Eye health state monitoring and evaluating method and device based on big data

Abstract

The invention provides an eye health state monitoring and evaluation method and device based on big data, and relates to the technical field of health monitoring, and the method comprises the steps: obtaining first information, second information and third information; performing three-dimensional reconstruction processing according to the first information to obtain three-dimensional eye feature data; performing pattern recognition processing according to the second information to obtain historical health feature data; performing time sequence analysis according to the historical health feature data and the third information to obtain behavior-environment cumulative effect data; performing data fusion processing according to the historical health feature data and the behavior-environment cumulative effect data, and obtaining potential risk data by predicting potential risk factors of child eye health; and performing evaluation processing according to the three-dimensional eye feature data and the risk data to obtain an evaluation result. According to the method, the modes and trends related to the specific eye diseases are recognized by using the big data mining and mode recognition technology, and potential health problems can be recognized in advance.

Description

Technical Field

[0001] The present invention relates to the field of health monitoring technology, and in particular to a method and device for monitoring and evaluating eye health status based on big data. Background Art

[0002] With the continuous development of computer technology, especially the progress of big data processing technology and artificial intelligence algorithms, the field of health monitoring is gradually moving towards intelligence and refinement. In terms of eye health status monitoring, traditional methods such as vision chart tests and regular eye examinations rely on manual observation and subjective judgment, and cannot obtain and process large amounts of health data in real time. This method not only has a long detection cycle and is difficult to achieve continuous monitoring of individual eye health, but also cannot capture subtle changes in structures such as the retina and cornea, especially for the early prediction and prevention of eye diseases such as myopia development.

[0003] The introduction of big data technology provides new possibilities for solving these problems. By integrating multiple data sources such as electronic health records, daily behavior monitoring, and environmental sensors, computer systems can process large amounts of multi-dimensional data and analyze changes in children's eye health in real time. Using machine learning and pattern recognition technology, potential health risks can be mined from massive amounts of health data to identify early abnormal patterns and trends. In addition, big data processing frameworks such as MapReduce and Hadoop can effectively support the storage and calculation of large-scale data, so that the monitoring of eye health is no longer limited to regular physical examinations, but has shifted to a more dynamic and real-time global health status analysis.

[0004] Therefore, the application of existing technologies in eye health monitoring is far from reaching the technological potential of the big data era. There is an urgent need for an eye health status monitoring and evaluation method and device based on big data, which can make full use of computer intelligence, big data processing and machine learning technology to improve the accuracy, real-time and coverage of eye health monitoring. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for monitoring and evaluating eye health status based on big data to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] On the one hand, the present application provides a method for monitoring and evaluating eye health status based on big data, including:

[0007] Acquiring first information, second information and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, eye examination records, treatment records and family medical history, and the third information includes daily behavior data and environmental data;

[0008] Performing three-dimensional reconstruction processing according to the first information, and obtaining three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data;

[0009] performing pattern recognition processing based on the second information, identifying patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records, and family medical history in the electronic health record, and obtaining historical health feature data;

[0010] Performing a time series analysis based on the historical health characteristic data and the third information, by quantitatively calculating the cumulative effects of various behavior-environment combinations on eye damage and mapping them to the transformation of the three-dimensional structure of the eye, to obtain behavior-environment cumulative effect data;

[0011] Performing data fusion processing based on the historical health characteristic data and the behavior-environment cumulative effect data, and training a prediction model using big data analysis and supervised learning algorithms to obtain potential risk data by predicting potential risk factors for children's eye health;

[0012] An evaluation result is obtained by performing evaluation processing based on the three-dimensional eye feature data and the risk data, and the evaluation result includes current monitoring results and future eye health management recommendations.

[0013] On the other hand, the present application also provides an eye health status monitoring and evaluation device based on big data, comprising:

[0014] an acquisition module, configured to acquire first information, second information and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, eye examination records, treatment records and family medical history, and the third information includes daily behavior data and environmental data;

[0015] a reconstruction module, configured to perform three-dimensional reconstruction processing according to the first information, and obtain three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data;

[0016] an identification module, configured to perform pattern recognition processing according to the second information, identify patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records and family medical history in the electronic health record, and obtain historical health feature data;

[0017] An analysis module, configured to perform time series analysis based on the historical health feature data and the third information, and obtain behavior-environment cumulative effect data by quantitatively calculating the cumulative effects of various behavior-environment combination scenarios on eye damage and mapping them to the transformation of the three-dimensional structure of the eye;

[0018] A fusion module, used to perform data fusion processing based on the historical health characteristic data and the behavior-environment cumulative effect data, and use big data analysis and supervised learning algorithms to train a prediction model to obtain potential risk data by predicting potential risk factors for children's eye health;

[0019] An evaluation module is used to perform evaluation processing based on the three-dimensional eye feature data and the risk data to obtain an evaluation result, wherein the evaluation result includes current monitoring results and future eye health management recommendations.

[0020] The beneficial effects of the present invention are:

[0021] The present invention constructs three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, thereby achieving high-precision three-dimensional reconstruction of children's eye structure; by performing correlation analysis on historical ophthalmic examination records, treatment records and family medical history in electronic health records, and utilizing big data mining and pattern recognition technology, patterns and trends related to specific eye diseases can be identified, thereby being able to identify potential health problems in advance.

[0022] Other features and advantages of the present invention will be set forth in the following description, and in part will become apparent from the description, or may be understood by practicing embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flow chart of the eye health status monitoring and assessment method based on big data described in an embodiment of the present invention;

[0025] Figure 2 Flow chart of the U-Net algorithm described in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0028] Embodiment 1:

[0029] This embodiment provides a method for monitoring and evaluating eye health status based on big data.

