Intelligent health management system for diabetics

Abstract

The invention relates to the technical field of patient information processing, in particular to an intelligent health management system for diabetic patients, which comprises the steps of acquiring continuous blood glucose monitoring data of patients, extracting blood glucose values, timestamps and patient identity information, establishing a blood glucose data set, performing data cleaning on the blood glucose data set, and removing missing values and repeated values. According to the method, after the data are cleaned and analyzed, the generated blood glucose fluctuation index can reveal the blood glucose fluctuation condition in a short time, and therefore the possible health risk is predicted. The fluctuation indexes are standardized and compared with a set threshold value, potential abnormal blood glucose events are recognized, and the response speed and preventability of disease management are improved. Meanwhile, the health risk of the patient can be quantitatively evaluated through calculation of the abnormal accumulated score, early warning is automatically triggered when the abnormal accumulated score exceeds a safety threshold value, and intervention is carried out in time. And the management efficiency and the treatment pertinence of diabetes mellitus are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of patient information processing, and in particular to an intelligent health management system for diabetic patients. Background Art

[0002] The field of patient information processing technology covers various methods and systems for managing and analyzing patient health information through technical means. This field focuses on how to use electronic health records (EHR), patient management systems, and data analysis tools to optimize patient health outcomes. By integrating and processing patient data collected from different sources, including clinical records, laboratory test results, and real-time monitoring device data. The smart health management system for diabetic patients is a comprehensive health management solution designed specifically for diabetic patients, aiming to improve patients' treatment outcomes and quality of life through highly personalized health monitoring and management tools.

[0003] However, existing technologies are relatively limited in their capabilities for automated processing and anomaly detection, and lack effective mechanisms to instantly warn patients or medical staff of possible health risks. This delayed identification may lead to worsening of the condition or require more expensive medical interventions, affecting the patient's health and economic benefits. For example, failure to adjust treatment plans or modify medication dosages in a timely manner may cause patients to experience severe blood sugar fluctuations, increasing the risk of hospitalization or leading to the development of long-term health problems. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent health management system for diabetic patients.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: an intelligent health management system for diabetic patients comprises:

[0006] Acquire the patient's continuous blood glucose monitoring data, extract the blood glucose value, timestamp and patient identity information, establish a blood glucose data set, perform data cleaning on the blood glucose data set, remove missing values ​​and duplicate values, and obtain cleaned blood glucose data;

[0007] Based on the cleaned blood sugar data, each blood sugar value is compared with the blood sugar values ​​in the previous and next time windows, the difference is calculated to obtain a blood sugar difference value, the blood sugar difference values ​​are arranged in time series, a moving average and a moving standard deviation are calculated to obtain a blood sugar fluctuation index, the blood sugar fluctuation index is standardized to obtain a standardized blood sugar fluctuation index, the standardized blood sugar fluctuation index is compared with a preset threshold value, the blood sugar value exceeding the threshold value is determined, and an abnormal blood sugar index is generated;

[0008] Based on the abnormal blood glucose index, the frequency, duration and amplitude of abnormal blood glucose values ​​are counted, the weighted sum of abnormal blood glucose values ​​is calculated, and an abnormal cumulative score is obtained. The abnormal cumulative score is time-weighted, and the abnormal trend of each patient in the target time period is calculated to obtain a patient risk score. The patient risk score is compared with a risk level threshold, the risk level is determined, and a risk level index is generated;

[0009] Based on the risk level indicator, the risk level indicator is compared with a warning threshold to determine whether a warning needs to be sent.

[0010] Preferably, the steps of obtaining the cleaned blood sugar data are:

[0011] Extract the patient's continuous blood glucose monitoring data, including the blood glucose value, timestamp and patient identity information at each measurement point, and integrate them into a preliminary blood glucose dataset;

[0012] Based on the preliminary blood glucose data set, data validation is performed to identify and mark all missing values ​​in the data, and missing values ​​are eliminated through data cleaning, and duplicate data entries are detected and deleted to generate cleaned blood glucose data.

