Method, device and equipment for assessing risk of myopia of teenagers by fusing multi-source data

Abstract

The application provides a kind of fusion multi-source data's myopia risk assessment method, device and equipment of adolescent, it is related to data processing technical field.The method comprises: obtaining the multi-source data of user about myopia risk and historical high-risk event data, multi-source data includes genetic factor, life habit, environmental factor, optometry parameter and intervention record dimension data;Based on multi-source data, determine the characteristic index deviation of each dimension data;According to the characteristic index deviation of each dimension data and historical high-risk event data, determine the comprehensive myopia health index of user;According to comprehensive myopia health index, determine the myopia risk level, predict myopia trend and myopia prevention plan of user.The application can consider the objectivity of evaluation, prediction accuracy, intervention scheme adaptability and engineering landing practicability, overcome the defects that existing myopia prevention and control technology research and judge dimension is single, trend can only be judged after the event, intervention scheme template solidification, difficult to scale landing.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, and device for assessing the risk of myopia in adolescents by integrating multi-source data. Background Technology

[0002] Current efforts to prevent and control myopia among teenagers are progressing steadily and showing initial results. Since the implementation of various myopia prevention and control measures, the overall myopia rate among children and adolescents nationwide has been declining continuously, effectively reversing the trend of continuous increase in myopia rates. Schools across the country have generally implemented prevention and control measures such as vision screening, looking into the distance during breaks, and outdoor activities. The grassroots vision health service system is continuously improving, and a collaborative prevention and control mechanism between families and schools is gradually taking shape. However, the overall prevention and control situation remains severe, and problems remain prominent.

[0003] Myopia among teenagers is characterized by its early onset, high incidence, and high degree of severity. The myopia rate among primary school students, junior high school students, and senior high school students increases progressively, with many young children experiencing vision problems prematurely, and the proportion of those with high myopia continuing to rise. Heavy academic pressure, excessive use of electronic products, insufficient outdoor lighting, and improper eye habits remain the core contributing factors to the high incidence of myopia. At the same time, weak awareness of prevention and control among some families and inadequate implementation of daily eye care practices have led to a widespread decline in students' vision during holidays. Therefore, the normalized prevention and control of myopia among teenagers remains a long and arduous task.

[0004] Existing myopia prevention and control platforms and technologies generally suffer from multiple shortcomings, including difficulty in implementation, low accuracy in assessment, and rigid decision-making. Various monitoring data belong to independent systems, lacking a unified coding mapping and standardized fusion mechanism. This makes it impossible to quantify the accelerated myopia risk caused by the combined effects of genetics, environment, and eye-use behavior, allowing only single-dimensional independent analysis. Risk assessment relies on fixed thresholds and doctors' subjective experience, failing to dynamically adjust benchmarks based on age and genetic susceptibility. This easily leads to missed warnings for high-risk children and excessive intervention for healthy children, and also lacks quantitative models to objectively assess cumulative eye damage. Existing models can only rely on post-examination reviews of myopia progression, lacking long-term quantitative prediction capabilities and making it difficult to predict future changes in axial length and refractive power, thus hindering proactive intervention. Intervention plans use uniform templates, failing to dynamically prioritize based on individual risk, myopia trends, and family conditions. Furthermore, intervention effects cannot be fed back to optimize the model, lacking adaptive closed-loop systems. Most academic models rely on niche, specialized testing data, lacking lightweight decomposition, making them difficult to adapt for large-scale deployment in multiple scenarios such as schools, homes, and hospitals, resulting in insufficient engineering practicality. Summary of the Invention

[0005] This invention provides a method, device, and equipment for assessing the risk of myopia in adolescents by integrating multi-source data, in order to solve the problems of existing myopia assessments being based on a single dimension, relying on human experience, being unable to predict in advance, having fixed intervention plans, and having models that are difficult to implement on a large scale.

[0006] In a first aspect, embodiments of the present invention provide a method for assessing the risk of myopia in adolescents by fusing multi-source data, including: Acquire multi-source data on myopia risk and historical high-risk event data of users; wherein the multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention record dimensions; Based on the multi-source data, the deviation of the characteristic indicators of each dimension of the data is determined; Based on the deviation of the characteristic indicators of each dimension of data and the historical high-risk event data, the user's comprehensive myopia health index is determined; Based on the comprehensive myopia health index, the user's myopia risk level is determined, myopia trends are predicted, and a myopia prevention and control plan is developed.

[0007] Secondly, embodiments of the present invention provide a device for assessing the risk of myopia in adolescents by integrating multi-source data, comprising: The acquisition module is used to acquire multi-source data on the user's myopia risk and historical high-risk event data; wherein, the multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention record dimensions; The first processing module is used to determine the deviation of the feature indicators of each dimension of the data based on the multi-source data. The second processing module is used to determine the user's comprehensive myopia health index based on the deviation of the characteristic indicators of each dimension of data and the historical high-risk event data; The myopia risk assessment module is used to determine the user's myopia risk level, predict myopia trends, and formulate myopia prevention and control plans based on the comprehensive myopia health index.