[0030] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300, step S400, step S500 and step S600.

[0031] Step S100, obtaining first information, second information and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, the electronic health records include eye examination records, treatment records and family medical history, and the third information includes daily behavior data and environmental data;

[0032] It is understood that the three-dimensional eye structure data of the children to be monitored is obtained through eye imaging devices, such as optical coherence tomography (OCT) or high-resolution magnetic resonance imaging (MRI). Electronic health records are a big data collection containing multi-dimensional health data, covering children's medical history, diagnosis results, treatment processes and medication. Eye examination records contain detailed data of each eye examination, such as vision test, intraocular pressure measurement, etc., all of which are aggregated and processed by big data systems. Treatment records and family medical history are integrated by integrating multiple data sources. Family medical history can be generated from the genetic information module in the electronic health record or through the automated data mining and analysis algorithms of the big data system to help identify potential genetic risk factors. Daily behavior data is obtained through wearable devices (such as smart watches) and smartphone applications, covering children's activity levels, screen time, reading habits and sleep quality. Environmental data is automatically collected through environmental sensors and smart home systems, involving external factors such as lighting conditions, air quality, noise levels, etc. in which children are located. These data can be collected and transmitted in real time through Internet of Things (IoT) devices, providing an important reference for the impact of the environment on eye health. The big data platform integrates multi-dimensional environmental data from different devices and sensors through real-time data stream processing technology and distributed computing framework, providing comprehensive support and reference for analyzing the long-term impact of the environment on eye health.

[0033] Step S200, performing three-dimensional reconstruction processing according to the first information, and obtaining three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data;

[0034] Preferably, the three main features of retinal thickness, corneal curvature and lens morphology are identified by image processing and computer vision technology. The measurement of retinal thickness is achieved by analyzing the density changes at different levels in the three-dimensional eye image; the measurement of corneal curvature is achieved by detecting the shape and curvature changes of the corneal surface; the lens morphology data is obtained by extracting the morphological features of the lens through image processing algorithms, and the lens morphology data includes the curvature, thickness and position of the front and back surfaces.

[0035] Step S300, performing pattern recognition processing according to the second information, identifying patterns and trends related to specific eye diseases by performing correlation analysis on historical ophthalmological examination records, treatment records and family medical history in the electronic health record, and obtaining historical health feature data;

[0036] It can be understood that this step aggregates and standardizes multiple data sources in the electronic health record, and uses association analysis algorithms to effectively identify frequent patterns and association rules in the data, helping to identify potential eye disease-related patterns.

[0037] Step S400, performing time series analysis based on the historical health characteristic data and the third information, quantitatively calculating the cumulative effects of various behavior-environment combinations on eye damage, and mapping them to the transformation of the three-dimensional structure of the eye, to obtain behavior-environment cumulative effect data;

[0038] It can be understood that this step concretizes the behavior-environment cumulative effects into visible three-dimensional structural changes by simulating and visualizing the specific manifestations of these cumulative effects on the three-dimensional structure of the eye and accurately calculating the changes in retinal thickness, corneal curvature and lens morphology.

[0039] Step S500: performing data fusion processing based on historical health characteristic data and behavior-environment cumulative effect data, and using big data analysis and supervised learning algorithms to train a prediction model, and obtaining potential risk data by predicting potential risk factors for children's eye health;

[0040] It can be understood that this step first fuses historical health feature data from multiple sources (including children's physical examination records, treatment records, and family medical history, etc.) with behavior-environment cumulative effect data (such as the duration of use of electronic devices, lighting conditions, etc.). With the support of big data technology, these heterogeneous data can be effectively integrated and standardized through parallel computing and distributed processing to ensure the relevance and consistency of multidimensional data. Next, supervised learning algorithms, such as decision trees or random forest models, are used to train and predict these fused data. The advantage of big data technology is that it can process large amounts of health and behavioral data, improve the accuracy of prediction models, and identify potential health risk factors. In this way, the system can efficiently extract important patterns related to children's eye health from large data sets, reveal potential health risks, and provide a basis for individualized health monitoring.

[0041] Step S600: performing evaluation processing based on the three-dimensional eye feature data and risk data to obtain an evaluation result, wherein the evaluation result includes current monitoring results and future eye health management recommendations.

[0042] These assessment results can help doctors and parents better understand children’s eye health status and take effective prevention and intervention measures to improve children’s eye health.

[0043] It should be noted that step S200 includes step S210, step S220, step S230 and step S240.

[0044] Step S210, performing voxelization processing according to the first information, and obtaining voxelized eye data by converting the eye structure data into three-dimensional voxel data;

[0045] Specifically, a voxel (volumetric pixel) is a small cube in three-dimensional space, which represents a data point in this spatial area. The value of each voxel usually represents a certain characteristic value of the location, such as density, color or intensity. In the eye structure data, voxelization will associate each voxel with specific eye tissue features, such as retinal thickness, corneal curvature, etc. In order to achieve this process, the input two-dimensional image needs to be preprocessed first, including denoising, contrast enhancement, etc. Next, an interpolation algorithm (such as bilinear interpolation and cubic spline interpolation) is used to fill the gaps in the three-dimensional space to ensure that the reconstructed three-dimensional model has high resolution and high accuracy. At this point, the generated three-dimensional voxel data can be used for further analysis and modeling. Through this voxelization process, a detailed three-dimensional eye structure model can be obtained. Voxelized eye data not only provides rich spatial information, but also enables more sophisticated three-dimensional analysis and visualization.

