Ophthalmology health management method and system based on artificial intelligence

Abstract

The invention discloses an ophthalmology health management method and system based on artificial intelligence, and relates to the technical field of artificial intelligence and medical health crossing, and the method comprises the following components: S1, a data collection step, S2, a data fusion and analysis step, S3, a risk assessment step, and S4, a prevention suggestion generation step. By collecting the gene data and the living environment data of the user and performing deep fusion and analysis, an accurate eye disease risk assessment model can be established for the user, based on the model, the system can quantify the risk level of the user suffering from various eye diseases and provide personalized prevention suggestions for the user, and the user experience is improved. Compared with the traditional one-step health management, the personalized health management scheme better meets the actual requirements of the user, and is beneficial to improving the eye health level of the user.

Description

Technical Field

[0001] The present invention relates to the cross - technical field of artificial intelligence and medical health, and specifically to an ophthalmic health management method and system based on artificial intelligence. Background Art

[0002] With the continuous development of artificial intelligence technology and the digital transformation of the medical health field, ophthalmic health management methods and systems based on artificial intelligence have gradually become a research hotspot. As an important part of human health, early prevention and management of ophthalmic health are crucial for improving people's quality of life.

[0003] In traditional ophthalmic health management, although doctors also consider patients' genetic factors and living environment factors, there is often a lack of systematicness and precision. On the one hand, the collection and processing of genetic data are relatively complex and require genetic testing and analysis techniques, and traditional methods often have difficulty obtaining and analyzing these data comprehensively and accurately. On the other hand, living environment factors such as air pollution index, ultraviolet intensity, and indoor lighting conditions have multiple - faceted impacts on eye health, but traditional methods often can only perform single and static analyses and cannot deeply explore the potential relationships and interactions between these factors. Therefore, traditional technologies have obvious deficiencies in the comprehensive analysis of genetic and environmental factors and are difficult to provide accurate and personalized health management solutions for users.

[0004] In summary, traditional technologies have many limitations and cannot meet the modern people's needs for personalized and precise health management. Therefore, it is particularly important to develop an ophthalmic health management method and system based on artificial intelligence. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the existing technology and provide an ophthalmic health management method and system based on artificial intelligence. It can achieve a comprehensive and in - depth analysis and integration of users' genetic data and living environment data by introducing advanced algorithms and technical means, establish an accurate risk assessment model for eye diseases for users, and provide personalized prevention suggestions. Such an intelligent health management solution not only helps to improve users' eye health level but also promotes innovation and development in the field of ophthalmic health management.

[0006] To solve the above - mentioned technical problems, the present invention provides the following technical solution: An ophthalmic health management method based on artificial intelligence, and the specific steps of this method are as follows:

[0007] S1. Data collection step: Collect users' genetic data, and use environmental monitoring devices and the way of users' independent upload to collect living environment data, including air pollution index, ultraviolet intensity, and indoor lighting conditions;

[0008] S2, data fusion and analysis step: using fusion analysis algorithm, the collected genetic data and living environment data are fused and analyzed;

[0009] S3, risk assessment step: Based on the data analysis results, establish an eye disease risk assessment model to quantify the user's risk level of various eye diseases according to different combinations of genes and environmental factors;

[0010] S4. Prevention suggestion generation step: Provide personalized prevention suggestions to users based on risk assessment results.

[0011] Furthermore, in the data collection step, the genetic data is collected through cooperation with genetic testing institutions, and the acquired genetic data is standardized and converted into a unified format to facilitate subsequent analysis. The collection of living environment data is also realized through the user's mobile phone APP. The user can record the special conditions of the environment in real time on the APP, and at the same time connect with the data of the environmental protection department to obtain macro air pollution index data to ensure the comprehensiveness and accuracy of the environmental data. During the data collection process, a data quality monitoring mechanism is set up to mark and verify abnormal data to ensure that the collected data is true and reliable.

[0012] Furthermore, in the data fusion and analysis step, a fusion analysis algorithm is used to define the gene data vector as G = (g1, g2, ..., g n ), where g i Represents the data of the ith gene locus, and the living environment data vector is E = (e1, e2, ..., e m ), where e j Represents the jth environmental factor data and constructs the fusion matrix M, M ij =α ij g i +β ij e j , where α ij and β ij is the weight coefficient. The determination of the weight coefficient is based on a large amount of experimental data and expert experience. The relative importance between different gene loci and environmental factors is determined by the hierarchical analysis method to obtain the weight coefficient. The self-attention mechanism in deep learning is used to extract features from the fusion matrix M to explore the potential relationship between genes and environmental factors. The formula of the self-attention mechanism is: Where Q, K, and V are query vectors, key vectors, and value vectors respectively. k is the dimension of the key vector. In this way, deep integration and analysis of genetic data and living environment data can be achieved.