[0008] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect or any possible implementation thereof.

[0009] In this embodiment of the invention, multi-source data on the user's myopia risk and historical high-risk event data are acquired. The multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention records. Based on the multi-source data, the deviation of characteristic indicators for each dimension is determined. Based on the deviation of characteristic indicators for each dimension and historical high-risk event data, the user's comprehensive myopia health index is determined. Based on the comprehensive myopia health index, the user's myopia risk level, predicted myopia trend, and myopia prevention and control plan are determined. This approach objectively measures the user's eye health by combining data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention records, reducing biases caused by doctors' subjective judgments and obtaining a more accurate comprehensive myopia health index. Based on this comprehensive myopia health index, the user's myopia risk level, predicted myopia trend, and myopia prevention and control plan are determined. This approach balances the objectivity of the assessment, the accuracy of the prediction, the adaptability of the intervention plan, and the practicality of engineering implementation, overcoming the shortcomings of existing myopia prevention and control technologies, such as single-dimensional assessment, the inability to determine trends retrospectively, fixed intervention plan templates, and difficulty in large-scale implementation. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the implementation of the adolescent myopia risk assessment method that integrates multi-source data, as provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the structure of the adolescent myopia risk assessment device that integrates multi-source data provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0011] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0012] See Figure 1 The document illustrates a flowchart of the implementation of the adolescent myopia risk assessment method that integrates multi-source data, as provided in an embodiment of the present invention, and details are as follows: Step 101: Obtain multi-source data on myopia risk and historical high-risk event data from users.

[0013] The multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention records.

[0014] For example, the multi-source influencing factors and characteristic indicators (i.e., multi-source data) of intelligent prevention and control of myopia risk in adolescents are shown in Table 1 below: Table 1. Multi-source influencing factors and characteristic indicators of intelligent prevention and control of myopia risk in adolescents Step 102: Based on multi-source data, determine the deviation of characteristic indicators for each dimension of the data.

[0015] In one embodiment, step 102 may include: For each dimension of data in the multi-source data, the deviation of the feature index of the data dimension is obtained based on the data dimension, the mean of the non-myopia samples corresponding to the data dimension, and the standard deviation of the non-myopia samples.

[0016] For example, the deviation of each feature indicator in each dimension of the data can be calculated using the following formula: ; : The deviation of the 0th adolescent and the i-th indicator in the scenario on day t (i.e., the deviation of the characteristic indicator), which measures the degree to which the indicator deviates from the healthy group of the same age and genetics. The larger the value, the higher the risk.

[0017] : t-day youth o-the measured value of the i-th indicator.

[0018] : The baseline mean of normal samples in the same age group (k) and genetic grade group (h) (that is, the mean of non-myopia samples corresponding to the i-th indicator). For example, h values ​​of 1, 2, and 3 represent mild, moderate, and high myopia in both parents, respectively.

[0019] The robust standard deviation of indicators in the same age and genetic group (i.e., the standard deviation of non-myopia samples corresponding to the i-th indicator) represents the normal fluctuation range of the indicator and distinguishes between physiological fluctuations and pathological abnormalities.

[0020] : Load correction item, corrects the amplifying effect of near-field eye load on index fluctuations.

[0021] : Load regression correction coefficient, typically taken as 0.22.

[0022] Numerical stability constant to prevent the denominator from being 0.

[0023] For example: a tester Age, high genetic grade Daily outdoor natural light duration Corresponding , , ,but: ; Taking the characteristic indicators including daily outdoor natural light duration, axial length (AL), electronic screen time, and indoor illuminance as examples, the deviations corresponding to daily outdoor natural light duration, AL, electronic screen time, and indoor illuminance are respectively: , , , .

[0024] Step 103: Determine the user's comprehensive myopia health index based on the deviation of characteristic indicators of each dimension of data and historical high-risk event data.

[0025] In one embodiment, step 103 may include: The user's dimensional space health index is determined based on the deviation of characteristic indicators of each dimension of data, as well as the dimensional coupling coefficient and average indicator deviation between each dimension of data.

[0026] Based on historical high-risk event data and a preset time decay model, the impact index of a user's historical high-risk events is determined.

[0027] The user's comprehensive myopia health index is determined based on the dimensional space health index and the historical high-risk event impact index.