[0046] Step S220, extracting and processing the retinal thickness of the voxelized eye data based on the U-Net segmentation algorithm, automatically segmenting the retinal area and calculating the thickness to obtain retinal thickness data by combining the thickness variation characteristics of the children's retina at different growth stages;

[0047] It can be understood that the U-Net algorithm is a deep learning network that is suitable for medical image segmentation tasks. Its structure is a symmetrical U-shape, including a contraction path to capture background information and a symmetrical expansion path to ensure precise positioning. This structure makes it suitable for the precise segmentation of small targets in medical images. When dealing with the extraction of retinal thickness, U-Net can effectively separate retinal areas from complex eye images, and maintain high-precision segmentation even when the structure of the retina changes at various growth stages. This step first inputs the voxelized three-dimensional eye data into the U-Net model. The model has been trained to recognize different features and boundaries of the retina by learning a large amount of similar medical image data. Specifically, the algorithm analyzes the features of each voxel point and automatically marks the retinal area. Then, by calculating the thickness of these marked areas, an accurate measure of retinal thickness is obtained. As Figure 2 As shown, Figure 2The flowchart of the U-Net algorithm is shown. The U-Net architecture includes an input layer, a descending convolution layer, a bottom layer (bottleneck layer), an ascending convolution layer, and an output layer. Among them, the input layer is the starting point of the network, and the input is voxelized three-dimensional eye data; the descending convolution layer includes descending convolution layer 1 and descending convolution layer 2, and extracts low-level features of the image, such as edges and textures, through convolution operations. After each convolution layer, the spatial dimension of the image is halved through maximum pooling, and the number of feature channels increases; the bottom layer (bottleneck layer) is located at the deepest part of the network and is designed to have the maximum feature extraction capability. Pooling is not performed here, but more convolution operations are used to deepen the learning of features and provide rich feature information for the ascending path. The ascending convolution layer includes ascending convolution layer 1 and ascending convolution layer 2: These layers restore the details and size of the image through upsampling (gradually increasing the spatial dimension of the image) and convolution operations. While upsampling, the feature maps in the corresponding descending path are merged into the ascending path (through jump connections). This is done to retain important spatial information that may be lost in the descending path, thereby improving the accuracy of segmentation. The output layer finally converts the feature map into a final segmentation map through a convolutional layer. This map marks the exact area of ​​the retina and can be used for further analysis such as thickness calculation. This design is intended to capture more abstract high-level features while reducing computational complexity. The technical effects of this method include the ability to provide fast and automated retinal thickness measurements, which is particularly important for monitoring children's vision development. The formula for calculating retinal thickness is:

[0048]

[0049] Where x, y represent the horizontal and vertical coordinates of the central pixel of the retinal image, d(x, y) represents the retinal thickness at the position (x, y), R represents the local neighborhood, i, j represent the horizontal and vertical coordinates of the points in the R region, G is a function used to simulate the changing characteristics of children's retina at different growth stages, w represents the weight assigned to each pixel in the R region by the U-Net model, and F ij represents the eigenvalue at (i, j), F xy represents the eigenvalue at (x,y), and σ represents the standard deviation.

[0050] Step S230, performing corneal curvature calculation processing based on the voxelized eye data, capturing the curvature characteristics of different development stages through an adaptive filtering algorithm and performing surface fitting processing to obtain corneal curvature data;

[0051] It is understood that the adaptive filtering algorithm can dynamically adjust its parameters according to the local characteristics of the data, so that it can better adapt to different areas in the image. For the measurement of corneal curvature, this means that the algorithm can adjust the processing method according to its unique morphology and curvature changes in different parts of the cornea (such as the center and edge areas). After the adaptive filtering process, surface fitting technology is used to create a continuous and smooth surface model. The calculation formula is:

[0052]

[0053] Where x, y represent the horizontal and vertical coordinates of the central pixel of the retinal image, K(x, y) represents the corneal curvature at position (x, y), N represents the local neighborhood, i, j represent the horizontal and vertical coordinates of the points in the N area, W represents the adaptive weight, and Z represents the height function of the eye structure. This calculation method uses differential geometry and image processing techniques to quantitatively describe the curvature of the cornea. Evaluating the local curvature of the cornea through a second-order partial differential equation can provide a detailed view of the corneal morphology, especially in areas with complex or irregular surface changes.

[0054] Step S240: Perform lens morphology analysis based on the voxelized eye data, and extract the main morphological features of the lens to obtain lens morphology data.

[0055] Specifically, image segmentation technology is first used to distinguish the lens from other eye structures. After that, a morphological analysis is performed on the segmented lens area, including its size, shape, and any morphological abnormalities. Through quantitative analysis, the geometric parameters of the lens, such as length, width, thickness, and the complexity of its three-dimensional structure are extracted, and three-dimensional reconstruction is performed using voxelized data, so that the morphology of the lens at different viewing angles can be observed.

[0056] It should be noted that step S300 includes step S310, step S320, step S330 and step S340.

[0057] Step S310: performing data integration processing according to the second information to obtain comprehensive health data;

[0058] Understandably, aggregating health data involves merging fields, creating new data dimensions (such as calculating long-term trends in certain health indicators), and transforming the data to suit subsequent analysis.

[0059] Step S320: performing feature extraction processing based on the comprehensive health data, and obtaining health feature data by identifying key health features;

[0060] Preferably, this step uses statistical analysis and machine learning algorithms to identify features that have a significant impact on health conditions. These features include the patient's biomarkers, historical symptoms, treatment responses, etc.