[0013] Furthermore, in the risk assessment step, the eye disease risk assessment model established is based on a method combining Bayesian network and deep learning. First, the Bayesian network is used to construct a causal relationship diagram between genes, environmental factors and eye diseases to determine the impact path of different factors on the occurrence of the disease. Then, the convolutional neural network in deep learning is combined to perform feature learning on the fused data. The convolutional layer calculation formula of CNN is: in represents the neuron value of the i-th row and j-th column of the l-th layer, is the convolution kernel weight, b l As a bias term, through the prior knowledge of the Bayesian network and the powerful feature learning ability of CNN, the risk level of users suffering from various eye diseases is accurately quantified, the risk level is divided into three levels: low, medium and high, and the corresponding risk probability value is given for each level.

[0014] Furthermore, in the prevention suggestion generation step, a prevention suggestion rule base is established by combining rule reasoning and case reasoning. When the risk assessment result is high risk and the ultraviolet intensity is high, the corresponding rule in the rule base is to recommend users to wear ultraviolet protection glasses and minimize outdoor activity time. At the same time, a case base is constructed to store successful prevention cases under similar risk assessment results in the past. When prevention suggestions are generated, preliminary suggestions are first generated according to the rule base, and then similar cases are retrieved in the case base to supplement and optimize the preliminary suggestions. If a case is retrieved in which a high-risk user's eye condition improved after adjusting the indoor light intensity and performing eye massage regularly, the eye massage suggestion will also be included in the prevention suggestion for the current user, thereby providing users with more comprehensive and personalized prevention suggestions.

[0015] Furthermore, the method also includes data updating and dynamic evaluation steps, which regularly updates the user's genetic data and living environment data. When new data is collected, data fusion and analysis, risk evaluation and prevention suggestions are generated again. The user's eye health risk is dynamically evaluated using a time series analysis algorithm. The risk evaluation value is set to R t , through the time series model R t-i +∈ t ,in is the autoregressive coefficient, p is the autoregressive order, ∈ t It is white noise. According to the results of time series analysis, the user's prevention recommendations are adjusted in time to ensure the effectiveness and timeliness of health management.

[0016] Further, before the data fusion and analysis step, there is also a data preprocessing step to clean the collected gene data and living environment data, remove duplicate data, error data, and missing values. For missing values, the multiple imputation method is used for processing. According to the correlation and distribution characteristics of the data, multiple imputed values are generated, and then the subsequent analysis is carried out by integrating multiple imputed values. The cleaned data is normalized to unify data of different magnitudes to the same scale range for the convenience of subsequent algorithm processing. The gene data normalization formula is:

[0017]

[0018] where is the data value of the i-th gene locus after normalization, g i represents the data value of the i-th gene locus in the original gene data, min(g) represents the minimum value in the entire gene data set, and max(g) represents the maximum value in the entire gene data set;

[0019] The living environment data normalization formula is:

[0020]

[0021] where is the data value of the j-th environmental factor after normalization, e j represents the data value of the j-th environmental factor in the original living environment data, min(e) is the minimum value in the entire living environment data set, and max(e) represents the maximum value in the entire living environment data set. Through data preprocessing, the data quality is improved, providing a reliable data basis for subsequent analysis.

[0022] Further, in the risk assessment step, the family medical history data of the user is also considered. The family medical history data is converted into a vector form F = (f1, f2, …, f k ), where f i represents the incidence of different eye diseases in the family. In the risk assessment model, interaction terms of the family medical history data with the gene data and the living environment data are added, and analysis is carried out through a logistic regression model. The logistic regression model formula is:

[0023]

[0024] where P(Y = 1|X) represents the probability of disease, X is the feature vector, including gene data, environmental data, and family medical history data, β i , γ j , δ l are regression coefficients. By considering the family medical history data, the accuracy of risk assessment is further improved.

