City community environment cognitive health risk assessment method based on intelligent technology

Abstract

The invention provides an urban community environment cognitive health risk assessment method based on an intelligent technology, and relates to the technical field of health community evaluation, and the method comprises the steps: obtaining a cognitive health feature vector; preprocessing the cognitive health feature vectors to obtain a training set and a test set with interactive feature data; inputting the training set into an extreme gradient boosting model for training to obtain a trained extreme gradient boosting model; inputting the test set into the trained extreme gradient boosting model to test the performance to obtain a final extreme gradient boosting model, and analyzing the cognitive health feature vector to obtain the gain importance of the corresponding feature; and performing normalization and sequential arrangement on the gain importance, constructing an environment cognitive health risk assessment index library, analyzing index data, obtaining an environment cognitive health risk level of the corresponding community, and completing assessment of the environment cognitive health risk of the urban community. The problem that the urban community environment cognitive health risk is difficult to accurately evaluate is solved.

Description

Technical Field

[0001] This specification relates to the technical field of health community evaluation, and particularly to a method for assessing the cognitive health risks of urban community environments based on intelligent technologies. Background Art

[0002] The global population aging is showing an increasing trend year by year. As people age, the elderly population will experience varying degrees of degradation both physically and mentally. Therefore, in the context of population aging, the incidence of cognitive impairment among the elderly population is also increasing rapidly. The community is the core living unit of the elderly population and an important node for upstream prevention of cognitive impairment. Rapidly predicting the cognitive health risks of the community environment is an important basis for effectively carrying out planning interventions to reduce the incidence of cognitive impairment. For a long time, urban community health risk assessment has focused on respiratory diseases, endocrine, nutritional and metabolic diseases, and mental and behavioral disorders, ignoring neurological diseases represented by cognitive impairment; the variable selection focuses on demographic factors, physiological factors, and lifestyle factors, making it difficult to assess the potential risks of the community environment on health and lacking a reasonable basis for planning interventions. Therefore, it is urgent to predict the potential risks of the environment on the cognitive health of the elderly population at the community scale. The community environment involves multiple dimensions such as population density, building density, public service facilities, transportation facilities, and blue-green spaces, and there are multi-level and non-linear associations and influence paths with cognitive health, resulting in a wide range of content, diverse data types, large data volumes, and high computational complexity in cognitive health risk assessment. How to collect and process different types of data information, identify potential cognitive health risk factors in the community environment, and achieve efficient collection, rapid processing, accurate prediction, and scientific planning are the difficult problems in current cognitive health risk assessment of the community environment. Summary of the Invention

[0003] Aiming at the above deficiencies in the prior art, the method for assessing the cognitive health risks of urban community environments based on intelligent technologies provided by the present invention solves the problem of difficult accurate assessment of the cognitive health risks of urban community environments.

[0004] To achieve the above invention purpose, the technical solution adopted by the present invention is: A method for assessing the cognitive health risks of urban community environments based on intelligent technologies, comprising: S1: Using the intelligent community governance platform, Montreal Cognitive Assessment Scale, facial emotion recognition database, and Health Promotion Lifestyle Scale, process the community environment data to obtain a cognitive health feature vector; S2: Preprocess the cognitive health feature vector to obtain a training set and a test set with interactive feature data; S3: Input the training set into the extreme gradient boosting model, and use the parameter tuning method based on Bayesian optimization and cross-validation for training to obtain a trained extreme gradient boosting model; wherein, the extreme gradient boosting model uses regression trees as the base learners. S4: Input the test set into the trained extreme gradient boosting model, and use the root mean square error to test the performance to obtain the final extreme gradient boosting model; the final extreme gradient boosting model is used to analyze the cognitive health feature vectors to obtain the gain importance of the corresponding features. S5: Normalize and sort the gain importance to construct an environmental cognitive health risk assessment index library; the environmental cognitive health risk assessment index library is used to analyze the index data to obtain the environmental cognitive health risk level of the corresponding community, and complete the assessment of the environmental cognitive health risk of urban communities.