[0013] Preferably, the step of obtaining the blood sugar difference value is:

[0014] Selecting each measurement point from the cleaned blood sugar data, extracting the blood sugar value and time stamp of the measurement point, and forming a time series blood sugar record;

[0015] Based on the blood glucose records of the time series, a time window is defined for each blood glucose value, and the difference between the time point and the blood glucose values ​​before and after the window is calculated. The calculation formula is:

[0016]

[0017] Where, ΔB i represents the blood glucose difference value at the i-th time point, B i represents the blood glucose value at the i-th time point, k represents the radius of the time window, and j is the index of the time point within the window;

[0018] According to the calculated blood glucose difference value at each time point, a blood glucose difference value list is formed by integration.

[0019] Preferably, the steps for obtaining the blood sugar fluctuation index are:

[0020] Arrange the blood glucose difference values ​​in order according to the timestamps to form a blood glucose difference value list arranged in time order;

[0021] Based on the list of blood glucose difference values ​​arranged in chronological order, a moving average and a moving standard deviation are calculated for each blood glucose difference value to form a blood glucose fluctuation index.

[0022] Preferably, the steps of obtaining the abnormal blood sugar index are:

[0023] Based on the blood sugar fluctuation index, performing Z-Score standardization to generate a standardized blood sugar fluctuation index;

[0024] Based on the standardized blood sugar fluctuation index, a preset threshold value YT is set, and it is calculated whether each standardized index exceeds the preset threshold value, which is expressed as:

[0025]

[0026] Among them, S i is the standardized blood sugar fluctuation index of the i-th data point, TY is the preset threshold, and Y i is the judgment result of exceeding the threshold value;

[0027] An abnormal blood sugar index is generated according to the determination result.

[0028] Preferably, the steps for obtaining the abnormal cumulative score are:

[0029] Based on the abnormal blood sugar index, the frequency, duration and amplitude of each abnormal point are counted to generate abnormal blood sugar value statistical data;

[0030] Based on the abnormal blood sugar value statistical data, the weighted sum of each abnormal blood sugar value is calculated, and the calculation formula is:

[0031]

[0032] Among them, f i represents the frequency of the i-th outlier, d i Represents duration, m i represents amplitude, YE is the weighted sum of abnormal blood glucose values;

[0033] The overall abnormal cumulative score is analyzed based on the calculated weighted sum YE of abnormal blood sugar levels to generate an abnormal cumulative score.

[0034] Preferably, the steps for obtaining the patient risk score are:

[0035] Select the abnormal cumulative score of each patient within the target time period, and generate abnormal cumulative score data within the target time period;

[0036] Based on the abnormal cumulative score data within the target time period, the risk score is calculated using the following formula:

[0037]

[0038] Among them, R i is the risk score of the i-th patient, E it is the abnormal cumulative score of patient i at time point t, w t is the weight of time point t, and T is the total number of time points included in the target time period.

[0039] Preferably, the steps for obtaining the risk level indicator are:

[0040] Collect risk scores of all patients to generate a risk score dataset;

[0041] Based on the risk score data set, comparing the risk score of each patient with a preset risk level threshold, determining the risk level of each patient, and generating a preliminary risk level judgment;

[0042] Based on the preliminary risk level judgment, a risk level label is assigned to each patient to generate a risk level indicator.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are:

[0044] In the present invention, after cleaning and analyzing the data, the generated blood sugar fluctuation index can reveal the blood sugar fluctuation in the short term, thereby predicting possible health risks. These fluctuation indicators are standardized and compared with the set thresholds to identify potential abnormal blood sugar events, thereby improving the response speed and preventiveness of disease management. At the same time, by calculating the abnormal cumulative score, the patient's health risk can be quantitatively assessed, and an early warning can be automatically triggered when the safety threshold is exceeded, so that timely intervention can be carried out. The management efficiency of diabetes and the targeted treatment are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0047] See also Figure 1 The present invention provides a technical solution: an intelligent health management system for diabetic patients includes:

[0048] Obtain the patient's continuous blood glucose monitoring data, extract the blood glucose value, timestamp and patient identity information, establish a blood glucose data set, clean the blood glucose data set, remove missing values ​​and duplicate values, and obtain cleaned blood glucose data;