[0061] Step S330, performing pattern recognition processing according to the health feature data, and obtaining disease pattern data by identifying patterns related to specific eye diseases based on a hidden Markov model;

[0062] It can be understood that a hidden Markov model is a statistical model that assumes that the system can be described by a series of hidden states that have Markov properties (i.e., the probability of the next state depends only on the current state). In the identification of pediatric ophthalmological diseases, each disease can be regarded as a series of state changes, which are observed and identified through the patient's health characteristic data. The process of using hidden Markov models for disease pattern recognition first involves training the model to identify state transition patterns in the data, which correspond to the development stages or characteristics of the disease. For example, in the process of analyzing the development of glaucoma, the state transition can be marked by the progression of the lesion (such as the gradual reduction of the visual field). In addition, the hidden Markov model is able to process time series data, making it particularly suitable for tracking health indicators that change over time, such as vision test results or intraocular pressure data.

[0063] Step S340: Perform trend analysis based on the disease pattern data, identify the potential trend of disease development through trend prediction of the disease pattern, and obtain historical health characteristic data.

[0064] Preferably, this step predicts the development trend of children's eye diseases by applying time series analysis methods, such as autoregressive moving average models, and ultimately generates historical health characteristic data. This analysis is particularly important for children's eye health management because it can help doctors detect signs of disease early, such as worsening myopia, astigmatism, or initial signs of other vision problems. For example, when analyzing children's vision test results, by capturing patterns and trends in vision changes, it is possible to predict the possible development paths of these diseases and take appropriate preventive measures in advance. This trend analysis enhances the ability to predict changes in children's eye health, allowing ophthalmologists to develop personalized monitoring and intervention strategies based on data-driven insights.

[0065] It should be noted that step S400 includes step S410, step S420, step S430 and step S440.

[0066] Step S410: performing time series processing according to the historical health characteristic data and the third information, and obtaining time series data by eliminating noise and trends in the data;

[0067] This step improves the quality of children's eye health monitoring data and enhances the ability to monitor and understand long-term changes in children's eye health.

[0068] Step S420: Perform scenario recognition processing based on the time series data, and obtain behavior-environment data by using a dynamic time warping algorithm to identify and match typical behaviors and environmental scenarios in the time series data;

[0069] It is understandable that the dynamic time warping algorithm is particularly suitable for analyzing sequence data that may be deformed on the time axis, such as sequences with inconsistent time lengths or different rates. In children's eye health monitoring, the dynamic time warping algorithm can effectively compare and analyze the relationship between health indicators (such as intraocular pressure or vision changes) recorded at different time points and environmental factors (such as lighting conditions, frequency of use of electronic products). Through the processing of the dynamic time warping algorithm, specific behaviors (such as study time, outdoor activities) and environmental factors (such as lighting conditions, frequency of use of electronic devices) in the time series data can be associated with eye health conditions (such as vision changes). This helps to determine which specific combinations of behaviors and environments affect children's vision health.

[0070] Step S430, performing quantification processing according to the behavior-environment data, and obtaining cumulative effect data by applying a long short-term memory network to quantify the cumulative damage effects on the eyes under various behavior-environment combination scenarios;

[0071] It is understandable that this step standardizes the data of various behaviors (such as reading, outdoor activities) and environmental factors (such as lighting conditions, frequency of using electronic products) to ensure that they are suitable for input into the long short-term memory network model. Next, the network structure of the long short-term memory network was adjusted according to these data characteristics, including increasing the depth of the network and the number of memory units, so that it can better capture children’s behavior patterns in different environments and their potential impact on vision. The core of the long short-term memory network is its ability to effectively process and remember long-term data dependencies, which is crucial for analyzing how behaviors and environments cumulatively affect children’s vision over time. Through training, the long short-term memory network model can identify the extent to which specific behaviors and environmental factors affect eye health and predict possible health trends in the future. Specifically, the model can identify vision problems that may be caused by reading in low light conditions for a long time, or evaluate how regular outdoor activities can effectively prevent the development of myopia.

[0072] Step S440: Perform data mapping processing according to the cumulative effect data, by reverse modeling the eye structure and mapping the cumulative effect data to specific eye structure changes, to obtain behavior-environment cumulative effect data.

[0073] It is understandable that the core of the cumulative effect data is to understand how various behavioral and environmental factors interact over time to jointly affect children's eye structure and health. It is possible to capture the additive effects between single factors, as well as the interactions between different factors, and even the offsetting effects they may produce. Specifically, the cumulative effect data can show that long-term use of electronic devices indoors (low light environment and close focus) may cause a gradual decline in vision, but this effect may be offset or slowed by regular outdoor activities (providing better lighting and long-distance focusing opportunities). These data not only record the impact of each independent factor, but also reveal their combined effects under different combinations through comprehensive analysis. Further, finite element analysis (FEA) and other modeling techniques are used for this reverse modeling. These techniques can simulate in detail the response of physical structures when subjected to external forces (in this scenario, physiological changes caused by behavioral and environmental factors). By inputting the cumulative effect data into these models, it is possible to simulate and predict in detail how behavioral and environmental factors affect specific structures of the eye over time, such as changes in retinal thickness, corneal curvature, and lens morphology, and predict how specific behavioral-environmental combinations specifically affect the health status of eye structures. By considering the cumulative effects of multiple variables, a more comprehensive and accurate health management tool is provided.

[0074] It should be noted that step S440 includes step S441, step S442, step S443 and step S444.

[0075] Step S441, performing statistical analysis on the cumulative effect data, quantifying the specific impact of each behavior and environmental factor on the eye structure change by using multivariate regression analysis, and obtaining the regression coefficient;

[0076] Through this analysis, the impact of each factor is not only measured individually, but also how they interact with other factors is considered. For example, multivariate regression can identify which specific behavioral patterns or environmental settings have the greatest impact on vision loss, or discover that certain seemingly unrelated combinations of behavioral and environmental factors may lead to unexpected eye health problems. Regression analysis provides specific regression coefficients that quantify the specific contribution of each factor to changes in eye structure, making it possible to accurately understand and predict how behavior and environment affect children's eye health.

