A campus health management system and method based on population benchmark analysis

Abstract

The application discloses a kind of campus health management system and method based on group benchmark analysis, comprising: data acquisition module, for collecting the health and behavior data of student individual, and implementing data anonymization processing;Data processing and analysis module, including group clustering unit, for clustering the anonymized student individual data into homogenization analysis group according to student group attribute;Benchmark generation unit, for calculating the dynamic statistical benchmark band of one or more health indicators in each analysis group based on historical data within the set time window;Individual comparative analysis unit, for comparing each student individual data with the dynamic statistical benchmark band of corresponding health indicators of its analysis group, and calculating the relative deviation of individual health indicators;Early warning and report generation module, for triggering hierarchical early warning according to individual comparative analysis result, and generating visual health report. It can be widely applied to the field of campus health management technology.

Description

Technical Field

[0001] This invention relates to the field of campus health management technology, specifically to a campus health management system and method based on population benchmark analysis. Background Technology

[0002] In recent years, with the continuous advancement of technology and the deepening of the concept of smart education, campus health management systems have gradually become an important tool for improving students' physical and mental health. Among existing technologies, some systems collect data such as student activity and heart rate using wearable devices and compare this data with preset, fixed health standards based on age, gender, etc., to achieve health early warning. Another approach collects multi-source behavioral data of students on campus, uses statistical models to calculate behavioral indices, and compares them with preset thresholds to detect anomalies. In addition, some systems record and provide tiered early warnings and responses to abnormal student behavior through manual reporting.

[0003] However, most of the existing technologies mentioned above rely on fixed absolute thresholds as the judgment standard, ignoring the actual differences in health habits among different student groups (such as different regions, classes, and grades), as well as the fluctuations in the health status of the same group at different times (such as exam week and normal times). This leads to warnings either lagging behind the actual situation or generating a large number of false alarms. Secondly, existing systems mainly focus on the deviation of individuals from abstract standards, failing to reflect the relative health status of individuals within their real communities. This results in inaccurate warning positioning and difficulty in accurately identifying those who truly need attention.

[0004] Therefore, there is an urgent need for a health management solution that can focus on group differences in order to improve the accuracy and effectiveness of early warning. Summary of the Invention

[0005] In order to solve the problems existing in the prior art, the purpose of this invention is to provide a campus health management system and method based on population benchmark analysis, which solves the problems of delayed early warning and inaccurate positioning in the existing campus health management system.

[0006] To achieve the above objectives, in a first aspect, embodiments of the present invention provide a campus health management system based on population benchmark analysis, comprising:

[0007] The data acquisition module is used to collect individual students' health and behavioral data and to perform data anonymization processing.

[0008] The data processing and analysis module includes:

[0009] The group clustering unit is used to cluster anonymized student individual data into homogeneous analysis groups based on student group attributes; the benchmark generation unit is used to calculate the dynamic statistical benchmark band of one or more health indicators within each analysis group based on historical data within a set time window; the individual comparison analysis unit is used to compare each student individual data with the dynamic statistical benchmark band of the corresponding health indicator of its analysis group and calculate the relative deviation of the individual health indicator.

[0010] The early warning and report generation module is used to trigger tiered early warnings based on individual comparative analysis results and generate visualized health reports.

[0011] Preferably, the benchmark generation unit uses a weighted time window method to calculate a dynamic statistical benchmark band for health indicators. The length of the time window is adaptively adjusted according to the intensity of the current calendar event. The calculation of the dynamic statistical benchmark band is based on the weighted 25th percentile (…). ) and the weighted 75th percentile ( The middle 50% of individuals are defined.

[0012] Preferably, the reference generation unit includes:

[0013] The time decay weighting subunit is used to assign weights to the data within the time window using an exponential decay function.

[0014] The weighted quantile baseline band sub-unit is used to sort the data by time and calculate the cumulative weight percentage, and then calculate the weighted quantile through linear interpolation.

[0015] Preferably, the time decay weight subunit adjusts the weight of each time point within the time window, wherein the determination of the weight depends on the distance from the current time point and the influence intensity of related calendar events.

[0016] Preferably, the individual comparison analysis unit adopts a weighted scoring model, which integrates the relative deviation of multiple health indicators to calculate a multidimensional health risk assessment score and trend risk.