[0049] Based on the cleaned blood sugar data, each blood sugar value is compared with the blood sugar values ​​in the previous and next time windows, the difference is calculated to obtain the blood sugar difference value, the blood sugar difference values ​​are arranged in time series, the moving average and the moving standard deviation are calculated to obtain the blood sugar fluctuation index, the blood sugar fluctuation index is standardized to obtain the standardized blood sugar fluctuation index, the standardized blood sugar fluctuation index is compared with the preset threshold value, the blood sugar value exceeding the threshold value is determined, and the abnormal blood sugar index is generated;

[0050] Based on the abnormal blood glucose index, the frequency, duration and amplitude of abnormal blood glucose values ​​are counted, the weighted sum of abnormal blood glucose values ​​is calculated, and the abnormal cumulative score is obtained. The abnormal cumulative score is time-weighted, and the abnormal trend of each patient in the target time period is calculated to obtain the patient risk score. The patient risk score is compared with the risk level threshold, the risk level is determined, and the risk level index is generated;

[0051] Based on the risk level indicator, the risk level indicator is compared with the warning threshold to determine whether a warning needs to be sent. Here, the warning threshold is medium risk or high risk. If it is determined to be medium risk or high risk, a warning is sent.

[0052] Specifically, after cleaning and analyzing the data, the generated blood sugar fluctuation index can reveal the short-term blood sugar fluctuations, thereby predicting possible health risks. By standardizing these fluctuation indicators and comparing them with the set thresholds, potential abnormal blood sugar events can be identified, which improves the responsiveness and preventiveness of disease management. At the same time, by calculating the abnormal cumulative score, the patient's health risk can be quantitatively assessed, and an early warning can be automatically triggered when the safety threshold is exceeded, so that timely intervention can be made. This improves the management efficiency of diabetes and the targeted treatment.

[0053] The steps for obtaining cleaned blood sugar data are:

[0054] Extract the patient's continuous blood glucose monitoring data, including the blood glucose value, timestamp and patient identity information at each measurement point, and integrate them into a preliminary blood glucose dataset;

[0055] Based on the preliminary blood glucose dataset, data validation was performed to identify and mark all missing values ​​in the data, missing values ​​were excluded through data cleaning, and duplicate data entries were detected and deleted to generate cleaned blood glucose data.

[0056] Specifically, the patient's continuous blood glucose monitoring data, including blood glucose values, timestamps and patient identity information, are extracted from the database; each data point includes the result of a blood glucose measurement and the corresponding measurement time; these data points are aggregated into a preliminary blood glucose data set. During this process, data verification is used to ensure that the timestamp of each data point is correct and complies with the ISO8601 format. At the same time, the data is checked for duplicate records, and if so, they are removed, thereby forming a complete and accurate preliminary blood glucose data set.

[0057] Data validation was performed on the preliminary blood glucose dataset. First, outliers in the data were marked, i.e., those records that significantly deviated from the normal physiological blood glucose range of the human body (usually 70 to 180 mg / dL), and all missing values ​​were marked, such as unrecorded blood glucose values ​​or missing timestamps; duplicate entries in the data were detected and deleted, especially those data points that were uploaded multiple times due to network failures or operational errors; these operations ensured that the information in the dataset was accurate and without redundancy, and cleaned blood glucose data was generated.

[0058] The steps for obtaining the blood sugar difference value are:

[0059] Select each measurement point from the cleaned blood sugar data, extract the blood sugar value and time stamp of the measurement point, and form a time series blood sugar record;

[0060] Based on the blood glucose records of time series, a time window is defined for each blood glucose value, and the difference between the time point and the blood glucose values ​​before and after the window is calculated. The calculation formula is:

[0061]

[0062] Where, ΔB i represents the blood glucose difference value at the i-th time point, B i represents the blood glucose value at the i-th time point, k represents the radius of the time window, and j is the index of the time point within the window;

[0063] According to the calculated blood glucose difference value at each time point, a blood glucose difference value list is formed by integration.

[0064] Specifically, after extracting the blood glucose value and timestamp of each measurement point, the data quality was checked, including checking the integrity and accuracy of each data point to ensure that there were no missing or erroneous records in the data set. The consistency of the timestamp was also checked to ensure that all records were correctly arranged in chronological order. A time series blood glucose record was formed, which not only reflected the blood glucose level at each time point, but also could accurately track and analyze the fluctuation trend of blood glucose through subsequent steps. The accuracy of these records was the basis for calculating the blood glucose difference value, ensuring that the obtained blood glucose difference value could accurately reflect the patient's blood glucose changes.