[0077] Step S442, performing reverse modeling of the eye structure according to the regression coefficient, simulating the stress and deformation of the eye structure under different external influences through finite element analysis and displaying the specific changes of the eye structure caused by different cumulative effects, and obtaining a reverse model;

[0078] It can be understood that inverse modeling is to infer the factors (behavioral and environmental factors) that cause these changes based on known results (changes in eye structure). Using regression coefficients, the contribution of each factor to changes in eye structure can be quantified, and these data can be used to set the input parameters of the finite element model to simulate in detail the various forces that the eye structure may be subjected to in real life. For example, the specific stress patterns caused by long periods of reading or using electronic products with the head down, and the impact of changes in light during outdoor activities on eye structures such as the cornea and retina. The model converts these factors into stress and deformation data of the eye structure, providing a visual picture of the changes.

[0079] Step S443, performing data mapping processing according to the inverse model, and obtaining a mapping result by associating the cumulative effect data with the changes in the eye structure using a support vector regression algorithm;

[0080] It is understandable that the support vector regression algorithm is suitable for processing data sets with complex nonlinear relationships and can find the best fit line in high-dimensional space. In this embodiment, the support vector regression algorithm minimizes the error between the actual observed value and the predicted value by defining a loss function, while trying to maintain the smoothness of the model and avoid overfitting. Specifically, first, the eye structure change data output by the inverse model is integrated with the cumulative effect data and converted into a format suitable for support vector regression analysis. The data is trained using the support vector regression algorithm, and parameters such as kernel function type, penalty parameter and kernel coefficient are adjusted to adapt to the characteristics of eye health data and optimize prediction accuracy. Through the trained model, the cumulative effect data is mapped to specific eye structure changes, such as specific indicators such as corneal thinning or retinal deformation, to provide quantitative association results. The support vector regression algorithm can provide highly accurate prediction results, which helps to accurately determine how different factors affect the eye structure individually and together.

[0081] Step S444: perform data conversion processing according to the mapping results, convert continuous variables into categorical data, and obtain behavior-environment cumulative effect data.

[0082] Understandably, converting continuous variables into categorical data simplifies the decision-making process and makes eye health management measures easier to implement, especially in situations that require a rapid response.

[0083] It should be noted that step S500 includes step S510, step S520, step S530 and step S540.

[0084] Step S510: performing data standardization processing based on the historical health characteristic data and the behavior-environment cumulative effect data, unifying the dimensions of the two types of data through minimum and maximum scaling, and performing parallel processing on heterogeneous data based on the MapReduce framework to obtain a normalized data set;

[0085] In this process, each data point is rescaled to the specified range so that all data items have the same scale when analyzed and compared. This processing not only improves the consistency of data processing, but also facilitates the application of statistical analysis and machine learning models. Furthermore, multi-source heterogeneous data is processed in parallel based on the MapReduce framework, including:

[0086] Step S511, Map phase: pre-process data from different sources (such as health monitoring data, behavior data, environmental data, etc.), divide them into multiple data segments (sp l its), and assign a mapping task to each data segment. Each mapping task maps the input data into the key-value pair format required for standardized processing, such as health characteristics as keys and numerical data as values. For multi-source heterogeneous data, the Map phase can also unify the data format, remove invalid data or missing data, and ensure consistent data format.

[0087] Step S512, Shuffle phase: Classify and sort the key-value pairs generated in the Map phase so that data with the same health characteristics are aggregated on the same Reducer node. This phase classifies similar data from different sources to ensure that health data, behavioral data, and environmental data from different sources can be efficiently integrated and analyzed in the Reduce phase.

[0088] Step S513, Reduce phase: Perform normalization processing on each Reducer node, and use the minimum and maximum scaling algorithm to perform normalization calculations on health data from different sources. Specifically, the Reducer will perform normalization processing based on the values ​​corresponding to each key (such as health characteristics), ensure that all health data, behavioral data, and environmental data have a unified dimension, and finally output a standardized normalized data set.

[0089] Step S520: Perform modeling processing according to the normalized data set to construct a decision tree model;

[0090] It can be understood that this step uses the normalized data set to build a decision tree model specifically for the children's eye health status monitoring project. The establishment of this model is based on decision tree algorithms, such as CART (Classification and Regression Tree) or C4.5, which can automatically learn and identify key decision paths from data sets. Each decision node represents an evaluation of a specific variable (such as living environment, reading habits, etc.), and the path from the root node to the leaf node reflects a series of decision points until the final health status classification. By calculating information gain or Gini impurity to select the best splitting attribute, the decision tree accurately analyzes how different behavioral and environmental factors collectively affect children's eye health. The main advantage of this model is its high interpretability, which allows doctors and parents to easily understand the key factors affecting children's eye health and formulate preventive or intervention measures accordingly. In addition, the classification results of the decision tree directly support clinical decision-making, help implement personalized health management plans, and improve the pertinence and timeliness of preventive measures.

[0091] Step S530, cross-validation processing is performed based on the decision tree model, by dividing the data set into at least two groups for training and validation and using the K-fold cross-validation method to evaluate the model to obtain the generalization ability and stability to obtain an optimized model;

[0092] It is understood that cross-validation is a statistical analysis method used to evaluate and improve the performance of a model on unseen data. In this process, the entire dataset is split into at least two subsets, usually using the K-fold cross-validation method, in which the dataset is divided into K subgroups, one set is set aside each time as a validation dataset, and the remaining K-1 groups are used to train the model. By repeatedly training and testing on different data subsets, the model can demonstrate its stability and reliability in a variety of data situations, thereby improving its generalization ability.