[0028] In this embodiment, considering the coupling risks of five dimensions—genetic factors, lifestyle habits, environmental factors, refraction parameters, and intervention records (e.g., high genetic predisposition combined with low outdoor activity can amplify the risk of myopia)—a dimensional coupling coefficient is introduced. Calculate the user's dimensional space health index : ; For single-indicator weighting, the entropy weighting method can be used to calculate the objective basis weight based on the dispersion of each indicator's data. The stronger the indicator's ability to distinguish risk, the greater its objective basis weight. Simultaneously, clinical constraint corrections are introduced, appropriately increasing the weight of indicators clinically proven to have a strong impact on myopia (such as axial length) to meet the following requirements. .

[0029] : This is a cross-dimensional coupling coefficient that quantifies the amplified effect of multiple factors, including genetics, environment, and eye use. For example, a value of 0.28 can be used.

[0030] The average deviation of the related dimension indicators (i.e., the average deviation of the data of each dimension) represents the synergistic deterioration effect of multiple risks. This means that the dimensions of physiological optometry, genetic susceptibility, behavioral habits, environmental lighting, and intervention implementation are not independent of each other and have an interactive effect. The risk of myopia caused by the deviation of a single dimension indicator will be amplified or inhibited by the status of other dimensions.

[0031] For example: the weighting of single-indicator dimensions such as daily outdoor natural light duration, axial length (AL), electronic screen time, and indoor illuminance. If we take values ​​of 0.25, 0.45, 0.20, and 0.10 respectively, then: ; Cross-dimensional coupling coefficient Considering the coupling of genetic and habitual dimensions, the average deviation between the two is... Dimensional coupling coefficient =0.7, then .

[0032] Then the dimensional space health index .

[0033] In this embodiment, the user's dimensional space health index is determined based on the deviation of characteristic indicators of each dimension of data, as well as the dimensional coupling coefficient and average indicator deviation between the data of each dimension. It not only reflects the effects of single-indicator factors such as daily outdoor natural light duration, axial length (AL), electronic screen time, and indoor illuminance, but also considers the multi-risk synergistic effects of cross-dimensional coupling.

[0034] Based on this, considering the persistence of myopia damage and the continued impact of past high-risk conditions on current health scores, a pre-defined time decay model is used to determine the impact index of a user's historical high-risk events. : ; Dates of historical high-risk events (such as the date of events involving rapid axial elongation or heavy eye strain).

[0035] Risk level coefficient, ranging from 1 to 6, such as 6 for sudden increase in axial length.

[0036] : Indicator type correction coefficient, such as 1.2 for events related to axial length and 1.0 for environmental events.

[0037] : Attenuation coefficient, an example value of 0.008, which means that the risk will decrease to 10% of the initial value after 3 months.

[0038] Among them, the dates of occurrence of historical high-risk events Risk level coefficients can be obtained from historical high-risk event data. Indicator type correction coefficient attenuation coefficient It can be determined based on a preset time decay model.

[0039] For example, rapid axial elongation of the eye during an eye exam 30 days ago (high-risk event), event level. Indicator correction coefficient Interval days ,but: =5.6635; Furthermore, based on the dimensional space health index and the historical high-risk event impact index, the user's comprehensive myopia health index can be determined. .

[0040] : This is the coefficient for historical memory items. An exemplary value can be 0.35. This embodiment does not limit the specific value.

[0041] Following the example above, we can obtain Mapped to ,but It is 40.7.

[0042] This embodiment proposes a multi-data fusion method for assessing myopia in adolescents, which integrates scene deviation, correlation dimensions, and historical risks to quantify the eye health index, abandons fixed thresholds, and dynamically adjusts the assessment benchmark according to age and genetic background.

[0043] Step 104: Based on the comprehensive myopia health index, determine the user's myopia risk level, predict myopia trends, and formulate a myopia prevention and control plan.

[0044] In one embodiment, determining a user's myopia risk level based on a comprehensive myopia health index includes: The user's comprehensive myopia risk score is calculated based on the comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index over a preset period, and the user's myopia genetic level.

[0045] The user's myopia risk level is determined based on the comprehensive myopia risk score.

[0046] In this embodiment, based on the obtained comprehensive myopia health index, The rate of change of the comprehensive myopia health index in the past 3 months (i.e., the preset period of 3 months). Myopia hereditary grade Three-dimensional indicators to construct a comprehensive myopia risk score : ; For example, four levels of risk criteria can be used (each level of risk criteria can use dynamic thresholds, such as lowering the overall threshold by 10 for children under 12 years old to adapt to the development of children's eyes): Normal risk (0≤Q<30): No risk of myopia progression, basic routine monitoring.

[0047] Pay attention to the risks (30≤Q<55): There is a risk of myopia inducing development, so increase outdoor intervention.

[0048] Warning of risk (55≤Q<80): High risk of rapid myopia progression, initiate optical correction intervention.

[0049] High risk (80≤Q≤100): Risk of conversion to high myopia, combined with ophthalmic drugs and optical dual intervention.