[0065] The benefit of the formula is that it emphasizes the impact of adjacent time points on the current blood glucose difference value, thereby increasing the sensitivity to short-term fluctuations in blood glucose; ΔB i represents the blood glucose difference value at the i-th time point, which measures the deviation between the instantaneous blood glucose value and the average of the nearest neighbors. i is the blood glucose value at the ith time point, obtained directly from the time series data, k is the radius of the time window, usually set to 5, indicating that the analysis is about the blood glucose changes within 10 minutes (if measured once a minute), and j is the index of the time point in the window;

[0066] Calculation process:

[0067] Blood glucose value B at the current time point i i =150 mg / dL; the time window radius k = 5; the blood glucose values ​​in the window are [145, 148, 150, 149, 152, 150, 155, 158, 160, 150, 145], a total of 2k+1 = 11 data points.

[0068] Calculate the weighted blood sugar value for each j:

[0069] When j = i-5 (145), weight 1 + |i-5-i| = 6, weighted value

[0070] When j = i-4 (148), weight 1 + |i-4-i| = 5, weighted value

[0071] When j = i-3 (150), weight 1 + |i-3-i| = 4, weighted value

[0072] When j = i-2 (149), weight 1 + |i-2-i| = 3, weighted value

[0073] When j = i-1 (152), weight 1 + |i-1-i| = 2, weighted value

[0074] When j = i (150), weight 1 + |ii| = 1, weighted value

[0075] When j=i+1(155), weight 1+|i+1-i|=2, weighted value

[0076] When j=i+2(158), weight 1+|i+2-i|=3, weighted value

[0077] When j=i+3(160), weight 1+|i+3-i|=4, weighted value

[0078] When j = i + 4 (150), the weight is 1 + |i + 4 - i| = 5, and the weighted value is

[0079] When j = i + 5 (145), the weight is 1 + |i + 5 - i| = 6, and the weighted value is

[0080] Calculate the weighted average:

[0081]

[0082] Calculation of ΔB i :

[0083] ΔB i =|150-59.3|=90.7

[0084] Blood glucose difference value ΔB i The blood sugar level at this time point is 90.7 mg / dL, indicating that the blood sugar value at this time point has a larger fluctuation compared with its adjacent time points. If such fluctuations are common, it may indicate that the patient's blood sugar control is unstable.

[0085] The steps to obtain blood sugar fluctuation indicators are:

[0086] Based on the blood glucose difference values, the values ​​are arranged in order according to the timestamps to form a blood glucose difference value list arranged in time order;

[0087] Based on the list of blood glucose difference values ​​arranged in chronological order, a moving average and a moving standard deviation are calculated for each blood glucose difference value to form a blood glucose fluctuation index.

[0088] Specifically, based on the blood glucose difference value list, it is arranged in timestamp order. First, each data point is time-stamped to ensure that all blood glucose measurement values ​​can reflect their exact measurement time points and identify the specific time of each blood glucose record. Then the data points are sorted. This step compares the timestamps of each data point and arranges them in order from earliest to latest to ensure the order and logic of the data in subsequent analysis. This sequential arrangement enables the subsequent blood glucose fluctuation analysis to accurately reflect the changing trend on the timeline without being disturbed by disordered data, thereby generating a structured and time-sorted blood glucose difference value list.

[0089] Data is extracted from a list of blood glucose difference values ​​arranged in chronological order, and a moving average and moving standard deviation are calculated for each blood glucose difference value. This calculation process uses a fixed window size, usually set to 5 data points before and after, and the average and standard deviation are calculated for the blood glucose difference values ​​within this window. This process smoothes short-term blood glucose fluctuations and quantifies blood glucose stability. The moving average reflects the average level of blood glucose during the time period, while the moving standard deviation measures the breadth of the fluctuation, providing a statistical measure of blood glucose fluctuations. These calculated data are summarized to form a blood glucose fluctuation index.