[0005] The beneficial effects of the present invention are as follows: By using intelligent technologies to efficiently collect community environmental data, demographic data, and physiological health data of the elderly population, it solves the problems commonly existing in traditional data collection methods such as questionnaire collection, image recognition, and on-site research and measurement, including low collection efficiency, high collection cost, and inconsistent data types. According to the Diagnostic and Statistical Manual of Mental Disorders (5th Edition), cognitive health covers six special fields: attention, executive function, learning and memory, language ability, perceptual-motor, and social cognition. Currently, the Montreal Cognitive Assessment Scale commonly used for rapid screening cannot evaluate the health status of the social cognition field. Therefore, by using the Montreal Cognitive Assessment Scale and the Facial Emotion Recognition Database to jointly collect cognitive health data, it is possible to analyze the health status of the elderly population in each cognitive special field, improving the accuracy and comprehensiveness of the assessment of cognitive health status.

[0006] (1) In this way, it is possible to improve the accuracy of community environmental cognitive health risk assessment and the scientific nature of planning intervention strategies, and achieve the community planning goal of reducing the incidence of cognitive impairment. (2) It effectively integrates community environmental data and the health data of the elderly population, considers various potential factors affecting cognitive health, and has stronger comprehensiveness compared to other community health risk assessment methods. (3) It solves the problem of environmental cognitive health risk assessment in community spatial planning. On the basis of considering demographic factors, physiological factors, life behavior factors, and mental health factors, it identifies the risk elements in the community environment, comprehensively assesses the cognitive health risk level of the community environment, and can provide a planning basis for the planning of elderly-friendly communities and the construction of optimized communities for cognitive impairment, thereby improving the quality of the community environment and reducing the incidence of cognitive impairment among the elderly population in the community from the upstream prevention level.

[0007] Further, the S1 includes: Calculate the community environment data using the intelligent community governance platform to obtain community environment variables; Collect the special field data of the elderly population using the Montreal Cognitive Assessment Scale and the Facial Emotion Recognition Database to obtain cognitive health characteristics; wherein, the Montreal Cognitive Assessment Scale is used to collect data in special fields such as attention, executive function, learning and memory, language ability, and perceptual motor, and the Facial Emotion Recognition Database is used to collect data in the cognitive special field; Collect the life behavior characteristic data of the elderly population in the community using the Health Promoting Lifestyle Scale to obtain life behavior characteristics; Collect the physiological data and demographic data of the elderly population in the community using the intelligent community governance platform to obtain the physiological characteristics and demographic characteristics of the elderly population respectively; wherein, the community environment variables, the cognitive health characteristics, the life behavior characteristics, the physiological characteristics and demographic characteristics of the elderly population belong to the cognitive health characteristic vector.

[0008] The impact of the community environment on cognitive health will vary due to factors such as the gender, medical history, and life behavior of individuals. The present invention fully considers various potential factors affecting cognitive health, and through the interactive feature data, tests the moderating effects of demographic characteristics, physiological characteristics, and life behavior characteristics on the cognitive health characteristics of the elderly population, and can reveal the general laws of the community environment affecting the cognitive health of the elderly population, and has stronger scientificity compared with other community health risk assessment methods.

[0009] Further, the S2 includes: Perform screening processing on the cognitive health characteristic vector, remove the missing values, outliers, and duplicate data therein to obtain the processed cognitive health characteristic vector; Perform feature encoding on the physiological characteristics and demographic characteristics of the elderly population in the processed cognitive health characteristic vector to obtain encoded features; Perform standardization and normalization on the encoded features to obtain a training set and a test set with interactive feature data; wherein, the interactive feature data includes "community environment variable × physiological characteristics of the elderly population", "community environment variable × demographic characteristics", and "community environment variable × life behavior characteristics".

[0010] Traditional parameter tuning methods such as grid search and random search have problems such as huge computational complexity, high risk of overfitting or underfitting, incomplete parameter selection, and poor model reliability. The present invention proposes a parameter tuning method based on Bayesian optimization and 10-fold cross-validation. On the one hand, by intelligently selecting parameter combinations and historical data feedback, it can find near-optimal parameters with fewer experimental times, effectively solving the overfitting problem caused by local optimality and significantly improving the computational efficiency of the model; on the other hand, through multiple trainings and validations, it avoids the risk of underfitting and enhances the reliability of the computational model.