[0050] For example, weight Comprehensive myopia health index Rate of change of comprehensive myopia health index in the past 30 days (The characterization index continues to rise, indicating accelerated myopia progression); Genetic grade (Indicating high myopia in both parents), then: ; Therefore, this sample is determined to be of normal risk.

[0051] This embodiment integrates the comprehensive myopia health index and its rate of change, as well as genetic level, to establish a standard for judging the risk of myopia in adolescents, classify dynamic myopia risk criteria, achieve objective quantitative grading, and reduce underreporting and false reporting.

[0052] In one embodiment, determining a user's predicted myopia trend based on a comprehensive myopia health index includes: Based on multi-source data within a preset time window, obtain the variance of the data gradient for each dimension and the covariance between the data gradients of every two dimensions.

[0053] Based on the covariance and the variance of the corresponding dimensional data gradient, the dimensional coupling gradient of the multi-source data is obtained.

[0054] Obtain the mean and standard deviation of the comprehensive myopia health index within a preset time window, as well as the incremental changes in the comprehensive myopia health index at the beginning and end of the preset time window, denoted as the window mean, window standard deviation, and window increment.

[0055] We perform a weighted summation of the window mean, window standard deviation, window change increment, and dimensional coupling gradient, and use the weighted summation result as the user's state evolution intensity regarding myopia trend.

[0056] Based on the intensity of state evolution and multi-source data, the predicted myopia trend of users is determined.

[0057] For example, predicting myopia trends includes refractive prediction values, axial length prediction values, myopia progression probability, and myopia progression rate; determining a user's predicted myopia trend based on state evolution intensity and multi-source data may include: Based on multi-source data, the annual changes in axial length and spherical lens power of users are calculated.

[0058] The input feature set is constructed based on multi-source data, annual changes in axial length, annual changes in spherical lens, and state evolution intensity.

[0059] Input the input feature set into a preset lightweight random forest model to obtain the model's prediction results.

[0060] The model prediction results are validated by clinical logic constraints, and the model prediction results are output as follows: refractive prediction value, axial length prediction value, myopia progression probability, and myopia progression rate.

[0061] In this embodiment, myopia trend prediction can obtain short-term (e.g., 3 months), medium-term (e.g., 6 months), and long-term (e.g., 12 months) prediction results. The purpose is to predict the risk of myopia onset and rapid myopia progression in advance, identify individuals with high risk of myopia progression, provide quantitative basis for risk assessment and dynamic adjustment of intervention plans, and especially provide an important basis for plan matching methods, avoiding relying solely on the results of a single examination for delayed intervention, and realizing the shift from post-screening to early warning.

[0062] Specifically, temporal features can be constructed, such as using a 30-day sliding window to extract features and construct the state evolution intensity. : ; in, This is the average health index within the window (i.e., the window mean). This represents the increment of the health index change at the beginning and end of the window (i.e., the window change increment). Standard deviation of health index fluctuation within the window (i.e., window standard deviation). Spatial coupling gradient between risk dimensions (i.e., dimensional coupling gradient).

[0063] in It mainly quantifies the interaction gain effect among multiple dimensions such as optometric physiological parameters, genetic susceptibility, lifestyle habits, ambient light, and intervention implementation. It is not a single variable rate of change, but rather a representation of how a change in one dimension amplifies or inhibits the driving force of another dimension on myopia progression.

[0064] For example, the covariance normalization of the sequence change trend can be used to describe the driving strength of the synchronous change in dimension j when dimension i changes, thus obtaining the cross-dimensional coupling gradient: The covariance of two temporal gradient sequences represents whether the directions of change in the two dimensions are synchronized. : Variance of the temporal gradient of dimension i.

[0065] For example, the weights can be The weights can be calibrated through tens of thousands of time-series ablation experiments on adolescents, corresponding to the current average eye condition, short-term progression rate, degree of state fluctuation, and multi-risk superposition gradient effect, and uniformly characterize the overall deterioration trend of myopia in adolescents in the past 30 days.

[0066] For example, the 30-day window average Increment Window fluctuation intensity dimensional gradient ,but: ; Next, we will establish a periodic prediction model: Model parameter sources: The training set consists of monthly time-series data of teenagers, with a training / test time-series ratio of 8:2; there are 200 decision trees with a maximum depth of 8. Each decision tree is traversed layer by layer along the feature threshold, and the single-tree prediction value is output when the leaf node is reached.

[0067] This trend prediction module receives standardized multi-source feature data output from the myopia health assessment module (i.e., during the comprehensive myopia health index assessment). The input data includes state evolution intensity. All data, including individual genetic characteristics, behavioral data, environmental factors, optometry parameters, and intervention compliance, can be obtained through ophthalmological optometry and recording.

[0068] Step a: Data preprocessing. Outlier removal and missing value imputation are performed on all input data. Qualitative information is converted into standardized numerical features, and redundant features with low contribution are removed. This completes the feature selection for model lightweighting, reducing the computational burden on the equipment.