[0093] Step S540: Perform risk assessment processing on the latest patient data in the historical health characteristic data according to the optimization model, and obtain potential risk data by predicting the risk of myopia development.

[0094] Specifically, the model compares new patient data with historical health feature data and uses trained parameters to assess possible future health problems, such as worsening myopia. Using the classification results of the decision tree model, it is possible to identify which children are at high risk of developing myopia. The model analyzes the impact of multiple factors such as genetic factors, living habits, and learning environment. The optimized model more accurately predicts the specific impact of various factors on children's eye health, especially the risk assessment of myopia development.

[0095] It should be noted that step S600 includes step S610, step S620, step S630 and step S640.

[0096] Step S610: Modeling is performed based on the three-dimensional eye feature data and the risk data, and the genetic factors and lifestyle data in the risk data are integrated with the three-dimensional eye structure data using a machine learning algorithm to construct a digital twin model that can dynamically simulate eye pathological changes;

[0097] It is understandable that this step uses machine learning technology to build a digital twin model, which integrates three-dimensional eye structure data and risk data including genetic factors and lifestyle habits to dynamically simulate and predict pathological changes in children's eyes. The model can accurately predict pathological conditions such as myopia progression and corneal thickness changes by combining detailed analysis of the individual's genetic background, environmental exposure and behavioral habits with sophisticated three-dimensional eye imaging. The model uses algorithms such as deep learning to identify and learn complex patterns hidden in health data, which can not only see the current state of eye lesions, but also foresee future trends.

[0098] Step S620: Perform simulation processing according to the digital twin model to obtain a pathology development prediction model by simulating the potential changes of different risk factors on the eye structure and simulating the long-term effects of different lifestyles and treatment plans on eye health;

[0099] Specifically, this step is simulated through a digital twin model. These simulations examine the possible changes in children's eye structures caused by various risk factors, as well as the long-term effects of different lifestyles and treatments on eye health. This model integrates data such as genetic factors, behavioral habits, and environmental influences, and uses machine learning technology to predict future pathological developments, such as the progression of myopia and changes in corneal diseases, providing a comprehensive method for assessing and predicting children's eye health. The advantage of the digital twin model is that it can simulate real-world medical scenarios and foresee the effects of different medical interventions. For example, by simulating different vision correction strategies, it is possible to predict the specific effects of these measures on children's vision and optimize treatment plans. This simulation is not only based on existing medical data, but can also be updated in real time, and the prediction results are continuously optimized as new data is added.

[0100] Step S630, generating a health risk assessment according to the pathology development prediction model, and using the SHAP value to explain the decision-making process of the prediction model to obtain a health assessment report;

[0101] It can be understood that this step uses the pathological development prediction model to conduct a risk assessment of children's eye health, and uses the SHAP value to explain the model's prediction decision process in detail. Specifically, first, the pathological development prediction model combines multi-source data, including personal genetic information, lifestyle habits, and environmental factors, which are analyzed by machine learning methods to identify and predict possible eye health problems. The model predicts the possibility of pathological conditions such as myopia exacerbation by learning the relationship between these data. Subsequently, the SHAP value is used to explain the model decision. The SHAP value provides an intuitive way to understand the operating mechanism of the model by quantifying the contribution of each feature to the prediction results. This explanatory ability not only improves the transparency of the model, but also enables doctors and parents to better understand the reasons for the prediction output. In this way, the potential effects of various health interventions, such as the long-term health effects of changing lifestyle habits or medical interventions, can be displayed and evaluated in detail.

[0102] Step S640: Perform language conversion processing according to the health assessment report, and obtain the assessment result by generating a coherent natural language text from the structured data explained by the SHAP value using natural language generation technology.

[0103] This process involves extracting key information from the data and expressing it in easy-to-understand language so that non-professionals can easily understand the model's decision-making process and results. Specifically, the structured SHAP value data was first parsed through natural language generation technology, which detailed how various variables affect the model's prediction results. Natural language generation technology can automatically generate text descriptions from these explanations, such as "Based on the patient's genetic background and current lifestyle habits, the model predicts that his risk of myopia increases." This technology not only saves a lot of time in writing reports, but also ensures the accuracy and consistency of information. This step significantly improves the acceptability and readability of the report, allowing doctors and patients to better understand and trust the output of the model. In this way, complex data analysis results are transformed into concise and clear health advice, greatly promoting the practical application and implementation of medical advice.

[0104] Embodiment 2:

[0105] This embodiment provides an eye health status monitoring and evaluation device based on big data, the device comprising:

[0106] an acquisition module, configured to acquire first information, second information, and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, the electronic health records include eye examination records, treatment records, and family medical history, and the third information includes daily behavior data and environmental data;

[0107] A reconstruction module, configured to perform three-dimensional reconstruction processing according to the first information, and obtain three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data;

[0108] an identification module, configured to perform pattern recognition processing according to the second information, identify patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records, and family medical history in the electronic health record, and obtain historical health feature data;

[0109] An analysis module is used to perform time series analysis based on historical health characteristic data and third-party information, and obtain behavior-environment cumulative effect data by quantitatively calculating the cumulative effects of eye damage under various behavior-environment combination scenarios and mapping them to the transformation of the three-dimensional structure of the eye;

[0110] The fusion module is used to perform data fusion processing based on historical health characteristic data and behavior-environment cumulative effect data, and use big data analysis and supervised learning algorithms to train the prediction model to obtain potential risk data by predicting potential risk factors for children's eye health;

[0111] The evaluation module is used to evaluate and process the three-dimensional eye feature data and risk data to obtain evaluation results, which include current monitoring results and future eye health management recommendations.