[0011] Further, the S3 includes: Based on the parameter search space, adjust the parameters of the initial extreme gradient boosting model to obtain a candidate extreme gradient boosting model; Input the training set into the candidate extreme gradient boosting model, and obtain the scores of each parameter combination by setting the parameter tuning process parameters, the number of search iterations, the number of cross-validation folds, and the performance measurement indicators; By comparing the scores of each parameter combination, obtain the optimal parameter combination; Input the optimal parameter combination into the initial extreme gradient boosting model to obtain a trained extreme gradient boosting model.

[0012] By combining demographic factors, physiological factors, and lifestyle behavior factors, identify the community environmental risks affecting cognitive health, and obtain the strength of the impact of each type of community environmental risk on cognitive health, solving the problem that the lack of evidence-based design basis for community planning due to unclear community environmental risks.

[0013] Further, the environmental cognitive health risk assessment index library is used for: Input the index data into the environmental cognitive health risk assessment index library, and based on the weight calculation, obtain the community environmental cognitive health risk score; Map the community environmental cognitive health risk score to the risk level interval to obtain the environmental cognitive health risk level of the corresponding community, completing the assessment of the urban community environmental cognitive health risk; wherein, the risk level interval is divided based on the determination threshold, including low risk level, medium-low risk level, medium risk level, medium-high risk level, and high risk level.

[0014] Determine the index weights based on the gain importance of the feature vectors, establish an environmental cognitive health risk assessment index library for community, which can accurately predict the cognitive health risk level, and determine the priority of the update project according to the cognitive health risk level, providing a planning basis for the planning of elderly-friendly communities and the construction of optimized communities for cognitive impairment, thereby improving the quality of the community environment and reducing the incidence of cognitive impairment among the elderly population in the community from the upstream prevention level. Description of the Drawings

[0015] This specification will be further described by way of exemplary embodiments, which will be described in detail through the accompanying drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, where: Figure 1 is an exemplary flowchart of a method for assessing cognitive health risks in the urban community environment based on intelligent technology according to some embodiments of this specification. Detailed implementation manners

[0016] The following describes the detailed implementation manners of the present invention to facilitate those skilled in the art of this technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the detailed implementation manners. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are within the scope of protection.

[0017] Embodiment Figure 1 is an exemplary flowchart of a method for assessing cognitive health risks in the urban community environment based on intelligent technology according to some embodiments of this specification. As Figure 1 shown, the process includes the following steps. In some embodiments, the process can be executed by a processor.

[0018] S1: Using the intelligent community governance platform, Montreal Cognitive Assessment Scale, facial emotion recognition database, and Health Promoting Lifestyle Scale, process the community environment data to obtain a cognitive health feature vector.

[0019] The intelligent community governance platform is a platform for processing community environment data.

[0020] The Montreal Cognitive Assessment Scale (MoCA) is an assessment scale that reflects the situation of the elderly in the fields of attention, executive function, learning and memory, language ability, and perceptual motor specialization.

[0021] The facial emotion recognition database is a database for evaluating the status of the elderly in the field of social cognition.

[0022] The Health Promoting Lifestyle Scale is a scale that reflects the health level of the elderly population's life.

[0023] The community environment data is environmental data related to the daily life of community residents. For example, the community environment data may include environmental data such as community density, community land use, community traffic infrastructure, and community blue-green space.

[0024] In some embodiments, the processor can sample multiple community environments through random sampling to obtain community environment data.

[0025] The cognitive health feature vector is a feature vector that reflects the cognitive health status of the elderly in the community environment. For example, the cognitive health feature vector may include community environment variables, cognitive health features, life behavior features, physiological features of the elderly population, and demographic features.

[0026] In some embodiments, the processor may implement S1 based on the following steps: using the intelligent community governance platform to calculate the community environment data to obtain community environment variables; using the Montreal Cognitive Assessment Scale and the facial emotion recognition database to collect the special field data of the elderly population to obtain cognitive health features, wherein the Montreal Cognitive Assessment Scale is used to collect the special field data of attention, executive function, learning and memory, language ability, and perceptual motor, and the facial emotion recognition database is used to collect the special field data of cognition; using the Health Promoting Lifestyle Scale to collect the life behavior feature data of the elderly population in the community to obtain life behavior features; using the intelligent community governance platform to collect the physiological data and demographic data of the elderly population in the community to obtain the physiological features of the elderly population and demographic features respectively.

[0027] The community environment variable is a variable that concretely reflects the community environment data.