[0017] Preferably, the individual comparative analysis unit converts the relative deviation of individual health indicators into risk sub-scores in the range of 0 to 1 using the Sigmoid function, and then calculates the individual's multidimensional health risk assessment score using a weighted scoring model. The selection of the Sigmoid function parameters is based on the dispersion of health indicators and the distribution characteristics of actual deviation.

[0018] Preferably, the preset group attributes on which the group clustering unit is based include at least the student's grade and class.

[0019] Preferably, the reports generated by the early warning and report generation module include: an individual report showing the comparison between individual data trends and the group dynamic baseline, and a group report showing the distribution of overall health indicators and baseline trends of the group.

[0020] Preferably, the system further includes an intervention suggestion knowledge base, and the early warning and report generation module matches and outputs personalized health intervention suggestions from the intervention suggestion knowledge base based on the specific deviation patterns identified by the individual comparison analysis unit.

[0021] Secondly, embodiments of the present invention also provide a campus health management method based on population benchmark analysis, comprising:

[0022] Collect individual student health and behavioral data and anonymize the data.

[0023] Based on the student group attributes, the anonymized individual student data were clustered into homogeneous analysis groups;

[0024] Within each of the analysis groups, a dynamic statistical baseline band for one or more health indicators is calculated based on historical data within a set time window.

[0025] Each student's individual data is compared with the dynamic statistical baseline of the corresponding health indicator of the analysis group to which it belongs, and the relative deviation of the individual health indicator is calculated.

[0026] Based on the results of individual comparative analysis, a tiered early warning system is triggered, and a visual health report is generated.

[0027] The present invention has the following beneficial effects:

[0028] The technical solution provided by this invention divides students into homogeneous analysis groups based on their group attributes. By generating dynamic statistical benchmark bands closely related to the current status of specific student groups, it replaces the traditional absolute standard model, making health assessments more aligned with the actual health characteristics of different groups. The individual comparison analysis unit accurately quantifies the degree of individual health deviation, truly reflecting an individual's relative health status within the group and achieving precise positioning of health risks. The early warning and report generation module triggers tiered early warnings and generates visual reports based on the analysis results, effectively improving the accuracy and timeliness of early warnings. Simultaneously, it presents individual and group health statuses to school administrators, assisting them in making scientific campus health management decisions and comprehensively enhancing the professionalism and effectiveness of campus health management. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the campus health management system based on population benchmark analysis disclosed in an embodiment of the present invention;

[0030] Figure 2This is a flowchart illustrating the campus health management method based on population benchmark analysis disclosed in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0032] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] like Figure 1 As shown, in a first aspect, embodiments of the present invention provide a campus health management system based on population benchmark analysis, comprising:

[0034] The data acquisition module is used to collect individual students' health and behavioral data and to perform data anonymization processing.

[0035] The data processing and analysis module includes:

[0036] The group clustering unit is used to cluster anonymized student individual data into homogeneous analysis groups based on student group attributes; the benchmark generation unit is used to calculate the dynamic statistical benchmark band of one or more health indicators within each analysis group based on historical data within a set time window; the individual comparison analysis unit is used to compare each student individual data with the dynamic statistical benchmark band of the corresponding health indicator of its analysis group and calculate the relative deviation of the individual health indicator.

[0037] The early warning and report generation module is used to trigger tiered early warnings based on individual comparative analysis results and generate visualized health reports.

[0038] This invention addresses the issues of delayed early warnings and inaccurate positioning in campus health management systems by employing a data acquisition module to obtain individual students' health and behavioral data. Simultaneously, it implements data anonymization, ensuring privacy and security during transmission and storage while laying the foundation for subsequent group analysis. The group clustering unit in the data processing and analysis module performs rule-based clustering based on student group attributes. The benchmark generation unit calculates the benchmark band for group health indicators, while the individual comparison analysis unit compares individual data with the dynamic benchmark of its respective analysis group, calculating the relative deviation of individual health indicators, quantifying the degree of individual health deviation, and assessing health risks. This combination of technical features enables the system to dynamically adjust the benchmark, more accurately reflecting the current normal range of a particular cluster, thereby improving the accuracy and timeliness of early warnings. The early warning and report generation module, based on the analysis results generated by the individual comparison analysis unit, executes a tiered early warning mechanism to generate visualized health reports reflecting the health status of individuals and groups. This effectively solves the limitations of traditional fixed-threshold early warning methods, achieving precise positioning of individual health risks while also providing a holistic view of the group's health situation. This helps school administrators make more scientific decisions and optimize resource allocation, significantly improving the level of campus health management.