[0069] Step b: Construct time-series derived features. Based on historical records, calculate the annual change in axial length ΔAL and the annual change in spherical lens ΔS, transforming static detection indicators into dynamic time-series features that reflect the rate of myopia progression. These features, together with the original features, form the model input feature set.

[0070] Step c: Lightweight Random Forest Model Inference. Input the processed feature set into the trained lightweight random forest model, with a training / test time split of 8:2; 200 decision trees, maximum depth 8. Each decision tree is traversed layer by layer along the feature threshold using binary search, and the single-tree prediction value is output when the leaf node is reached.

[0071] Step d: Clinical logic constraint verification. Perform a rationality check on the model's original output, and remove abnormal predictions that exceed physiological boundaries; when the prediction results deviate significantly from the normal range for the corresponding age group, mark them for manual review to prompt ophthalmologists to intervene and make corrections, thus avoiding erroneous results that are out of touch with clinical reality due to purely algorithmic output.

[0072] Step e: Module result output. After verification, the output includes: 1) short-, medium- and long-term refractive prediction values ​​and axial length prediction values; 2) probability of myopia onset / progression P; and 3) myopia progression rate label.

[0073] Step f: Closed-loop iterative update. After obtaining the subject's subsequent actual refraction data, the actual measured values ​​are compared with the previous model predictions, the absolute error of the deviation is calculated, and the model parameters are fine-tuned to achieve continuous iterative optimization of the prediction model.

[0074] Verification showed that the prediction model obtained using this embodiment had a 3-month axial length prediction deviation of 0.07 mm and a 12-month high myopia prediction accuracy of 0.917, which met the clinical prevention and control requirements.

[0075] This embodiment proposes a trend prediction method based on a lightweight random forest regression model. It combines a prediction model that integrates full-dimensional time-series indicators with the intensity characteristics of myopia state evolution to predict the development trend of myopia. The prediction model itself does not directly output the risk level, but it outputs quantitative prediction values, progression probability, and progression rate. Combining the above parameters can correct the risk assessment results and serve as the development trend of the risk assessment.

[0076] In one embodiment, a myopia prevention and control plan for a user is determined based on a comprehensive myopia health index, including: The user's comprehensive myopia risk score is calculated based on the comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index over a preset period, and the user's myopia genetic level.

[0077] Obtain the individual risk scenario matching degree and family implementation feasibility indicators of each preset myopia prevention plan for users.

[0078] Based on individual risk scenario matching degree, comprehensive myopia risk score, myopia progression probability and family implementation feasibility indicators, the matching priority of users with each preset myopia prevention and control plan is obtained.

[0079] The user's myopia prevention and control plan is determined based on the matching priority between the user and each myopia prevention and control plan.

[0080] For example, obtaining the user's individual risk scenario matching degree for each preset myopia prevention and control plan includes: comparing the user's individual risk scenario with the matching conditions of each preset myopia prevention and control plan, and obtaining the user's individual risk scenario matching degree for each preset myopia prevention and control plan based on the comparison results; Based on individual risk scenario matching degree, comprehensive myopia risk score, myopia progression probability, and family implementation feasibility indicators, the matching priority between users and various preset myopia prevention and control plans is obtained, including: The matching priority of users with each preset myopia prevention and control plan is obtained by weighting and summing the individual risk scenario matching degree, myopia comprehensive risk score, myopia progression probability and family implementation feasibility indicators.

[0081] For example, standardized intervention plans (i.e., pre-set approximate prevention plans) may include outdoor light intervention, defocus optical correction, near-vision control, low-concentration atropine drug intervention, and eye accommodation training. The matching priority of the pre-set myopia prevention plan for adolescents can be calculated according to the following formula. The higher the matching priority value, the more recommended the plan is: ; The match degree between plan o and the individual risk scenario of adolescent p at time t, with a value of 0 to 1. It can be achieved by comparing the individual risk scenario of adolescent p with the fit conditions of plan o one by one. If a fit condition is met, the corresponding weight score is added; otherwise, it is deducted. The original scores are normalized to a maximum and minimum value and mapped to an interval. ,get .

[0082] Current comprehensive myopia risk score.

[0083] The 12-month high myopia prediction probability can be obtained through the above-mentioned periodic prediction model.

[0084] Feasibility indicators for family implementation (such as parental time spent with family, economic conditions, and normalized values ​​of equipment ownership). First, each sub-indicator can be quantitatively scored, and the objective weight of each indicator can be calculated using the entropy weight method. The weighted sum is then used to obtain the original score. Finally, the results can be mapped to a more suitable value using max-min normalization. The interval is obtained .

[0085] Dynamic weights Adjusted according to risk level: Enhanced in high-risk scenarios Weighting focuses on improving the matching degree of outdoor interventions in specific scenarios.