[0090] The steps to obtain abnormal blood sugar index are:

[0091] Based on the blood sugar fluctuation index, Z-Score standardization is performed to generate a standardized blood sugar fluctuation index;

[0092] Based on the standardized blood sugar fluctuation index, a preset threshold value YT is set, and each standardized index is calculated to see whether it exceeds the preset threshold value, which is expressed as:

[0093]

[0094] Among them, S i is the standardized blood sugar fluctuation index of the ith data point, YT is the preset threshold, and Y i is the judgment result of exceeding the threshold value;

[0095] Based on the judgment result, an abnormal blood sugar index is generated.

[0096] Specifically, Z-Score standardization is performed to adjust the distribution of blood sugar fluctuation indicators. The standardization formula is: Where X is the original data point, μ is the mean, and σ is the standard deviation. In this way, the data is transformed into a distribution centered at 0 and with a standard deviation of 1. The processed data is more suitable for further statistical analysis because the dimension effect in the original data is eliminated.

[0097] In the expression, the step of obtaining the T parameter is to determine the threshold that can distinguish normal from abnormal blood sugar fluctuations by analyzing historical case data, which needs to be set by statistics and experience;

[0098] Calculation process: The blood sugar fluctuation data of a certain day is standardized to S i =1.5, preset threshold YT=1.2, then: Y i =1; this means that the blood sugar fluctuation on that day exceeds the normal range; this result shows that the standardized blood sugar fluctuation value exceeds the set safety threshold on that day, and the patient's blood sugar control on that day is not good;

[0099] According to the judgment result that exceeds the threshold, all data points with the result of 1 are marked as abnormal. This step involves calculating each Y i The value is logically judged. If the value is 1, the corresponding time point is recorded as an exception.

[0100] The steps to obtain the abnormal cumulative score are:

[0101] Based on the abnormal blood sugar index, the frequency, duration and amplitude of each abnormal point are counted to generate abnormal blood sugar value statistics;

[0102] Based on the abnormal blood glucose value statistics, the weighted sum of each abnormal blood glucose value is calculated using the following formula:

[0103]

[0104] Among them, f i represents the frequency of the i-th outlier, d i Represents duration, m i represents amplitude, YE is the weighted sum of abnormal blood glucose values;

[0105] The overall abnormal cumulative score is analyzed based on the calculated weighted sum YE of abnormal blood sugar levels to generate an abnormal cumulative score.

[0106] Specifically, based on the abnormal blood glucose index, frequency, duration and amplitude data statistics are performed on each abnormal point. During the statistical process, the blood glucose monitoring data of the relevant patients are first extracted from the medical database. These data include but are not limited to timestamps, blood glucose values ​​and patient identification information. Data points marked as abnormal are screened out, and then frequency statistics are performed on each marked data point, that is, the frequency of occurrence within a certain period of time is calculated. The duration is calculated by comparing the timestamps of the start and end of the abnormality. The amplitude analysis requires calculating the difference between the blood glucose value at the abnormal point and the average value of normal blood glucose.

[0107] The benefit of the formula is that it increases the impact of duration on the cumulative score of abnormal blood glucose through a weighted mechanism, so that abnormal points with longer duration contribute more to the final result, thereby improving the flexibility and accuracy of dealing with sudden and persistent abnormal blood glucose situations; i The parameter acquisition step is to determine the frequency of each abnormal point within the set time window by counting the records in the medical database, d i The parameter acquisition step is to calculate m by performing difference calculation on the records of the start and end time of the abnormal state. i The step of obtaining the parameter is to calculate the difference between the blood glucose value at the abnormal point and the set normal blood glucose threshold value;

[0108] Calculation process: Assume that the frequency of abnormal point 1 is 2 times / hour, the duration is 30 minutes, and the amplitude is 50 mg / dl. Calculate according to the formula and substitute the specific values ​​to get:

[0109]

[0110] A lower score (e.g., less than 10): indicates that the abnormal event may be brief, less frequent, or less severe, and has little impact on the patient's overall blood sugar control. This situation may not require an immediate adjustment to the treatment plan, but continuous monitoring should be performed to confirm whether the trend changes.