[0039] In embodiments of this application, the data acquisition module relies on wearable devices used by students, such as smart bracelets that support PPG (photoplethysmography) and accelerometers. The acquired data may include:

[0040] Exercise data: raw signals from the triaxial accelerometer (used to calculate cadence and intensity), daily steps (pedometer output), and duration of moderate to high intensity activity (determined by using acceleration signals and the metabolic equivalents (METs) model).

[0041] Sleep data: Based on accelerometer and heart rate variability (HRV) sleep stage algorithm, outputs total sleep duration, deep sleep / light sleep / REM sleep duration, sleep onset time, wake-up time, and sleep efficiency (total sleep duration / bedtime).

[0042] Stress and cardiovascular data: resting heart rate (HR), heart rate variability (HRV, time-domain index SDNN or frequency-domain index LF / HF), and skin conductance (EDA).

[0043] It is also possible to periodically retrieve academic data, such as daily attendance records and management tags, through the secure API interface of the campus information system. Management tags can include the student's grade, class, and dormitory.

[0044] In embodiments of the present invention, the specific process of anonymized data collection may include the following steps:

[0045] Device registration and anonymous ID generation. When students receive their wristbands, they scan the device's QR code using the school's dedicated app. The app sends the student's student ID (SID) and the device's physical address (MAC address) to the registration server. The server generates a high-strength random string as an anonymous unique identifier (AID), establishes a mapping relationship from SID to AID to MAC, and sends the AID-MAC pair to both the wristband and the app. Thereafter, only the AID is used in all communications.

[0046] Data transmission. The wristband transmits encrypted packaged data ({AID, timestamp, sensor type, data value}) to an edge server or cloud access point via a Bluetooth gateway or campus Wi-Fi. The data transmission frequency is configurable; for example, exercise data can be batched every minute, and sleep data can be synchronized once every morning.

[0047] Identifier Reset. To prevent long-term tracking risks, the system can trigger an anonymous ID reset process once per semester (or at a fixed interval). The server generates a new AID' for each AID, updates the mapping relationship SID->AID', and sends the new AID'-MAC pair to the wristband and the app. Old AIDs in historical data will be correctly associated with the same SID through the mapping table during analysis, but to external attackers, the data flow is broken.

[0048] Furthermore, before proceeding with the analysis, the data can be preprocessed, including:

[0049] Time alignment unifies all data streams to the Beijing timestamp and corrects for minor clock discrepancies that may exist between devices;

[0050] Standardize the format, parse and convert heterogeneous data formats (such as JSON, Protobuf) from different equipment manufacturers into a unified data model within the system;

[0051] Initial validity check: Check data packet integrity and discard data with invalid AID or seriously incorrect timestamp.

[0052] The preprocessed data is published to a unified data bus in streaming or micro-batch format for use by downstream data processing and analysis modules.

[0053] In embodiments of the present invention, the group clustering unit performs deterministic allocation of all students based on group attributes, forming a health behavior analysis group with inherent consistency.

[0054] Furthermore, to ensure the quality of the analyzed data, the data is cleaned before clustering, including:

[0055] 1) Outlier Detection and Handling:

[0056] Static range filtering: For physiological parameters such as heart rate, records that exceed human physiological limits are directly removed;

[0057] Dynamic Z-score filtering: For example, daily steps, calculate the mean (μ) and standard deviation (σ) of its data over the past N days. For the current value x, if |(x - μ)| > 3σ, it is marked as an anomaly. If an anomaly is found, it is replaced with μ or a random value within the interval [μ - σ, μ + σ] to maintain data continuity;

[0058] Multi-dimensional anomaly detection: For example, the isolated forest algorithm can be used to quickly identify sample points that are significantly alienated from other points in the feature space by simultaneously examining three dimensions: sleep onset time, sleep duration, and resting heart rate.

[0059] 2) Missing value imputation:

[0060] Random missing data: For short-term data missing due to occasional device disconnection, time series linear interpolation is used to fill in the missing data.

[0061] Systematic missing data: For missing data such as sleep data for an entire night, if it is determined that the device was not worn, no imputation will be made, and the day will be marked as a missing data day. When calculating the daily average in subsequent calculations, the denominator will be reduced accordingly.

[0062] Data smoothing: For high-frequency noise data such as steps per minute, a moving average window is used for smoothing to better reflect activity trends rather than instantaneous fluctuations.

[0063] The cleaned data is organized by AID and time, stored in the cleaned details table, and marked with a cleaned operation log.