[0028] In some embodiments, the processor may obtain the community environment variables based on the manner in Table 1.

[0029] Table 1 Selection and calculation method of community environment variables

[0030] The cognitive health feature is a feature that reflects the health status of the elderly in each special field. For example, the cognitive health feature may include the health features of attention, executive function, learning and memory, language ability, perceptual motor, and cognitive special fields.

[0031] In some embodiments, the processor may use the Montreal Cognitive Assessment Scale (MoCA) to collect the cognitive health features of the elderly population in 5 special fields of attention, executive function, learning and memory, language ability, and perceptual motor.

[0032] In some embodiments, the processor may use the facial emotion recognition database to collect the cognitive health data of the elderly population in the social cognition special field; wherein the facial emotion recognition database consists of a large number of face photos, and each test is randomly composed of 10 different black and white photos of women and 10 different black and white photos of men, and each photo randomly reflects 1 kind of emotion among happy, sad, angry, surprised, fearful, and ashamed emotions; during the test, each photo is played for 15 seconds, and after the playback ends, the subject is asked to select the emotion reflected by the photo, and 1 point can be obtained for a correct answer, with a full score of 20 points.

[0033] The characteristics of life behaviors are the characteristics that reflect the physical activities, eating habits and social activities of the elderly.

[0034] In some embodiments, as shown in Table 2, the processor can use the improved Health-Promoting Lifestyle Profile to collect the life behaviors of the elderly population in the community and obtain the characteristics of life behaviors.

[0035] Table 2 Improved Health-Promoting Lifestyle Profile

[0036] The physiological characteristics of the elderly population are the physiological characteristics that reflect the disease problems of the elderly population. For example, the physiological characteristics of the elderly population can include obesity, hypertension, diabetes, cardiovascular and cerebrovascular diseases, hearing impairment and brain injury, etc.

[0037] Demographic characteristics are the relevant characteristics that reflect the population structure of the community. For example, demographic characteristics can include gender, age, education level, household register, residential structure and family history of cognitive impairment, etc.

[0038] In some embodiments, the processor can use the intelligent community governance platform to collect the physiological data and demographic data of the elderly population in the community and obtain the physiological characteristics and demographic characteristics of the elderly population.

[0039] In some embodiments, the processor can output the community environmental variables, the cognitive health characteristics, the life behavior characteristics, the physiological characteristics and demographic characteristics of the elderly population as a CSV format file and use the Pandas library of Python to read the file to obtain the cognitive health feature vector.

[0040] S2: Preprocess the cognitive health feature vector to obtain a training set and a test set with interactive feature data.

[0041] The training set is a data set used to train the extreme gradient boosting model.

[0042] The test set is a data set used to test the extreme gradient boosting model.

[0043] Interactive feature data is feature data with interactive properties. For example, interactive feature data can include "community environmental variables × physiological characteristics of the elderly population", "community environmental variables × demographic characteristics" and "community environmental variables × life behavior characteristics" In some embodiments, the processor may implement S2 based on the following steps: screening the cognitive health feature vector to remove missing values, outliers, and duplicate data therein, obtaining a processed cognitive health feature vector; performing feature encoding on the physiological characteristics and demographic characteristics of the elderly population in the processed cognitive health feature vector, obtaining encoded features; performing standardization and normalization on the encoded features, obtaining a training set and a test set with interactive feature data.

[0044] The processed cognitive health feature vector is the cognitive health feature vector from which missing values, outliers, and duplicate data have been removed.

[0045] In some embodiments, the processor may use the mean filling method to process missing values, replacing the missing values with the mean, and the calculation formula is: ; where, represents the mean of the cognitive health feature vector data, represents the number of cognitive health feature vector data, represents the th cognitive health feature vector data.

[0046] In some embodiments, the processor may use the box plot method to remove outliers from the cognitive health feature vector.

[0047] The encoded features are the processed cognitive health feature vectors obtained by encoding the physiological characteristics and demographic characteristics of the elderly population. For example, the encoded features may include community environment variables, cognitive health features, life behavior features, encoded physiological characteristics of the elderly population, and encoded demographic characteristics.

[0048] In some embodiments, the processor may, based on the physiological characteristics and demographic characteristics of the elderly population in the processed cognitive health feature vector, use One - Hot encoding to convert physiological feature data, gender, household register, living structure, and family history of cognitive impairment into binary vectors, and use Label encoding to convert the education level into numerical labels, obtaining encoded features.