[0112] In a specific embodiment of the present disclosure, the reconstruction module includes:

[0113] A first processing unit, configured to perform voxelization processing according to the first information, and obtain voxelized eye data by converting the eye structure data into three-dimensional voxel data;

[0114] The first extraction unit extracts and processes the retinal thickness of voxelized eye data based on the U-Net segmentation algorithm, automatically segments the retinal area and calculates the thickness to obtain the retinal thickness data by combining the thickness variation characteristics of the children's retina at different growth stages;

[0115] A first calculation unit is used to calculate and process corneal curvature according to voxelized eye data, capture curvature characteristics of different development stages through an adaptive filtering algorithm, and perform surface fitting processing to obtain corneal curvature data;

[0116] The second extraction unit is used to perform lens morphology analysis and processing based on the voxelized eye data, and extract the main morphological features of the lens to obtain lens morphology data.

[0117] In a specific embodiment of the present disclosure, the identification module includes:

[0118] A first integration unit, configured to perform data integration processing according to the second information to obtain comprehensive health data;

[0119] A third extraction unit is used to perform feature extraction processing based on the comprehensive health data, and obtain health feature data by identifying key health features;

[0120] A first recognition unit is used for performing pattern recognition processing according to the health characteristic data, and obtaining disease pattern data by identifying patterns related to specific eye diseases based on a hidden Markov model;

[0121] The second identification unit is used to perform trend analysis and processing based on the disease pattern data, and to identify the potential trend of disease development by predicting the trend of the disease pattern to obtain historical health characteristic data.

[0122] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0124] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for monitoring and evaluating eye health status based on big data, characterized in that: include: Acquire first information, second information and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, the electronic health records include eye examination records, treatment records and family medical history, and the third information includes daily behavior data and environmental data; Performing three-dimensional reconstruction processing according to the first information, and constructing three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data; performing pattern recognition processing based on the second information, identifying patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records, and family medical history in the electronic health record, and obtaining historical health feature data; Performing a time series analysis based on the historical health characteristic data and the third information, by quantitatively calculating the cumulative effects of various behavior-environment combinations on eye damage and mapping them to the transformation of the three-dimensional structure of the eye, to obtain behavior-environment cumulative effect data; Performing data fusion processing based on the historical health characteristic data and the behavior-environment cumulative effect data, and training a prediction model using big data analysis and supervised learning algorithms to obtain potential risk data by predicting potential risk factors for children's eye health; An evaluation result is obtained by performing evaluation processing based on the three-dimensional eye feature data and the risk data, and the evaluation result includes current monitoring results and future eye health management recommendations.

2. The eye health status monitoring and evaluation method based on big data according to claim 1 is characterized in that , performing three-dimensional reconstruction processing according to the first information, and obtaining three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, including: Performing voxel processing according to the first information, and obtaining voxelized eye data by converting the eye structure data into three-dimensional voxel data; Extracting and processing the retinal thickness of the voxelized eye data based on the U-Net segmentation algorithm, automatically segmenting the retinal area and calculating the thickness to obtain retinal thickness data by combining the thickness variation characteristics of the children's retina at different growth stages; Performing corneal curvature calculation processing based on the voxelized eye data, capturing curvature characteristics of different developmental stages through an adaptive filtering algorithm and performing surface fitting processing to obtain corneal curvature data; Lens morphology analysis is performed based on the voxelized eye data, and main morphological features of the lens are extracted to obtain lens morphology data.

3. The eye health status monitoring and evaluation method based on big data according to claim 1 is characterized in that , performing pattern recognition processing based on the second information, identifying patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records and family medical history in the electronic health record, and obtaining historical health feature data, including: Performing data integration processing according to the second information to obtain comprehensive health data; Performing feature extraction processing according to the comprehensive health data, and obtaining health feature data by identifying key health features; Performing pattern recognition processing on the health feature data, and obtaining disease pattern data by identifying patterns related to specific eye diseases based on a hidden Markov model; Trend analysis is performed on the disease pattern data, and historical health characteristic data is obtained by identifying the potential trend of disease development through trend prediction of the disease pattern.

4. The eye health status monitoring and evaluation method based on big data according to claim 1 is characterized in that , performing time series analysis based on the historical health feature data and the third information, by quantitatively calculating the cumulative effects of various behavior-environment combinations on eye damage and mapping them to the transformation of the three-dimensional structure of the eye, obtaining behavior-environment cumulative effect data, including: Performing time series processing according to the historical health characteristic data and the third information to obtain time series data by eliminating noise and trends in the data; Performing scenario recognition processing according to the time series data, and obtaining behavior-environment data by using a dynamic time warping algorithm to identify and match typical behaviors and environmental scenarios in the time series data; Quantitative processing is performed according to the behavior-environment data, and cumulative effect data is obtained by applying a long short-term memory network to quantify the cumulative damage effects on the eyes under various behavior-environment combination scenarios; Data mapping processing is performed according to the cumulative effect data, and the behavior-environment cumulative effect data is obtained by reverse modeling the eye structure and mapping the cumulative effect data to specific eye structure changes.