[0025] On the other hand, an artificial intelligence-based ophthalmic health management system, characterized in that the system includes a data collection module, a data fusion and analysis module, a risk assessment module, and a prevention recommendation generation module:

[0026] The data collection module: collects the user's genetic data, and uses environmental monitoring devices and the user's independent upload method to collect living environment data, including air pollution index, ultraviolet intensity, and indoor lighting conditions;

[0027] The data fusion and analysis module: performs fusion analysis on the collected genetic data and living environment data through a fusion analysis algorithm;

[0028] The risk assessment module: based on the data analysis results, establishes an ophthalmic disease risk assessment model, and quantifies the risk levels of users suffering from various ophthalmic diseases according to different combinations of genes and environmental factors;

[0029] The prevention recommendation generation module: provides personalized prevention recommendations for users according to the risk assessment results.

[0030] Compared with the prior art, the artificial intelligence-based ophthalmic health management method and system have the following beneficial effects:

[0031] First, by collecting the user's genetic data and living environment data, and performing in-depth fusion and analysis, the invention can establish an accurate ophthalmic disease risk assessment model for users. Based on this model, the system can quantify the risk levels of users suffering from various ophthalmic diseases, and provide personalized prevention recommendations for users. This personalized health management plan is more in line with the actual needs of users than the traditional one-size-fits-all health management, and helps to improve the user's ophthalmic health level.

[0032] Second, the invention also has a data update and dynamic assessment function. It can regularly update the user's genetic data and living environment data. When new data is collected, the system will re-perform data fusion and analysis, risk assessment, and prevention recommendation generation. Using time series analysis algorithms, the system can dynamically assess the user's ophthalmic health risks, and adjust the user's prevention recommendations in a timely manner according to the assessment results. This dynamic assessment and timely adjustment mechanism ensures the effectiveness and timeliness of health management, and helps users to discover and respond to potential ophthalmic health risks in a timely manner.

[0033] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0035] Figure 1 It is a flow operation diagram of an ophthalmic health management method based on artificial intelligence;

[0036] Figure 2 It is a flow operation diagram of an ophthalmic health management system based on artificial intelligence. Detailed implementation manners

[0037] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following will, in combination with the accompanying drawings and preferred embodiments, detail the specific implementation manners, structures, features, and effects of the present invention as follows.

[0038] Embodiment 1

[0039] Xiaowang cooperated with a gene testing institution to obtain his own gene data and performed standardized processing. Usually, he independently uploaded living environment data through a mobile APP. If he found that the office lights were flickering (special situation recorded), at the same time, the APP obtained the air pollution index by connecting with the environmental protection department. When collecting data, the system, according to the quality monitoring mechanism, marked and verified abnormal data. For example, if the ultraviolet intensity data reported by the APP far exceeded the normal range on the same day and in the same place, the system marked the data. After Xiaowang confirmed that it was caused by a malfunction of the APP sensor resulting in incorrect data, it was corrected.

[0040] There are some missing values and incorrect values in the collected data. For the missing values in the gene data, the multiple imputation method is used. Multiple imputation values are generated based on data correlation and distribution characteristics and then comprehensively processed. Then, the gene data and living environment data are normalized. The gene data is normalized according to the formula for normalization, and the living environment data is normalized according to the formula for normalization, so that the data is on the same scale for subsequent analysis.

[0041] Define the gene data vector G = (g1, g2,..., g n ), the living environment data vector E = (e1, e2,..., e m ), and construct the fusion matrix M, where M ij = α ij g i + β ij e j, using the formula of the deep learning self-attention mechanism Extract features from the fusion matrix M to explore the potential relationships between genes and environmental factors. For example, it is found that a certain gene locus g of Xiao Wang x Combined with long-term high-intensity indoor lighting e y May affect eye health.

[0042] Construct an eye disease risk assessment model based on Bayesian network and deep learning. First, use the Bayesian network to construct a causal relationship graph of genes, environmental factors and eye diseases to determine the influence path, and then combine it with a convolutional neural network. The calculation formula of its convolutional layer is Perform feature learning on the fusion data. Considering that there is a history of glaucoma in Xiao Wang's family, convert the family history data into a vector F=(f1, f2,..., f k ), add an interaction term in the risk assessment model, and analyze through the logistic regression model formula It is evaluated that the risk level of Xiao Wang suffering from glaucoma is medium, and the risk probability is 40%.

[0043] The system uses a method that combines rule-based reasoning and case-based reasoning to initially generate suggestions in the rule base, such as "Reduce the continuous eye use time and rest for 10 minutes every hour", and then retrieve the case base to find cases that are similar to Xiao Wang's risk assessment results and have been successfully prevented, and supplement the suggestions with "Use anti-blue light glasses and perform eye massages regularly".