[0064] The group clustering unit, through rule-based clustering, directly uses structured attributes highly correlated with health behaviors in campus management as clustering keys. Individuals are divided into homogeneous groups with the same attributes, which serve as the baseline for defining the range.

[0065] Preferably, the preset group attributes on which the group clustering unit is based include at least the student's grade and class.

[0066] In embodiments of the present invention, the primary clustering keys are grade level and class. Because classmates share the same timetable, schedule, teacher management, and class culture, their health patterns are the most comparable.

[0067] Optional secondary clustering keys may include gender and dormitory. In some specific embodiments, this can be used for specific analysis scenarios, such as analyzing differences in physical activity between males and females, or the impact of dormitory environment on sleep.

[0068] For example, the group clustering unit uses a mapping dictionary of AID -> (grade, class, gender, dormitory). Every morning, the group clustering unit reads the list of AIDs with active data in the past 24 hours from the cleaned details table, and assigns a group label Group_ID to each AID according to the mapping dictionary. For example, G2025_3_2 represents Class 2 of Grade 3 in 2025.

[0069] Furthermore, it can also include dynamic group member management. When a student changes classes or moves up a grade, the school information system updates, and this unit dynamically updates the AID's group affiliation by monitoring these change events.

[0070] Preferably, the benchmark generation unit uses a weighted time window method to calculate a dynamic statistical benchmark band for health indicators. The length of the time window is adaptively adjusted according to the intensity of the current calendar event. The calculation of the dynamic statistical benchmark band is based on the weighted 25th percentile (…). ) and the weighted 75th percentile ( The middle 50% of individuals are defined.

[0071] In an embodiment of the present invention, the benchmark generation unit uses a weighted time window method to adaptively adjust the window length to reflect the intensity of current calendar events, and calculates a benchmark band for the health indicators of a specific group. This benchmark band is […]. The system aims to capture the health behavior range of the middle 50% of individuals, providing a more flexible and realistic assessment standard. It can dynamically generate and update health benchmarks based on the characteristics of different groups, thereby accurately assessing the degree to which individuals deviate from group health behaviors. This enhances the timeliness and accuracy of the early warning mechanism and supports campus administrators in making scientific decisions based on the group's health status, effectively supporting campus health management.

[0072] Furthermore, the reference generation unit includes:

[0073] The time decay weighting subunit is used to assign weights to the data within the time window using an exponential decay function.

[0074] The weighted quantile baseline band sub-unit is used to sort the data by time and calculate the cumulative weight percentage, and then calculate the weighted quantile through linear interpolation.

[0075] In an embodiment of the present invention, the system predefines a set of health indicators to be monitored. ,For example, Average daily effective steps Nighttime sleep time, The percentage of deep sleepers Average resting heart rate during the day Stress index.

[0076] The system integrates the academic calendar. When it recognizes that the current date falls within preset scenarios such as final exam week, sports meet preparation period, or the first week after a holiday, it automatically adjusts the window length and weight.

[0077] The formula for calculating the time window can be: in, Based on the time window; To enable adaptive adjustment, the larger this value is set, the more sensitive it is to calendar events; It is a Sigmoid type adjustment function. Historical average volatility This is the intensity factor for calendar events.

[0078] During periods of high event intensity, such as the stressful period of exam preparation, It can be set to 0.9 to achieve this during exam week. Shorter, benchmarks reflect current tensions more quickly; normal days It can be set to 0.5 to use a longer window, resulting in a more stable benchmark.

[0079] Preferably, the time decay weight subunit adjusts the weight of each time point within the time window, wherein the determination of the weight depends on the distance from the current time point.

[0080] Within the window, data closer to the current date has a higher weight. Exponentially decaying weights can be used. ,in, As the attenuation factor, , For the current time, Let be the time of the i-th historical data point.

[0081] λ is preset by the system based on the type of health indicator. Taking λ=0.1 as an example, today's data weight... 0=1, weight of data from 1 day prior =0.9, weight of data from 3 days ago 3=0.7. By assigning dynamic weights to the data within the time window, the baseline prioritizes reflecting changes in data such as the number of steps taken within the past 3 days.