[0086] For example: Focus on risk scenario weighting Two candidate plans were selected for calculation and comparison: Plan P1: Daily intensive outdoor intervention (matching the risks to be monitored): Scene matching Comprehensive risks Probability of high myopia Feasibility of implementation in families ,but: ; Contingency Plan P2: Dual intervention with defocus lens and atropine (suitable for high-risk situations): Scene matching ; Probability of high myopia Feasibility of implementation in families ,but: ; The matching conclusion M1>M2, so P1 is recommended as the optimal intervention plan.

[0087] In this embodiment, dynamic scenario weights are adjusted by combining risk level, probability prediction, and feasibility of family implementation to form a matching scheme for the prevention and control of myopia in adolescents, thereby obtaining a closed-loop prevention and control strategy.

[0088] The above embodiments construct four collaborative modules: myopia health assessment, risk determination, development trend prediction, and contingency plan matching. The myopia health assessment module abandons fixed judgment standards, objectively measuring eye health by incorporating factors such as age and genetics, reducing biases caused by doctors' subjective judgments. The risk determination module comprehensively classifies risk levels based on current eye condition, myopia progression rate, and congenital genetic factors, reducing the problem of missed diagnoses of high-risk cases and excessive intervention for ordinary children. The development trend prediction module constructs myopia state evolution intensity characteristics based on a 30-day sliding window, using lightweight random forests to quantitatively predict short-, medium-, and long-term myopia progression, achieving early warning of myopia risk. The contingency plan matching module provides suitable prevention and control plans based on individual circumstances and can be continuously optimized based on intervention effects. This invention balances objective assessment, accurate prediction, adaptability of intervention plans, and practicality for engineering implementation, overcoming the shortcomings of existing myopia prevention and control technologies, such as single-dimensional assessment, retrospective trend judgment, fixed intervention plan templates, and difficulty in large-scale implementation.

[0089] In this embodiment of the invention, data such as genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention execution records are used as data sources. By calculating the deviation of characteristic indicators and uniformly mapping them to dimensionless deviations, a comprehensive myopia health index is obtained by analyzing the dimensional space health status and historical myopia risk memory decay, and a quantitative assessment of eye health is performed. Combining the comprehensive myopia health index, progression rate, and genetic level, a comprehensive risk score is obtained to achieve four-level risk determination. Temporal features are extracted based on a 30-day sliding window and the intensity of state evolution is calculated. A lightweight random forest model is used to complete myopia trend prediction. Then, a personalized intervention plan is generated by matching through a dynamic weighted priority algorithm. Screening reports, risk warnings, and intervention guidance are output externally. At the same time, feedback data from re-examination and intervention check-in are sent back to correct the model parameters of each module, forming a complete closed-loop process. This approach aims to construct an intelligent risk prevention and control method for adolescent myopia assessment, risk determination, trend prediction, and contingency plan decision-making. Based on factors such as genetics, behavior, environment, testing, and intervention, it integrates multi-source data to build a standardized quantitative model. This model accurately classifies myopia risk, quantitatively predicts the evolution of axial length and refractive error, and dynamically generates individualized prevention and control plans based on individual risk and family conditions. Simultaneously, it establishes a closed-loop feedback model for intervention effects for continuous iteration. This addresses the industry pain points of current myopia assessment methods, such as single dimensions, reliance on human experience, inability to predict in advance, rigid intervention plans, and difficulty in scaling up models. It can scientifically slow the progression of myopia and reduce the increase in the number of people with high myopia. Furthermore, it supports collaborative prevention and control among schools, families, and medical institutions, reduces the long-term medical burden of eye diseases for the entire population, and improves the public service system for adolescent eye health.

[0090] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0091] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0092] Figure 2 A schematic diagram of the structure of the adolescent myopia risk assessment device that integrates multi-source data provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 2 As shown, the adolescent myopia risk assessment device that integrates multi-source data includes: The acquisition module 21 is used to acquire multi-source data on myopia risk and historical high-risk event data of users; among which, multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters and intervention record dimensions; The first processing module 22 is used to determine the deviation of characteristic indicators of data in each dimension based on multi-source data; The second processing module 23 is used to determine the user's comprehensive myopia health index based on the deviation of characteristic indicators of data from each dimension and historical high-risk event data. The myopia risk assessment module 24 is used to determine the user's myopia risk level, predict myopia trends, and develop myopia prevention and control plans based on the comprehensive myopia health index.

[0093] In one possible implementation, the first processing module 22 is specifically used for: For each dimension of data in the multi-source data, the deviation of the feature index of the data dimension is obtained based on the data dimension, the mean of the non-myopia samples corresponding to the data dimension, and the standard deviation of the non-myopia samples.