5. The eye health status monitoring and evaluation method based on big data according to claim 4 is characterized in that , performing data mapping processing according to the cumulative effect data, by reverse modeling the eye structure and mapping the cumulative effect data to specific eye structure changes, obtaining behavior-environment cumulative effect data, including: Statistical analysis is performed on the cumulative effect data to quantify the specific effects of each behavior and environmental factor on the changes in eye structure by using multivariate regression analysis to obtain regression coefficients; Performing reverse modeling of the eye structure according to the regression coefficient, simulating the stress and deformation of the eye structure under different external influences through finite element analysis and showing the specific changes of the eye structure caused by different cumulative effects, thereby obtaining a reverse model; Performing data mapping processing according to the inverse model, and obtaining a mapping result by associating the cumulative effect data with changes in eye structure using a support vector regression algorithm; Data conversion processing is performed according to the mapping results, and the behavior-environment cumulative effect data are obtained by converting continuous variables into categorical data.

6. The eye health status monitoring and evaluation method based on big data according to claim 1 is characterized in that , perform data fusion processing based on the historical health characteristic data and the behavior-environment cumulative effect data, and use big data analysis and supervised learning algorithms to train the prediction model, and obtain potential risk data by predicting potential risk factors for children's eye health, including: Performing data standardization processing on the historical health characteristic data and the behavior-environment cumulative effect data, unifying the dimensions of the two types of data through minimum and maximum scaling, and performing parallel processing on heterogeneous data based on a MapReduce framework to obtain a normalized data set; Performing modeling processing according to the normalized data set to construct a decision tree model; Performing cross-validation processing based on the decision tree model, splitting the data set into at least two groups for training and validation, and using a K-fold cross-validation method to evaluate the model to obtain generalization ability and stability to obtain an optimized model; The latest patient data in the historical health characteristic data is subjected to risk assessment processing according to the optimization model, and potential risk data is obtained by predicting the risk of myopia development.

7. The eye health status monitoring and evaluation method based on big data according to claim 1 is characterized in that , performing evaluation processing according to the three-dimensional eye feature data and the risk data to obtain an evaluation result, including: Modeling is performed based on the three-dimensional eye feature data and the risk data, and the genetic factors and lifestyle data in the risk data are integrated with the three-dimensional eye structure data using a machine learning algorithm to construct a digital twin model that can dynamically simulate eye pathological changes; Performing simulation processing according to the digital twin model, by simulating the potential changes of eye structure caused by different risk factors and simulating the long-term effects of different lifestyles and treatment plans on eye health, a pathology development prediction model is obtained; Generate a health risk assessment based on the pathology development prediction model, and use SHAP values ​​to explain the decision-making process of the prediction model to obtain a health assessment report; Language conversion processing is performed according to the health assessment report, and a coherent natural language text is generated from the structured data explained by the SHAP value by using natural language generation technology to obtain an assessment result.

8. A device for monitoring and evaluating eye health based on big data, characterized in that: include: an acquisition module, configured to acquire first information, second information, and third information, wherein the first information includes three-dimensional eye structure data of the child to be monitored, the second information includes electronic health records, the electronic health records include eye examination records, treatment records, and family medical history, and the third information includes daily behavior data and environmental data; a reconstruction module, configured to perform three-dimensional reconstruction processing according to the first information, and obtain three-dimensional eye feature data by identifying retinal thickness, corneal curvature and lens morphology, wherein the three-dimensional eye feature data includes retinal thickness data, corneal curvature data and lens morphology data; an identification module, configured to perform pattern recognition processing according to the second information, identify patterns and trends related to specific eye diseases by performing correlation analysis on historical eye examination records, treatment records and family medical history in the electronic health record, and obtain historical health feature data; An analysis module, configured to perform time series analysis based on the historical health feature data and the third information, and obtain behavior-environment cumulative effect data by quantitatively calculating the cumulative effects of various behavior-environment combination scenarios on eye damage and mapping them to the transformation of the three-dimensional structure of the eye; A fusion module, used to perform data fusion processing based on the historical health characteristic data and the behavior-environment cumulative effect data, and use big data analysis and supervised learning algorithms to train a prediction model to obtain potential risk data by predicting potential risk factors for children's eye health; An evaluation module is used to perform evaluation processing based on the three-dimensional eye feature data and the risk data to obtain an evaluation result, wherein the evaluation result includes current monitoring results and future eye health management recommendations.

9. The eye health status monitoring and evaluation device based on big data according to claim 8 is characterized in that , the reconstruction module includes: A first processing unit, configured to perform voxelization processing according to the first information, and obtain voxelized eye data by converting the eye structure data into three-dimensional voxel data; A first extraction unit extracts and processes the retinal thickness of the voxelized eye data based on a U-Net segmentation algorithm, automatically segments the retinal area and calculates the thickness to obtain retinal thickness data by combining the thickness variation characteristics of the children's retina at different growth stages; A first calculation unit is used to calculate and process corneal curvature according to the voxelized eye data, capture curvature characteristics of different development stages through an adaptive filtering algorithm, and perform surface fitting processing to obtain corneal curvature data; The second extraction unit is used to perform lens morphology analysis processing according to the voxelized eye data, and extract the main morphological features of the lens to obtain lens morphology data.

10. The eye health status monitoring and evaluation device based on big data according to claim 8, characterized in that , the identification module includes: A first integration unit, configured to perform data integration processing according to the second information to obtain comprehensive health data; A third extraction unit is used to perform feature extraction processing according to the comprehensive health data, and obtain health feature data by identifying key health features; A first recognition unit is used to perform pattern recognition processing according to the health feature data, and obtain disease pattern data by identifying patterns related to specific eye diseases based on a hidden Markov model; The second identification unit is used to perform trend analysis processing based on the disease pattern data, and identify the potential trend of disease development by trend prediction on the disease pattern to obtain historical health characteristic data.

Images