[0049] In some embodiments, the processor may divide the encoded features in a ratio of 4:1 to obtain a training set and a test set respectively.

[0050] S3: Input the training set into the extreme gradient boosting model, and perform training using a parameter tuning method based on Bayesian optimization and cross - validation to obtain a trained extreme gradient boosting model.

[0051] The Extreme Gradient Boosting model (XGBoost) is a neural network model used to analyze the importance of various community environmental variables. For example, the Extreme Gradient Boosting model can be used to extract variables that significantly affect the cognitive health of the elderly population and serve as the basis for selecting indicators and calculating weights in the indicator library for assessing the cognitive health risks of the community environment. It is an enhanced decision tree model based on the gradient boosting framework, and its core structure is the CART regression tree as the base learner.

[0052] In some embodiments, the processor can use the Scikit-learn library to build an XGBoost regression model, set model parameters such as booster, objective, max_depth, learning_rate, subsample, eval_metric, etc., to obtain the Extreme Gradient Boosting model.

[0053] In some embodiments, the processor can fit each tree according to the residuals of the previous round, minimize the loss function by optimizing the objective function, and introduce a regularization term to control the complexity of the model to obtain the pre-trained Extreme Gradient Boosting model.

[0054] In some embodiments, the expression of the Extreme Gradient Boosting model can be: ; where, represents the regression prediction value of the th sample, represents the prediction function of the th tree, represents the feature vector of the th sample, represents the number of trees.

[0055] In some embodiments, the expression of the objective function can be: ; ; ; where, represents the result of the objective function, represents the loss function, which is used to evaluate the difference between the predicted value and the true value, represents the true value of the th sample, represents the regularization term, represents the number of samples, represents the number of leaf nodes, and represent the regularization coefficients, represents the leaf node weight.

[0056] In some embodiments, the expression of the pre-trained extreme gradient boosting model can be: ; where represents the regression prediction value of the th sample in the th iteration, represents the prediction function of the th iteration, and represents the total number of iterations.

[0057] In some embodiments, the processor can implement S3 based on the following steps: adjusting the parameters of the initial extreme gradient boosting model based on the parameter search space to obtain a candidate extreme gradient boosting model; inputting the training set into the candidate extreme gradient boosting model, and obtaining the scores of each parameter combination by setting the parameters of the tuning process, the number of iterations of the search, the number of cross-validation folds, and the performance measurement index; obtaining the optimal parameter combination by comparing the scores of each parameter combination; and inputting the optimal parameter combination into the initial extreme gradient boosting model to obtain a trained extreme gradient boosting model.

[0058] The initial extreme gradient boosting model is an extreme gradient boosting model using default parameters.

[0059] In some embodiments, the processor can use the community environment variables and interaction feature data in the training set as the input data of the initial extreme gradient boosting model, and the cognitive health features as the labels of the initial extreme gradient boosting model for training to obtain a trained extreme gradient boosting model.

[0060] The candidate extreme gradient boosting model is an extreme gradient boosting model used to test the parameter scores after parameter adjustment.

[0061] In some embodiments, as shown in Table 3, the processor can adjust the parameters of the initial extreme gradient boosting model based on the parameter search space to obtain a candidate extreme gradient boosting model.

[0062] Table 3 Parameter Search Space Settings

[0063] The parameter combination score is a score reflecting the performance of each parameter combination.

[0064] In some embodiments, the processor can 2) set the parameters of the tuning process, set the number of iterations of the search to 50 times, set the number of cross-validation folds to 10 folds, and set the model performance measurement index to neg_mean_squared_error to obtain the parameter combination score.

[0065] The optimal parameter combination is the parameter combination with the highest parameter combination score.

[0066] S4: Input the test set into the trained extreme gradient boosting model, use the root mean square error to test the performance, and obtain the final extreme gradient boosting model; the final extreme gradient boosting model is used to analyze the cognitive health feature vector to obtain the gain importance of the corresponding feature.

[0067] In some embodiments, the processor can input the test set into the trained extreme gradient boosting model, use the community environmental variables and interaction feature data as the input data of the trained extreme gradient boosting model, and the cognitive health feature as the label of the trained extreme gradient boosting model for testing. By viewing the model prediction effect through the root mean square error (RMSE), if the root mean square error is within 10% to 20% of the predicted value range, it indicates that the model has a good effect, and the final extreme gradient boosting model is obtained; otherwise, the model is readjusted.