[0094] In one possible implementation, the second processing module 23 is specifically used for: Based on the deviation of characteristic indicators of each dimension of data, as well as the dimensional coupling coefficient and average indicator deviation between each dimension of data, the user's dimensional space health index is determined. Based on historical high-risk event data and a preset time decay model, determine the user's historical high-risk event impact index; The user's comprehensive myopia health index is determined based on the dimensional space health index and the historical high-risk event impact index.

[0095] In one possible implementation, the myopia risk assessment module 24 is specifically used for: The user's comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index during a preset period, and the user's myopia genetic level are used to calculate the user's comprehensive myopia risk score. The user's myopia risk level is determined based on the comprehensive myopia risk score.

[0096] In one possible implementation, the myopia risk assessment module 24 is specifically used for: Based on multi-source data within a preset time window, obtain the variance of the data gradient for each dimension and the covariance between the data gradients of every two dimensions; Based on the covariance and the variance of the corresponding dimensional data gradient, the dimensional coupling gradient of the multi-source data is obtained. The mean and standard deviation of the comprehensive myopia health index within a preset time window, as well as the incremental changes in the comprehensive myopia health index at the beginning and end of the window within the preset time window, are recorded as the window mean, window standard deviation, and window increment. We perform a weighted summation of the window mean, window standard deviation, window change increment, and dimensional coupling gradient, and use the weighted summation result as the intensity of the user's state evolution regarding myopia trend. Based on the intensity of state evolution and multi-source data, the predicted myopia trend of users is determined.

[0097] In one possible implementation, predicting myopia trends includes refractive prediction values, axial length prediction values, myopia progression probability, and myopia progression rate; the myopia risk assessment module 24 is specifically used for: Based on multi-source data, calculate the annual change in axial length and spherical lens for users; The input feature set is constructed based on multi-source data, annual changes in axial length, annual changes in spherical lens, and state evolution intensity. Input the input feature set into a preset lightweight random forest model to obtain the model prediction results; The model prediction results are validated by clinical logic constraints, and the model prediction results are output as follows: refractive prediction value, axial length prediction value, myopia progression probability, and myopia progression rate.

[0098] In one possible implementation, the myopia risk assessment module 24 is specifically used for: The user's comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index during a preset period, and the user's myopia genetic level are used to calculate the user's comprehensive myopia risk score. Obtain the individual risk scenario matching degree and family implementation feasibility indicators of each preset myopia prevention plan; Based on individual risk scenario matching degree, myopia comprehensive risk score, myopia progression probability and family implementation feasibility indicators, the matching priority of users with each preset myopia prevention and control plan is obtained; The user's myopia prevention and control plan is determined based on the matching priority between the user and each myopia prevention and control plan.

[0099] In one possible implementation, the myopia risk assessment module 24 is specifically used for: The user's individual risk scenario is compared with the matching conditions of each preset myopia prevention and control plan. Based on the comparison results, the user's individual risk scenario matching degree with each preset myopia prevention and control plan is obtained. The matching priority of users with each preset myopia prevention and control plan is obtained by weighting and summing the individual risk scenario matching degree, myopia comprehensive risk score, myopia progression probability and family implementation feasibility indicators.

[0100] Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Figure 3 As shown, the electronic device 3 of this embodiment includes a processor 30 and a memory 31. The memory 31 stores a computer program 32. When the processor 30 executes the computer program 32, it implements the steps in the various method embodiments described above. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module / unit in the various device embodiments described above.

[0101] For example, computer program 32 may be divided into one or more modules / units, which are stored in memory 31 and executed by processor 30 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 32 in electronic device 3.

[0102] Electronic device 3 may include, but is not limited to, processor 30 and memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 3 may also include input / output devices, network access devices, buses, etc.

[0103] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0104] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0105] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for assessing the risk of myopia in adolescents by integrating multi-source data, characterized in that, include: Acquire multi-source data on users' myopia risk and historical high-risk event data; wherein, the multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention records. Based on the multi-source data, the deviation of the characteristic indicators of each dimension of the data is determined; Based on the deviation of the characteristic indicators of each dimension of data and the historical high-risk event data, the user's comprehensive myopia health index is determined; Based on the comprehensive myopia health index, the user's myopia risk level is determined, myopia trends are predicted, and a myopia prevention and control plan is developed.

2. The method for assessing the risk of myopia in adolescents by fusing multi-source data according to claim 1, characterized in that, Based on the multi-source data, the deviation of characteristic indicators for each dimension of the data is determined, including: For each dimension of the multi-source data, the deviation of the feature index of the data dimension is obtained based on the data dimension, the mean of the non-myopia samples corresponding to the data dimension, and the standard deviation of the non-myopia samples.