[0082] In this embodiment, the weighted dynamic time window calculation method enables accurate diagnosis of individual health data and dynamic capture of group health status. This method adjusts the weight of each time point within the past N days, reflecting not only the freshness of the data but also the impact of specific calendar events such as exam weeks or sports meet preparation periods on student health behaviors. This strategy ensures that the system can quickly adapt to and generate more realistic health benchmarks based on the specific circumstances of the student group. The weights are determined by their distance from the current time point; more recent time points have higher weights, making the benchmark more sensitive to recent changes. By dynamically adjusting the length and weight of the time window, the system can more accurately assess the degree to which an individual deviates from the dynamic benchmark of their group, thereby improving the timeliness and accuracy of early warnings and effectively promoting the scientific and refined decision-making in campus health management.

[0083] In other embodiments not shown, the time window adjustment strategy can be further optimized, for example by introducing machine learning models to predict future health trends, or by adjusting weights based on real-time feedback, to achieve more intelligent and adaptive health management.

[0084] For a specific group Group_ID, a specific metric In the adjusted time window The set of indicator values ​​X = {x1, x2, ..., xN} for all members within the group, and their corresponding date weights. ={ 1, 2,..., N}.

[0085] The calculation process of the dynamic statistical baseline band requires sorting X by value and accumulating the corresponding weights until the total weights reach 25% and 75%. Then, precise values ​​are obtained through interpolation, ultimately yielding the dynamic statistical baseline band. ].

[0086] For example, on a typical Tuesday in the middle of the semester, the system identifies the intensity of the calendar event corresponding to the current date. The data was relatively low, and data was collected over a 7-day time window. For the specific group of Class 2, Grade 11, the baseline generating unit calculated the average daily steps of all students over the past 7 days, applying exponentially decaying weights. λ is set to 0.8. After sorting and weighting, the weighted 25th percentile is obtained. For 5500 steps, the weighted 75th percentile The daily average step count for Class 2, Grade 11 is 7500 steps. Therefore, for this day, the baseline daily step count for Class 2, Grade 11 is […]. The range is defined as [5500, 7500] steps, which is used to assess the individual's current normal range.

[0087] Based on this, the baseline for each indicator, each group, and each day is calculated and stored in the group baseline table. The system can calculate the previous day's baseline at midnight each day and use it for individual comparisons on that day.

[0088] Preferably, the individual comparison analysis unit adopts a weighted scoring model, which integrates the relative deviation of multiple health indicators to calculate a multidimensional health risk assessment score and trend risk.

[0089] Furthermore, the individual comparative analysis unit uses the Sigmoid function to convert the relative deviation of individual health indicators into risk sub-scores in the range of 0 to 1, and then calculates the individual's multidimensional health risk assessment score through a weighted scoring model. The selection of the Sigmoid function parameters is based on the dispersion of health indicators and the distribution characteristics of actual deviation.

[0090] The individual comparative analysis unit is responsible for comparing the data of an individual student with the dynamic benchmark of their group, quantifying the degree of deviation, and ultimately comprehensively assessing the risk, specifically including:

[0091] Single indicator deviation calculation:

[0092] According to students On date t Health indicator values and the baseline band of the indicator for its group on date t. ].like Falling on [ Within ], then the deviation degree ;like This is considered a negative deviation, such as taking too few steps or going to bed too late. Calculate the relative negative deviation: The relative deviation result is a negative value; the larger the absolute value, the greater the deviation from the lower limit of the group baseline.

[0093] like This is considered a positive deviation, such as an abnormally high number of steps or an abnormally high heart rate. Calculate the relative positive deviation: The relative deviation result is a positive value; the larger the value, the more it exceeds the upper limit of the group baseline.

[0094] relative deviation It is a standardized, dimensionless ratio used to quantify the relative severity of an individual's deviation from the range of their current normal peers, allowing for comparison and aggregation of deviations between different indicators.

[0095] Multidimensional health risk assessment:

[0096] The system for each indicator Assign a weight to an indicator , The weights of this indicator can be jointly set by the school doctor and physical education teacher based on experience (for example, exercise weight 0.3, sleep weight 0.3, stress weight 0.4).

[0097] indicators Daily deviation The health indicator is compressed and smoothed using the Sigmoid function, mapped to a risk sub-score range of 0-1, and the impact of significant deviations is amplified. The risk score is: γ is a sensitivity parameter used to adjust the steepness of the Sigmoid function. The larger γ is, the faster the function rises when the deviation increases, and the stronger the amplification effect on significant deviations.