[0111] Moderate scores (e.g., 10 to 30): indicate that abnormal events have a certain duration, frequency, or magnitude, and have a moderate impact on the patient's blood sugar control. This may indicate the need for moderate adjustments to the patient's lifestyle or treatment regimen, such as changing diet or adjusting medication schedules.

[0112] Higher scores (e.g., higher than 30): indicate that abnormal events may be very frequent, long-lasting, or large in magnitude, affecting the patient's blood sugar control;

[0113] The results show that during the monitoring period, this abnormal point had a significant impact on overall blood sugar control.

[0114] The steps to obtain the patient risk score are:

[0115] Select the abnormal cumulative score of each patient within the target time period, and generate abnormal cumulative score data within the target time period;

[0116] Based on the abnormal cumulative score data within the target time period, the risk score is calculated using the following formula:

[0117]

[0118] Among them, R i is the risk score of the i-th patient, E it is the abnormal cumulative score of patient i at time point t, w t is the weight of time point t, and T is the total number of time points included in the target time period.

[0119] Specifically, the abnormal cumulative score of each patient in the target time period is selected, and for each patient, all abnormal event records related to the patient are retrieved from the database, which include the timestamp, duration, frequency and magnitude of the event;

[0120] The benefit of the formula is that it not only enhances the sensitivity to outliers through the square root of the sum of squares, but also reduces the impact of a single extreme value on the overall score through smoothing, making the risk score more balanced and reflecting the actual situation; R i is the risk score of the ith patient, representing the patient’s overall health risk during the observation period, E it is the abnormal cumulative score of patient i at time point t, which is calculated by monitoring the deviation of blood glucose measurement values ​​at each time point relative to the normal value, w t is the weight of time point t, and its value is set in descending order according to the distance from the current time point to ensure that recent data has a greater impact on the score than long-term data. T is the total number of time points examined;

[0121] Calculation process: A patient has 7 data points in a week, and the abnormal cumulative score E of each point it They are {10, 20, 30, 40, 50, 60, 70} respectively, and the time weight w t Set it as {7,6,5,4,3,2,1} and calculate the risk score according to the formula:

[0122]

[0123] Low value range (such as R i <50): Indicates that the patient's blood sugar fluctuations are small, blood sugar is well controlled, and is in the low-risk area.

[0124] Median interval (e.g. 50≤R i <75): This indicates that the patient has moderate blood sugar fluctuations and may need to adjust diet, exercise, or medication to better control blood sugar.

[0125] High value interval (such as R i ≥75): This indicates that the patient has significant problems with blood sugar control, large blood sugar fluctuations, and is in a high-risk area.

[0126] The results showed that the patient's risk score was 86.77, which means that the patient's health status during the observation period may be at medium to high risk.

[0127] The steps to obtain the risk level indicator are:

[0128] Collect risk scores of all patients to generate a risk score dataset;

[0129] Based on the risk score data set, the risk score of each patient is compared with the preset risk level threshold, the risk level of each patient is determined, and a preliminary risk level judgment is generated;

[0130] Based on the preliminary risk level judgment, a risk level label is assigned to each patient and a risk level indicator is generated.

[0131] Specifically, the setting of risk level thresholds is determined based on historical data analysis and expert clinical experience. First, the frequency and severity of risk events are extracted from historical patient data, and the threshold dividing points of different risk levels are determined through statistics. These thresholds are intended to divide patients into three levels: low risk, medium risk and high risk. The determination of each threshold needs to be verified and adjusted to match the health risk characteristics of different groups.

[0132] Based on the preliminary risk level judgment, the next step is to assign a specific risk level label to each patient. This step involves parsing the relationship between each patient's risk score and the preset threshold, and by comparing whether the score exceeds each threshold, it is clear which risk level the patient belongs to.