[0068] The gain importance is a score reflecting the importance of community environmental variables.

[0069] S5: Normalize and sort the gain importance to construct an environmental cognitive health risk assessment index library; the environmental cognitive health risk assessment index library is used to analyze the index data to obtain the environmental cognitive health risk level of the corresponding community, and complete the assessment of the environmental cognitive health risk of urban communities.

[0070] The environmental cognitive health risk assessment index library standardizes the importance scores of community environmental variables into weights within the range of 0 - 1, forming an index library based on community density, community land use, community transportation infrastructure, and community blue - green space, and is used for comprehensive evaluation of the community environment.

[0071] In some embodiments, the processor can, based on the relative importance ranking of the gain importance, select the variables with significant influence and standardize their importance scores into weights within the range of 0 - 1. The calculation method is: ; where represents the weight standardized within the range of 0 - 1, represents the importance score.

[0072] The index data is the community environmental variable data of the current urban community. For example, the index data can include community density, community land use, community transportation infrastructure, and community blue - green space data, etc.

[0073] The environmental cognitive health risk level is a level reflecting the degree of cognitive health risk in the current urban community. For example, the environmental cognitive health risk can include low - risk level, medium - low - risk level, medium - risk level, medium - high - risk level, and high - risk level. In some embodiments, the processor may divide the cognitive health risk into five levels: low risk, medium - low risk, medium risk, medium - high risk, and high risk by using the Equal Interval Method based on determination thresholds P1 and P2, where P1 < P2.

[0074] In some embodiments, the processor may use the environmental cognitive health risk assessment index library to obtain the environmental cognitive health risk level: input the index data into the environmental cognitive health risk assessment index library, and based on weight calculation, obtain the community environmental cognitive health risk score; map the community environmental cognitive health risk score to the risk level interval to obtain the environmental cognitive health risk level of the corresponding community, thus completing the assessment of the environmental cognitive health risk of urban communities.

[0075] In some embodiments, the thresholds P1 and P2 can be set according to the environmental cognitive health risk scores of communities with relatively low and high incidence of cognitive impairment.

[0076] The community environmental cognitive health risk score is a score reflecting the level of environmental cognitive health risk of the corresponding urban community.

[0077] In some embodiments, the processor may map the community environmental cognitive health risk score P to the risk level interval to obtain the community environmental cognitive health risk score of the community. For example, the processor may calculate the community environmental cognitive health risk score based on the index data and index weights, and the calculation method is: ; where, represents the community environmental cognitive health risk score, represents the index data; The community environmental cognitive health risk score is mapped to the risk level interval to obtain the environmental cognitive health risk level of the evaluated community.

[0078] In some embodiments, the processor may derive a community environmental cognitive health risk assessment report based on the environmental cognitive health risk level and the community environmental cognitive health risk score, and determine the goals and priorities of planning interventions.

[0079] In some embodiments of this specification, a method for assessing the environmental cognitive health risk of urban communities based on intelligent technology efficiently collects community environmental data by using intelligent technology, combines demographic factors, physiological factors, and lifestyle factors, identifies environmental risks affecting cognitive health, establishes a community environmental cognitive health risk assessment index library, and accurately predicts the cognitive health risk level according to the indexes, solving the problem that the community planning lacks evidence - based design basis due to unclear community environmental risks and unknown risk levels.

[0080] (1) In this way, the accuracy of the cognitive health risk assessment of the community environment can be improved, the scientific nature of the planning intervention strategy can be realized, and the community planning goal of reducing the incidence of cognitive impairment can be achieved; (2) It effectively integrates community environmental data and the health data of the elderly population, takes into account various potential factors affecting cognitive health, and has stronger comprehensiveness compared with other community health risk assessment methods; (3) It solves the problem of cognitive health risk assessment in community spatial planning. On the basis of considering demographic factors, physiological factors, lifestyle behavior factors and mental health factors, it identifies risk elements in the community environment, comprehensively assesses the cognitive health risk level of the community environment, and can provide a planning basis for the planning of elderly-friendly communities and the construction of optimized communities for cognitive impairment, thereby improving the quality of the community environment and reducing the incidence of cognitive impairment among the elderly population in the community from the upstream prevention level.