[0044] The system regularly updates Xiao Wang's gene data and living environment data, such as the lighting and air quality data of the new environment after Xiao Wang changes his office location. Using the time series analysis algorithm, through the formula Dynamically evaluate the eye health risk of Xiao Wang. If it is evaluated that Xiao Wang has overused his eyes due to overtime recently and the risk value has increased, adjust the preventive suggestions in time and add the content of "Increase the outdoor activity time".

[0045] Example 2

[0046] Li, who works in a first-tier city, participated in an artificial intelligence-based ophthalmic health management project in order to pay attention to his eye health. He cooperated with a gene testing agency to provide saliva samples to obtain gene data and perform standardized processing. Li independently uploaded living environment data through a mobile APP. For example, when in the office, he recorded the situation of dim indoor lighting and long-term use of electronic devices. When going out, he used the APP to record the relatively high air pollution index and strong ultraviolet rays on that day. The APP is also connected to the data of the environmental protection department to obtain macro air pollution index data. At the same time, the data quality monitoring mechanism monitors the collected data, marks and verifies abnormal data.

[0047] The collected data enters the data fusion and analysis module. A fusion matrix is constructed from the gene data vector and the living environment data vector. The self-attention mechanism in deep learning is used to extract features from the fusion matrix. For example, analysis reveals that the combination of specific gene locus data of Xiao Li and factors such as long-term exposure to air pollution and poor indoor lighting may have an impact on eye health.

[0048] The eye disease risk assessment model is based on a method combining Bayesian network and deep learning. The Bayesian network constructs a causal relationship graph among genes, environmental factors, and eye diseases. The convolutional neural network conducts feature learning on the fusion data. Considering that there is a history of glaucoma in Xiao Li's family, the family history data is converted into vector form and incorporated into the risk assessment model. Through analysis using a logistic regression model, it is finally evaluated that the risk level of Xiao Li suffering from glaucoma is medium, and the risk probability is 40%.

[0049] According to the risk assessment results, the prevention advice generation module uses a method combining rule-based reasoning and case-based reasoning. The rule base gives preliminary advice, such as having regular eye examinations and eating more foods rich in vitamin A. The case base retrieves successful prevention cases similar to Xiao Li's situation, and supplements the advice that Xiao Li should use eye protection light sources in the office and wear anti-ultraviolet glasses when going out.

[0050] Every three months, Xiao Li updates the gene data and the living environment data. After the new data is collected, data fusion and analysis, risk assessment, and prevention advice generation are carried out again. The time series analysis algorithm dynamically assesses Xiao Li's eye health risk. If it is found that Xiao Li has high work pressure, frequent overtime, and an increased time of using electronic devices recently, and the risk assessment value rises, the prevention advice is adjusted in a timely manner. For example, it is recommended that Xiao Li increase the eye relaxation time and use artificial tears to relieve eye fatigue.

[0051] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications within the scope of the technical solution of the present invention. These equivalent embodiments with the same changes and effects, as long as they do not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments according to the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. An artificial intelligence-based ophthalmic health management method, characterized in that: The specific steps of this method are: S1. Data collection steps: Collect users' genetic data, use environmental monitoring equipment and user-uploaded methods to collect living environment data, including air pollution index, ultraviolet intensity, and indoor lighting conditions; S2, data fusion and analysis step: using fusion analysis algorithm, the collected genetic data and living environment data are fused and analyzed; S3, risk assessment step: Based on the data analysis results, establish an eye disease risk assessment model to quantify the user's risk level of various eye diseases according to different combinations of genes and environmental factors; S4. Prevention suggestion generation step: Provide personalized prevention suggestions to users based on risk assessment results.

2. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: In the data collection step, the genetic data is collected through cooperation with genetic testing institutions, and the acquired genetic data is standardized and converted into a unified format. The collection of living environment data is also realized through the user's mobile phone APP. The user can record the special conditions of the environment in real time on the APP, and at the same time connect with the data of the environmental protection department to obtain macro air pollution index data. During the data collection process, a data quality monitoring mechanism is set up to mark and verify abnormal data to ensure that the collected data is true and reliable.

3. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: In the data fusion and analysis step, a fusion analysis algorithm is used to define the gene data vector as G = (g1, g2, ..., g n ), where g i Represents the data of the ith gene locus, and the living environment data vector is E = (e1, e2, ..., e m ), where e j Represents the jth environmental factor data and constructs the fusion matrix M, M ij =α ij g i +β ij e j , where α ij and β ij is the weight coefficient, and the self-attention mechanism in deep learning is used to extract features from the fusion matrix M to explore the potential relationship between genes and environmental factors. The formula of the self-attention mechanism is: Where Q, K, and V are query vectors, key vectors, and value vectors respectively. k is the dimension of the key vector.

4. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: In the risk assessment step, the eye disease risk assessment model established is based on a method combining Bayesian network and deep learning. First, the Bayesian network is used to construct a causal relationship diagram between genes, environmental factors and eye diseases to determine the impact path of different factors on the occurrence of the disease. Then, the convolutional neural network in deep learning is combined to perform feature learning on the fused data. The convolutional layer calculation formula of CNN is: in represents the neuron value of the i-th row and j-th column of the l-th layer, is the convolution kernel weight, b l As a bias term, through the prior knowledge of the Bayesian network and the powerful feature learning ability of CNN, the risk level of users suffering from various eye diseases is accurately quantified, the risk level is divided into three levels: low, medium and high, and the corresponding risk probability value is given for each level.

5. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: In the prevention suggestion generation step, a prevention suggestion rule base is established by combining rule reasoning and case reasoning, and a case base is constructed at the same time to store successful prevention cases under similar risk assessment results in the past. When generating prevention suggestions, preliminary suggestions are first generated according to the rule base, and then similar cases are retrieved in the case base to supplement and optimize the preliminary suggestions.

6. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: The method also includes data updating and dynamic evaluation steps, which regularly update the user's genetic data and living environment data. When new data is collected, data fusion and analysis, risk evaluation and prevention suggestions are generated again. The user's eye health risk is dynamically evaluated using a time series analysis algorithm. The risk evaluation value is set to R t , through the time series model in is the autoregressive coefficient, p is the autoregressive order, ∈ t It is white noise. According to the time series analysis results, the user's prevention suggestions are adjusted in time.

7. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: Before the data fusion and analysis steps, the data preprocessing step is also included, in which the collected genetic data and living environment data are cleaned to remove duplicate data, erroneous data and missing values. For missing values, a multiple filling method is used to process them. According to the correlation and distribution characteristics of the data, multiple filling values ​​are generated, and then the multiple filling values ​​are combined for subsequent analysis. The cleaned data is normalized to unify data of different magnitudes into the same scale range to facilitate the processing of subsequent algorithms. The genetic data normalization formula is: in is the data value of the ith gene locus after normalization, g i represents the data value of the ith gene locus in the original gene data, min(g) represents the minimum value in the entire gene data set, and max(g) represents the maximum value in the entire gene data set; The normalization formula for living environment data is: in is the data value of the jth environmental factor after normalization, e j Represents the data value of the jth environmental factor in the original living environment data, min(e) is the minimum value in the entire living environment data set, and max(e) represents the maximum value in the entire living environment data set.

8. The method for ophthalmic health management based on artificial intelligence according to claim 1, characterized in that: In the risk assessment step, the user's family medical history data is also considered, and the family medical history data is converted into a vector form F = (f1, f2, ..., f k ), where f i Indicates the incidence of different eye diseases in the family. In the risk assessment model, the interaction terms of family medical history data, genetic data, and living environment data are added, and the analysis is performed through a logistic regression model. The formula of the logistic regression model is: Where P(Y=1|X) represents the probability of disease, X is the feature vector, including genetic data, environmental data and family history data, β i , γ j , δ l is the regression coefficient, and the accuracy of risk assessment can be further improved by taking family history data into account.

9. An artificial intelligence-based ophthalmic health management system, characterized in that: The system includes data acquisition module, data fusion and analysis module, risk assessment module and prevention suggestion generation module: The data collection module collects the user's genetic data and uses environmental monitoring equipment and user-uploaded data to collect living environment data, including air pollution index, ultraviolet intensity, and indoor lighting conditions; The data fusion and analysis module: performs fusion analysis on the collected gene data and living environment data through fusion analysis algorithm; The risk assessment module: establishes an eye disease risk assessment model based on the data analysis results, and quantifies the risk level of users suffering from various eye diseases according to the combination of different genes and environmental factors; The prevention suggestion generation module provides personalized prevention suggestions to users based on risk assessment results.

Images