[0098] In embodiments of the present invention, the Sigmoid function is used in the conversion of single-indicator deviation to risk scores. For health indicators such as step count, system analysis reveals high dispersion; most students' step counts fall near the baseline, while a few extremely active or sedentary students exhibit significant deviations. Based on this characteristic, the system can select a higher Sigmoid function sensitivity parameter γ to ensure that the risk scores of students with abnormally low or high step counts quickly reflect the severity of the deviation. For students with step counts close to the baseline, even slight deviations result in very slow increases in risk scores, avoiding overreaction to normal fluctuations. Similarly, for relatively concentrated and less dispersed health indicators such as sleep duration, the system can select a lower γ value, ensuring that even small changes in sleep duration can be reasonably converted into appropriate risk scores, guaranteeing the comprehensiveness and accuracy of the assessment.

[0099] Calculate the overall risk score for the day: ,here Used to retain deviation from the direction, if ,but ;like ,but A negative value indicates that the overall health is poor.

[0100] In summary, the individual comparative analysis unit outputs a risk vector for each active student daily, containing the following: {AID, date, [deviation of each indicator], overall risk score, and trend risk assessment score}.

[0101] Furthermore, considering the continuity of health status, trend risk calculations can be performed, such as calculating the moving average and trend slope of the composite risk score over the past M days. The moving average is the average of the composite risk scores over the past M days. The trend slope is obtained by analyzing the past M days. Linear regression is performed on the sequence to obtain the slope. A slope > 0 indicates that the risk is worsening, while a slope < 0 indicates that the risk is improving.

[0102] The trend risk assessment score, which combines moving average and trend slope, is as follows: Where α, β, and θ are harmonic coefficients used to balance immediate risk, recent average risk, and deterioration trend.

[0103] In this invention, the individual comparative analysis unit not only assesses the deviation of health indicators on a single day, but also introduces the moving average and trend slope analysis of the comprehensive risk score over the past M days to comprehensively capture the long-term trend changes in an individual's health status. This design allows the system to identify short-term health fluctuations while focusing more on the continuity and stability of health status, effectively avoiding false alarms that may result from relying on only one day's data. By calculating the time series of individual health risks, the system can identify individuals whose health status is slowly deteriorating, even if they have not yet reached the daily warning threshold, as well as special cases with abnormal indicators in the short term but stable overall trends. This provides campus health managers with more comprehensive and detailed early warning information. Timely detection of potential health risks facilitates early intervention and prevention, while also reducing unnecessary alarms and improving the accuracy and practicality of early warnings.

[0104] The early warning and report generation module is used to trigger tiered early warnings based on the results of individual comparative analysis.

[0105] In an embodiment of the present invention, a three-level early warning threshold can be set based on the risk assessment score, including:

[0106] Low-risk blue alert: Individuals whose risk assessment scores are in the bottom 10%-20% of the group will be automatically added to the watchlist. No alerts will be sent proactively, but the information will be reflected in the report.

[0107] Yellow alert for medium risk: If an individual's risk assessment score is in the bottom 5%-10% of the group, or if they are in the blue "watch" range for N consecutive days, an alert will be triggered, and the homeroom teacher and school doctor will be notified to pay attention.

[0108] High-risk red alert: If an individual's risk assessment score is in the bottom 5% of the group, or if a single indicator shows an extreme negative deviation, an alert will be triggered immediately and sent to the homeroom teacher, school doctor, and vice principal in charge.

[0109] It should be noted that the warning status is not permanent. When a student's individual risk assessment score returns to the normal range for P consecutive days, the system will automatically downgrade or close the warning and record the de-escalation log.

[0110] Preferably, the reports generated by the early warning and report generation module include: an individual report showing the comparison between individual data trends and the group dynamic baseline, and a group report showing the distribution of overall health indicators and baseline trends of the group.

[0111] Individual reports may include: a trend comparison chart showing the change curve of a student's indicator over the past several days, while overlaying the change of the dynamic baseline of their group during the same period. It directly shows whether the student is within, above, or below the baseline, and whether the trend deviates from the baseline; a radar chart showing the deviation of each health indicator; and warning status and history, showing the current warning level, trigger time, and historical warning records.

[0112] Group reports may include: a baseline evolution graph showing the trend of changes in the baseline bands of core health indicators for the class / grade, such as average sleep time and average steps, over the past semester, reflecting the overall shift in collective sleep or exercise habits; and distribution statistics showing the distribution of students at the current warning level through pie charts or bar charts.

[0113] Preferably, the system further includes an intervention suggestion knowledge base, and the early warning and report generation module matches and outputs personalized health intervention suggestions from the intervention suggestion knowledge base based on the specific deviation patterns identified by the individual comparison analysis unit.