Claims

1. A method for assessing the environmental cognitive health risks of urban communities based on intelligent technology, characterized in that: include: S1: Using the smart community governance platform, the Montreal Cognitive Assessment Scale, the Facial Emotion Recognition Database, and the Health Promotion Lifestyle Scale, the community environment data was processed to obtain the cognitive health feature vector; S2: preprocessing the cognitive health feature vector to obtain a training set and a test set with interactive feature data; S3: inputting the training set into the extreme gradient boosting model, and training using a parameter tuning method based on Bayesian optimization and cross-validation to obtain a trained extreme gradient boosting model; wherein the extreme gradient boosting model uses a regression tree as a base learner; S4: inputting the test set into the trained extreme gradient boosting model, using the root mean square error to test the performance, and obtaining the final extreme gradient boosting model; the final extreme gradient boosting model is used to analyze the cognitive health feature vector to obtain the gain importance of the corresponding feature; S5: Normalize and arrange the gain importance in order to construct an environmental cognitive health risk assessment index library; the environmental cognitive health risk assessment index library is used to analyze the index data, obtain the environmental cognitive health risk level of the corresponding community, and complete the assessment of the environmental cognitive health risk of the urban community.

2. The method for evaluating the environmental cognitive health risks of urban communities based on intelligent technology according to claim 1 is characterized in that: The S1 includes: Use the smart community governance platform to calculate community environmental data and obtain community environmental variables; Using the Montreal Cognitive Assessment Scale and the Facial Emotion Recognition Database, special field data of the elderly population are collected to obtain cognitive health characteristics; wherein the Montreal Cognitive Assessment Scale is used to collect special field data of attention, executive function, learning and memory, language ability, and perception and movement, and the Facial Emotion Recognition Database is used to collect special field data of cognition; The health-promoting lifestyle scale was used to collect data on the life behavior characteristics of the elderly population in the community and obtain the life behavior characteristics; By using the smart community governance platform, the physiological data and demographic data of the elderly population in the community are collected to obtain the physiological characteristics and demographic characteristics of the elderly population, respectively; wherein the community environmental variables, the cognitive health characteristics, the living behavior characteristics, the physiological characteristics and demographic characteristics of the elderly population belong to the cognitive health feature vector.

3. The method for evaluating the environmental cognitive health risks of urban communities based on intelligent technology according to claim 2 is characterized in that: The S2 includes: Screening the cognitive health feature vector to remove missing values, outliers and duplicate data to obtain a processed cognitive health feature vector; Perform feature coding on the physiological characteristics and demographic characteristics of the elderly population in the processed cognitive health feature vector to obtain coding features; The coding features are standardized and normalized to obtain a training set and a test set with interactive feature data; wherein the interactive feature data includes "community environment variables × physiological characteristics of the elderly population", "community environment variables × demographic characteristics" and "community environment variables × life behavior characteristics".

4. The method for evaluating the environmental cognitive health risks of urban communities based on intelligent technology according to claim 1 is characterized in that: The S3 includes: Based on the parameter search space, adjust the parameters of the initial extreme gradient boosting model to obtain a candidate extreme gradient boosting model; Inputting the training set into the candidate extreme gradient boosting model, and obtaining the score of each parameter combination by setting tuning process parameters, the number of search iterations, the number of cross-validation folds, and the performance measurement index; By comparing the scores of the parameter combinations, an optimal parameter combination is obtained; The optimal parameter combination is input into the initial extreme gradient boosting model to obtain a trained extreme gradient boosting model.

5. The method for evaluating the environmental cognitive health risks of urban communities based on intelligent technology according to claim 1 is characterized in that: The environmental cognitive health risk assessment indicator library is used for: Inputting the indicator data into the environmental cognitive health risk assessment indicator library, and obtaining a community environmental cognitive health risk score based on weight calculation; The community environmental cognitive health risk score is mapped to a risk level interval to obtain the environmental cognitive health risk level of the corresponding community, thereby completing the assessment of the environmental cognitive health risk of the urban community; wherein the risk level interval is divided based on a determination threshold, including a low risk level, a medium-low risk level, a medium risk level, a medium-high risk level and a high risk level.

Images