Claims

1. An intelligent health management system for diabetic patients, characterized in that: The system comprises: Acquire the patient's continuous blood glucose monitoring data, extract the blood glucose value, timestamp and patient identity information, establish a blood glucose data set, perform data cleaning on the blood glucose data set, remove missing values ​​and duplicate values, and obtain cleaned blood glucose data; Based on the cleaned blood sugar data, each blood sugar value is compared with the blood sugar values ​​in the previous and next time windows, the difference is calculated to obtain a blood sugar difference value, the blood sugar difference values ​​are arranged in time series, a moving average and a moving standard deviation are calculated to obtain a blood sugar fluctuation index, the blood sugar fluctuation index is standardized to obtain a standardized blood sugar fluctuation index, the standardized blood sugar fluctuation index is compared with a preset threshold value, the blood sugar value exceeding the threshold value is determined, and an abnormal blood sugar index is generated; Based on the abnormal blood glucose index, the frequency, duration and amplitude of abnormal blood glucose values ​​are counted, the weighted sum of abnormal blood glucose values ​​is calculated, and an abnormal cumulative score is obtained. The abnormal cumulative score is time-weighted, and the abnormal trend of each patient in the target time period is calculated to obtain a patient risk score. The patient risk score is compared with a risk level threshold, the risk level is determined, and a risk level index is generated; Based on the risk level indicator, the risk level indicator is compared with a warning threshold to determine whether a warning needs to be sent.

2. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the cleaned blood sugar data are as follows: Extract the patient's continuous blood glucose monitoring data, including the blood glucose value, timestamp and patient identity information at each measurement point, and integrate them into a preliminary blood glucose dataset; Based on the preliminary blood glucose data set, data validation is performed to identify and mark all missing values ​​in the data, and missing values ​​are eliminated through data cleaning, and duplicate data entries are detected and deleted to generate cleaned blood glucose data.

3. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the blood sugar difference value are: Selecting each measurement point from the cleaned blood sugar data, extracting the blood sugar value and time stamp of the measurement point, and forming a time series blood sugar record; Based on the blood glucose records of the time series, a time window is defined for each blood glucose value, and the difference between the time point and the blood glucose values ​​before and after the window is calculated. The calculation formula is: Where, ΔB i represents the blood glucose difference value at the i-th time point, B i represents the blood glucose value at the i-th time point, k represents the radius of the time window, and j is the index of the time point within the window; According to the calculated blood glucose difference value at each time point, a blood glucose difference value list is formed by integration.

4. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the blood sugar fluctuation index are as follows: Arrange the blood glucose difference values ​​in order according to the timestamps to form a blood glucose difference value list arranged in time order; Based on the list of blood glucose difference values ​​arranged in chronological order, a moving average and a moving standard deviation are calculated for each blood glucose difference value to form a blood glucose fluctuation index.

5. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the abnormal blood sugar index are: Based on the blood sugar fluctuation index, performing Z-Score standardization to generate a standardized blood sugar fluctuation index; Based on the standardized blood sugar fluctuation index, a preset threshold value YT is set, and it is calculated whether each standardized index exceeds the preset threshold value, which is expressed as: Among them, S i is the standardized blood sugar fluctuation index of the ith data point, YT is the preset threshold, and Y i is the judgment result of exceeding the threshold value; An abnormal blood sugar index is generated according to the determination result.

6. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the abnormal cumulative score are: Based on the abnormal blood sugar index, the frequency, duration and amplitude of each abnormal point are counted to generate abnormal blood sugar value statistical data; Based on the abnormal blood sugar value statistical data, the weighted sum of each abnormal blood sugar value is calculated, and the calculation formula is: Among them, f i represents the frequency of the i-th outlier, d i Represents duration, m i represents amplitude, YE is the weighted sum of abnormal blood glucose values; The overall abnormal cumulative score is analyzed based on the calculated weighted sum YE of abnormal blood sugar levels to generate an abnormal cumulative score.

7. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the patient risk score are: Select the abnormal cumulative score of each patient within the target time period, and generate abnormal cumulative score data within the target time period; Based on the abnormal cumulative score data within the target time period, the risk score is calculated using the following formula: Among them, R i is the risk score of the i-th patient, E it is the abnormal cumulative score of patient i at time point t, w t is the weight of time point t, and T is the total number of time points included in the target time period.

8. The intelligent health management system for diabetic patients according to claim 1, characterized in that: The steps for obtaining the risk level indicator are: Collect risk scores of all patients to generate a risk score dataset; Based on the risk score data set, comparing the risk score of each patient with a preset risk level threshold, determining the risk level of each patient, and generating a preliminary risk level judgment; Based on the preliminary risk level judgment, a risk level label is assigned to each patient to generate a risk level indicator.

Images