[0114] In an embodiment of the invention, the intervention suggestion knowledge base is constructed by educational experts, psychologists, and physical education research groups jointly inputting experience-based intervention rules. After each manual intervention (such as a teacher's conversation), the administrator can record the intervention measures and the effect feedback after a period of time in the system. The system associates these cases with the student's risk pattern at the time of triggering.

[0115] By statistically analyzing the improvement rates of different intervention measures on similar risk patterns (the rate of decline in risk scores and the proportion of warnings lifted), the system can automatically sort and optimize the suggestions in the knowledge base, prioritizing the most effective suggestions and forming a closed loop of data-driven intervention, feedback, and optimization.

[0116] The intervention recommendation knowledge base stores rules such as IF (exercise deviation < -30% AND sleep deviation ~ 0) THEN (recommendation list = [encourage participation in activities during breaks, communicate with the physical education teacher]). The system can generate specific and actionable recommendation texts based on the current risk vector matching rules. For example, a group report could include health intervention recommendations: "The overall deep sleep rate of third graders decreased by 5% this week; consider adjusting the end time of evening self-study." "More than 15% of students in this class lack exercise; increase fun sports activities."

[0117] In this invention, the tiered early warning mechanism implemented by the early warning and report generation module automatically generates corresponding early warning signals based on different degrees of individual health deviation and comprehensive risk scores. Through tiered early warning, not only can individual health risks be accurately identified and responded to, but resource waste is also effectively avoided, achieving a balance between early intervention and crisis management, and significantly improving the effectiveness of the campus health management system. Simultaneously, the system provides customized visual reports for individuals and groups, presenting trend comparisons, risk distribution, and resource allocation suggestions, providing educators and administrators with an intuitive and comprehensive overview of the health situation, assisting them in making more scientific decisions. Furthermore, the establishment of an intervention suggestion knowledge base, through continuous learning and optimization, ensures the effectiveness and relevance of intervention measures, further enhancing the system's adaptability and intelligence. This significantly improves the accuracy and efficiency of the campus health management system in assessment, early warning, and response, providing strong technical support for safeguarding students' physical and mental health.

[0118] The workflow of the above embodiments is as follows:

[0119] Data collection: The bracelet worn by Zhang San (MAC:05:33:85) encrypted and uploaded his AID (A7f3e9) and the time he fell asleep at 23:00 on the same day.

[0120] Data cleaning and clustering: The data cleaning unit verifies and receives the data. The clustering unit assigns the group label G2025_3 to the data entry based on the mapping A7f3e9 -> (Class of 2025, Class 3, Male).

[0121] Benchmark calculation (performed daily at midnight): The benchmark generation unit reads the sleep times of all students in Class G2025_3 over the past 7 days and calculates the weighted average. 22:40, weighted The time is 23:20. Therefore, the baseline for bedtime on that day is [22:40, 23:20].

[0122] Individual Comparison: Zhang San went to sleep at 11:00 PM that day, falling within the baseline range, with a single indicator deviation D (sleep) of 0. However, the system also detected that his average daily steps were 4000, while the baseline range for the class's average daily steps was [5500, 8000]. The calculated D (average daily steps) = (5500-4000) / 5500 ≈ -0.27. After weighted calculation combining other health indicators, his overall risk score for that day was relatively high, placing him in the top 10% of the class.

[0123] Warning and Report: The system triggered a yellow warning and sent it to the homeroom teacher, Ms. Li.

[0124] Teacher Li reviewed Zhang San's individual report: The report chart showed that Zhang San's step count had been declining continuously over the past two weeks, falling below the lower limit of the class baseline, while his sleep time was generally normal. Based on the knowledge base rules, the report provided the following suggestions: This student's recent exercise level is significantly lower than the class average, but his daily routine is still regular. Recommendations: 1. Actively inquire about his physical condition during breaks; 2. Encourage him to participate in the afternoon class physical activities.

[0125] The grade leader reviewed the group report: The report showed a comparison of the average steps taken by each class in Grade 3 this week. Class 3 had the lowest median baseline and a wide baseline band. The report indicated that the overall physical activity level of Class 3 in Grade 3 was low, with significant individual differences. It was recommended that the grade's physical education team pay attention to the class's participation in physical education classes.

