynamic feature, determine whether the time interval of the behavior is in the fluctuation clustering interval, screen eating behavior, and combine diet, exercise and blood glucose decline to compare the impact of different behaviors and obtain the priority level of behavior intervention.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An AI-driven intelligent computer-aided blood glucose monitoring and analysis system, characterized in that: The system includes: The fluctuation trend recognition module, based on blood glucose monitoring data, determines the direction of blood glucose changes every hour. By comparing the trends of high and low points in each time period, it identifies the fluctuation characteristics of each time period and obtains cross-segment fluctuation characteristic groups. The clustered window screening module compares the current fluctuation pattern with the historical window based on the cross-segment fluctuation feature group, determines whether there is a concentration trend in multiple consecutive windows, identifies windows that meet the features, and obtains the fluctuation clustering interval. The feeding response discrimination module analyzes the feeding behavior data within the fluctuation clustering interval based on the fluctuation clustering interval, combines dietary structure, body temperature and exercise status, compares the blood glucose change process after eating, identifies key nodes of blood glucose fluctuation and decline, and obtains the dynamic stable interval of behavior. The behavioral deviation analysis module compares the blood glucose performance after newly collected food intake items based on the dynamic stability interval of the behavior, analyzes the differences between the new behavior and the previous pattern, judges the relationship between the behavior and the abnormal response, and obtains the dynamic feature quantity of the deviation. The intervention level determination module determines whether the behavior time interval is within the fluctuation cluster interval based on the offset dynamic feature quantity, screens eating behavior, and compares the impact of different behaviors in combination with diet, exercise and blood glucose decline to obtain the behavior intervention priority level.

2. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The cross-segment fluctuation feature group includes fluctuation morphology, interval persistence structure, and change frequency distribution; the fluctuation clustering interval includes high-density period, risk classification number, and cluster affiliation attribute; the behavioral dynamic stability interval includes response stability index, fluctuation recovery parameter, and duration attribute; the offset dynamic feature quantity includes offset pattern type, response anomaly factor, and behavioral feature label; and the behavioral intervention priority level includes level order parameter, intervention implementation level, and priority grouping.

3. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The fluctuation trend identification module includes: The data stream receiving submodule analyzes the blood glucose monitoring data collected continuously for hours, compares the monitoring data information within each hour, determines the fluctuation boundary of blood glucose level in each time period, summarizes the monitoring time of the maximum and minimum points in each time period, and obtains the blood glucose boundary reference factor. The trend inflection point extraction submodule, based on the blood glucose boundary reference factor, compares the temporal changes of the boundaries of each time period, analyzes the switching of blood glucose trends between each two adjacent time periods, determines the time when the blood glucose rising and falling trends change, and obtains the trend inflection sequence. The cross-segment feature construction submodule analyzes the blood glucose fluctuation performance in each consecutive time period based on the trend turning sequence, determines the pattern type of trend change, classifies the fluctuation structure characteristics of each time period, and obtains the cross-segment fluctuation feature group.

4. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The aggregation window screening module includes: The window structure construction submodule, based on the cross-segment fluctuation feature group, compares the blood glucose fluctuation structure in each time period, judges the continuity of fluctuation performance between consecutive time periods, identifies time segments with consistent fluctuation patterns, optimizes the changes in blood glucose fluctuation curves within the time window, and obtains continuous fluctuation groups. The trend pattern comparison submodule compares the trend characteristics of fluctuation changes in each window based on the continuous fluctuation group, analyzes the dominant direction of blood glucose level changes within the window, judges the correlation of trend paths between windows, identifies the distribution sequence of trend change points, and calculates the continuity index of trend patterns to obtain a trend consistency sequence. The clustering interval identification submodule, based on the trend consistency sequence, determines the segments in the continuous window sequence that exhibit synchronous fluctuation characteristics, identifies time intervals with consistent trends, optimizes the time range marking of the window sequence, and obtains the fluctuation clustering interval.

5. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The feeding response discrimination module includes: The behavioral data extraction submodule compares the dietary structure, eating time and exercise performance recorded in the eating behavior data based on the fluctuation cluster interval, determines the time correspondence between eating events and blood glucose fluctuation cycles, identifies the synchronization of dietary components, body temperature and exercise records in each time period, and obtains a set of eating behavior features. The dynamic curve comparison submodule compares the eating behavior feature set with the eating event and the monitored blood glucose curve, analyzes the dynamic changes in the continuous blood glucose monitoring data after eating, determines the blood glucose level rise and fall segments, filters the change trend of each curve segment, and obtains the blood glucose response curve index. The stable interval identification submodule, based on the blood glucose response curve index, determines the time interval in which blood glucose fluctuations gradually tend to stabilize, filters out segments where the fluctuation range continuously converges, sorts out the food event numbers and the time sequence of each interval, and obtains the dynamic stable interval of behavior.

6. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The behavioral deviation analysis module includes: The new behavior comparison submodule analyzes the blood glucose monitoring sequence corresponding to the newly collected eating behavior based on the dynamic stable interval of the behavior, compares the similarities and differences between the blood glucose change trend under the new behavior and the stable interval, judges the blood glucose response change caused by the new behavior, identifies the monitoring parameters that differ in different time periods, and obtains the new behavior difference feature quantity. The cooperation relationship determination submodule, based on the new behavioral difference feature quantity, judges the medication records and exercise interruption information within the corresponding time period, analyzes the synergy of food, drug and exercise factors at each time node, identifies the interaction performance of behavioral factors under the combination of different time sequences, and obtains a set of behavioral synergy indicators. The abnormal feature identification submodule analyzes the relationship between each behavior and blood glucose response based on the behavioral coordination index set, judges the blood glucose deviation performance under each interaction combination, identifies the intervals and features exhibiting abnormal fluctuations, and obtains the dynamic feature quantity of the deviation.

7. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The intervention level determination module includes: The time interval discrimination submodule determines whether the time period of each behavioral event coincides with the fluctuation aggregation interval based on the offset dynamic feature quantity, filters the feeding behaviors associated within the time interval, and obtains the behavioral event interval index. The factor correlation analysis submodule, based on the behavioral event interval index, compares the dietary categories, exercise performance, and blood glucose decline process involved in eating behavior, analyzes the correlation between changes in dietary structure, exercise intervention, and blood glucose fluctuations, determines the strength of the synergistic effect of each behavioral factor, and obtains the correlation degree of behavioral effects. The priority output submodule determines the driving strength of multiple behavioral interventions on blood glucose dynamics based on the correlation of the behavioral effects, compares the differences in the performance of various intervention measures under different combinations of behavioral factors, identifies the intervention content with the dominant influence, and obtains the priority level of the behavioral intervention.

8. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The direction of blood glucose change refers to the point in continuous data where blood glucose changes from rising to falling or from falling to rising, i.e., the inflection point of blood glucose trend. The fluctuation pattern refers to the trend and structural characteristics of blood glucose level changes over time within the current time window.

9. The AI-driven intelligent blood glucose monitoring system according to claim 1, characterized in that, The central tendency refers to the consistent fluctuations of multiple time windows within a certain period of time. The eating behavior data refers to the eating information entered by the user in the recording platform or device. The blood glucose fluctuation refers to the situation where the blood glucose value rises or falls significantly in a short period of time after eating.

10. An AI-driven intelligent computer analysis method for blood glucose monitoring, the method being used to implement the AI-driven intelligent computer analysis system for blood glucose monitoring as described in any one of claims 1-9, characterized in that, Includes the following steps: S1: Based on blood glucose monitoring data, determine the direction of blood glucose changes every hour, identify the fluctuation characteristics of each time period by comparing the trends of high and low points in each time period, and obtain the cross-segment fluctuation characteristic group. S2: Based on the cross-segment fluctuation feature group, compare the fluctuation patterns of the current window with those of the same period in history, determine whether there is a concentrated trend in multiple consecutive windows, identify windows that meet the features, and obtain the fluctuation clustering interval. S3: Based on the fluctuation clustering interval, analyze the eating behavior data within the interval, combine dietary structure, body temperature and exercise, compare the blood glucose change process after eating, identify the key nodes of blood glucose fluctuation and decline, and obtain the dynamic stable interval of behavior. S4: Based on the dynamic stable interval of the behavior, compare the blood glucose performance after the newly collected food intake item by item, analyze the difference between the new behavior and the previous pattern, determine the relationship between the behavior and the abnormal response, and obtain the offset dynamic feature quantity; S5: Based on the offset dynamic feature, determine whether the behavior time interval is within the fluctuation cluster interval, screen eating behavior, and compare the impact of different behaviors in combination with diet, exercise and blood glucose decline to obtain the behavior intervention priority level.