3. The method for assessing the risk of myopia in adolescents by fusing multi-source data according to claim 1, characterized in that, Based on the deviation of the characteristic indicators of each dimension of data and the historical high-risk event data, the user's comprehensive myopia health index is determined, including: Based on the deviation of the characteristic indicators of each dimension of data, as well as the dimensional coupling coefficient and average indicator deviation between each dimension of data, the user's dimensional space health index is determined. Based on the historical high-risk event data and the preset time decay model, the user's historical high-risk event impact index is determined; The user's comprehensive myopia health index is determined based on the dimensional space health index and the historical high-risk event impact index.

4. The method for assessing the risk of myopia in adolescents by fusing multi-source data according to claim 1, characterized in that, Based on the comprehensive myopia health index, the user's myopia risk level is determined, including: Based on the comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index during a preset period, and the user's myopia genetic level, calculate the user's comprehensive myopia risk score; The user's myopia risk level is determined based on the comprehensive myopia risk score.

5. The method for assessing the risk of myopia in adolescents by integrating multi-source data according to claim 1, characterized in that, Based on the comprehensive myopia health index, the user's predicted myopia trend is determined, including: Based on the multi-source data within a preset time window, obtain the variance of the data gradient for each dimension and the covariance between the data gradients of every two dimensions; Based on the covariance and the variance of the corresponding dimensional data gradient, the dimensional coupling gradient of the multi-source data is obtained; The mean and standard deviation of the comprehensive myopia health index within the preset time window, as well as the incremental changes of the comprehensive myopia health index at the beginning and end of the window within the preset time window, are recorded as the window mean, window standard deviation, and window increment. The window mean, the window standard deviation, the window change increment, and the dimensional coupling gradient are weighted and summed, and the weighted summation result is used as the user's state evolution intensity regarding myopia trend. Based on the state evolution intensity and the multi-source data, the user's predicted myopia trend is determined.

6. The method for assessing the risk of myopia in adolescents by fusing multi-source data according to claim 5, characterized in that, The predicted myopia trend includes refractive error prediction, axial length prediction, probability of myopia progression, and rate of myopia progression. Based on the state evolution intensity and the multi-source data, the user's predicted myopia trend is determined, including: Based on the multi-source data, calculate the annual change in the user's axial length and the annual change in spherical lens; The input feature set is constructed based on the multi-source data, the annual variation of axial length, the annual variation of spherical lens, and the state evolution intensity; The input feature set is input into a preset lightweight random forest model to obtain the model prediction results; The model prediction results are validated by clinical logic constraints, and the refractive prediction value, axial length prediction value, myopia progression probability, and myopia progression rate are output based on the validated model prediction results.

7. The method for assessing the risk of myopia in adolescents by integrating multi-source data according to claim 6, characterized in that, Based on the comprehensive myopia health index, a myopia prevention and control plan for the user is determined, including: Based on the comprehensive myopia health index, the rate of change of the user's comprehensive myopia health index during a preset period, and the user's myopia genetic level, calculate the user's comprehensive myopia risk score; Obtain the individual risk scenario matching degree and family implementation feasibility indicators of each preset myopia prevention plan; Based on the individual risk scenario matching degree, the comprehensive myopia risk score, the myopia progression probability, and the family implementation feasibility index, the matching priority of users with each preset myopia prevention and control plan is obtained; The user's myopia prevention and control plan is determined based on the matching priority between the user and each myopia prevention and control plan.

8. The method for assessing the risk of myopia in adolescents by integrating multi-source data according to claim 7, characterized in that, Obtain the individual risk scenario matching degree of each user for each preset myopia prevention plan, including: The user's individual risk scenario is compared with the matching conditions of each preset myopia prevention and control plan. Based on the comparison results, the user's individual risk scenario matching degree with each preset myopia prevention and control plan is obtained. Based on the individual risk scenario matching degree, the comprehensive myopia risk score, the myopia progression probability, and the family implementation feasibility index, the matching priority between the user and each preset myopia prevention and control plan is obtained, including: The matching priority of the user with each preset myopia prevention and control plan is obtained by weighting and summing the individual risk scenario matching degree, the comprehensive myopia risk score, the myopia progression probability, and the family implementation feasibility index.

9. A device for assessing the risk of myopia in adolescents by integrating multi-source data, characterized in that, include: The acquisition module is used to acquire multi-source data on the user's myopia risk and historical high-risk event data; wherein, the multi-source data includes data on genetic factors, lifestyle habits, environmental factors, optometry parameters, and intervention record dimensions; The first processing module is used to determine the deviation of the feature indicators of each dimension of the data based on the multi-source data. The second processing module is used to determine the user's comprehensive myopia health index based on the deviation of the characteristic indicators of each dimension of data and the historical high-risk event data; The myopia risk assessment module is used to determine the user's myopia risk level, predict myopia trends, and formulate myopia prevention and control plans based on the comprehensive myopia health index.

10. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 8.

Images