[0126] Based on the advice given, Teacher Li spoke with Zhang San and learned that he had reduced his activity due to a minor ankle sprain. The teacher showed concern and arranged for him to do some upper limb exercises. A week later, Zhang San's step count rebounded, and the warning was automatically lifted. This minor injury was stored as valid context by the system for future interpretation of similar patterns.

[0127] like Figure 2 As shown, in a second aspect, embodiments of the present invention also provide a campus health management method based on population benchmark analysis, comprising:

[0128] Collect individual student health and behavioral data and anonymize the data.

[0129] Based on the student group attributes, the anonymized individual student data were clustered into homogeneous analysis groups;

[0130] Within each of the analysis groups, a dynamic statistical baseline band for one or more health indicators is calculated based on historical data within a set time window.

[0131] Each student's individual data is compared with the dynamic statistical baseline of the corresponding health indicator of their analysis group, and the relative deviation of the individual health indicator is calculated.

[0132] Based on the results of individual comparative analysis, a tiered early warning system is triggered, and a visual health report is generated.

[0133] Finally, it should be noted that the above embodiments are only preferred embodiments of the present invention, and the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A campus health management system based on population benchmark analysis, characterized in that, include: The data acquisition module is used to collect individual students' health and behavioral data and to perform data anonymization processing. The data processing and analysis module includes: The group clustering unit is used to cluster anonymized student individual data into homogeneous analysis groups based on student group attributes; the benchmark generation unit is used to calculate the dynamic statistical benchmark band of one or more health indicators within each analysis group based on historical data within a set time window; the individual comparison analysis unit is used to compare each student individual data with the dynamic statistical benchmark band of the corresponding health indicator of its analysis group and calculate the relative deviation of the individual health indicator. The early warning and report generation module is used to trigger tiered early warnings based on individual comparative analysis results and generate visualized health reports.

2. The campus health management system according to claim 1, characterized in that, The benchmark generation unit uses a weighted time window method to calculate a dynamic statistical benchmark band for health indicators. The length of the time window is adaptively adjusted according to the intensity of the current calendar event. The calculation of the dynamic statistical benchmark band is based on the weighted 25th percentile. ) and the weighted 75th percentile ( The middle 50% of individuals are defined.

3. The campus health management system according to claim 2, characterized in that, The reference generation unit includes: The time decay weighting subunit is used to assign weights to the data within the time window using an exponential decay function. The weighted quantile baseline band sub-unit is used to sort the data by time and calculate the cumulative weight percentage, and then calculate the weighted quantile through linear interpolation.

4. The campus health management system according to claim 3, characterized in that, The time decay weight subunit adjusts the weight of each time point within the time window, wherein the determination of the weight depends on the distance from the current time point and the influence intensity of related calendar events.

5. The campus health management system according to claim 1, characterized in that, The individual comparative analysis unit uses a weighted scoring model to calculate a multidimensional health risk assessment score and trend risk by comprehensively considering the relative deviation of multiple health indicators.

6. The campus health management system according to claim 5, characterized in that, The individual comparative analysis unit uses the Sigmoid function to convert the relative deviation of individual health indicators into risk sub-scores in the range of 0 to 1, and then calculates the individual's multidimensional health risk assessment score through a weighted scoring model. The selection of the Sigmoid function parameters is based on the dispersion of health indicators and the distribution characteristics of actual deviation.

7. The campus health management system according to claim 1, characterized in that, The preset group attributes on which the group clustering unit is based include at least the student's grade and class.

8. The campus health management system according to claim 1, characterized in that, The reports generated by the early warning and report generation module include: individual reports that show the trend of individual data and the comparison of the dynamic baseline of the group, and group reports that show the distribution of the overall health indicators of the group and the baseline trend.

9. The campus health management system according to claim 1, characterized in that, The system also includes an intervention suggestion knowledge base. The early warning and report generation module matches and outputs personalized health intervention suggestions from the intervention suggestion knowledge base based on the specific deviation patterns identified by the individual comparison analysis unit.

10. A campus health management method based on population benchmark analysis, characterized in that, include: Collect individual student health and behavioral data and anonymize the data. Based on the student group attributes, the anonymized individual student data were clustered into homogeneous analysis groups; Within each of the analysis groups, a dynamic statistical baseline band for one or more health indicators is calculated based on historical data within a set time window. Each student's individual data is compared with the dynamic statistical baseline of the corresponding health indicator of their analysis group, and the relative deviation of the individual health indicator is calculated. Based on the results of individual comparative analysis, a tiered early warning system is triggered, and a visual health report is generated.

Images