A regional user health risk assessment method, system and device

Abstract

The present invention discloses a regional user health risk assessment method, system and device, the method comprising: obtaining physical examination data and collating vulnerability indicators; obtaining regional environmental data and constructing element combinations; associating to obtain environmental tags; associating the vulnerability indicators with living habits; classifying the vulnerability indicators and living habits to obtain a data pool; obtaining the environmental tags corresponding to the user to be tested and combining the actual living habits as input, matching with the data pool, and obtaining a health risk assessment result. The system comprises: a first data collating module, a second data collating module, a first association module, a second association module, a classification module and a matching module. The device comprises a memory and a processor for executing the above method. By using the present invention, a more comprehensive health risk assessment of users can be performed. The present invention can be widely used in the field of data analysis.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a method, system and device for regional user health risk assessment. Background Art

[0002] With the development of society and the improvement of people's living standards, people are paying more and more attention to health issues, and human health has become a key issue in modern society. More and more people are beginning to pay attention to their own health. However, going to the hospital for cumbersome or unnecessary examinations to check for potential health problems not only wastes personal time and money, but also causes a certain degree of waste of medical resources. Therefore, there is an urgent need for a technology that can assist users in completing the assessment of health problems independently and provide health risk warnings. In addition, most of the existing health risk assessment methods do not take into account regional environmental factors. Summary of the invention

[0003] In view of this, in order to solve the technical problem that the existing user health risk assessment method fails to effectively integrate regional environmental factors, thereby failing to comprehensively assess user health issues, in a first aspect, the present invention proposes a regional user health risk assessment method, the method comprising the following steps:

[0004] Obtain physical examination data and collate frailty indicators;

[0005] Obtain regional environmental data and construct element combinations;

[0006] Associating the vulnerability indicator with the element combination to obtain an environmental label;

[0007] Associating the vulnerability indicators with the living habits to obtain a relationship network;

[0008] Based on the relationship network, the vulnerability indicators and living habits are classified to obtain a data pool;

[0009] The environmental label corresponding to the user to be tested is obtained and combined with the actual living habits as input, and matched with the data pool to obtain the health risk assessment result.

[0010] In some embodiments, the step of obtaining regional environment data and constructing element combinations specifically includes:

[0011] The regional environmental data include atmospheric particulate metal concentrations, drinking water metal concentrations, and soil metal concentrations;

[0012] A preset number of chemical elements are respectively taken from the environmental data of different categories to construct an element combination.

[0013] In some embodiments, the step of associating the vulnerability indicator with the element combination to obtain an environmental label specifically includes:

[0014] According to the regional environmental data, a first environmental matrix is ​​obtained by merging;

[0015] Combining the physical examination data and the fragility index to obtain a first index matrix;

[0016] Based on the first environment matrix and the first indicator matrix, establishing a relationship model by using machine learning;

[0017] Scrambling the first environment matrix to obtain a second environment matrix;

[0018] Inputting the second environment matrix into the relationship model to obtain a second indicator matrix;

[0019] Subtracting the second indicator matrix from the first indicator matrix to obtain a subtraction result;

[0020] Based on the subtraction result, the influence degree of the corresponding element combination and the vulnerability index is calculated, and the combination is screened to obtain the environmental label.

[0021] The calculation formula of the impact degree is as follows:

[0022]

[0023] in, Indicates the difference between the data before and after the operation in the nth row and jth column of the subtraction result, N represents the total number of rows, and B j Indicates the impact degree of the corresponding vulnerability indicator.

[0024] Through this preferred step, the extraction of the combination of regional environmental features is completed, which is helpful for subsequent matching and classification in combination with the label and output of the evaluation result.

[0025] In some embodiments, the step of obtaining the environmental label corresponding to the user to be tested and combining it with the actual living habits as input, matching it with the data pool, and obtaining the health risk assessment result specifically includes:

[0026] According to the data pool, determining anchor points, positive samples and negative samples;

[0027] Define the twin network;

[0028] Inputting the positive sample and the negative sample into the twin network respectively and generating low-dimensional feature representation, and training them in combination with the anchor point to obtain a trained twin network;

[0029] Processing the data pool based on a clustering algorithm to obtain corresponding category center points;

[0030] Obtain the environmental labels corresponding to the users to be tested and construct data points based on their actual living habits;

[0031] Input the data points into the trained twin network to obtain a low-dimensional representation to be tested;

[0032] Calculating the cosine similarity between the low-dimensional representation to be tested and the center point of the category to obtain the similarity of the data point for each category;

[0033] The health risk assessment result of the user to be tested is obtained according to the similarity of the data points to each category.

[0034] Among them, the loss function of the twin network training process is expressed as follows:

[0035]

[0036] Among them, Y represents the label of the same category; D represents the distance between the low-dimensional feature representation of the sample and the anchor point, where the samples include positive samples and negative samples; m represents the preset edge threshold.

[0037] Through this preferred step, the establishment of a health risk assessment and early warning model is completed, and health risk early warnings can be made based on real-time input data.

[0038] In some embodiments, it also includes:

[0039] A correction compensation is introduced to correct the similarity of the data points for each category.

[0040] Through this preferred step, different types of corrections are considered and used to correct the final result, which helps to obtain a more accurate matching classification, including age correction, gender correction and BMI correction.

[0041] In a second aspect, the present invention further proposes a regional user health risk assessment system, the system comprising:

[0042] A first data sorting module is used to obtain physical examination data and sort out vulnerability indicators;

[0043] The second data sorting module is used to obtain regional environmental data and construct element combinations;

[0044] A first association module, used to associate the vulnerability indicator with the element combination to obtain an environmental label;

[0045] A second association module is used to associate the vulnerability indicators with living habits to obtain a relationship network;

[0046] A classification module, based on the relationship network, classifies the vulnerability indicators and living habits to obtain a data pool;

[0047] The matching module is used to obtain the environmental label corresponding to the user to be tested and combine it with the actual living habits as input, match it with the data pool, and obtain the health risk assessment result.

[0048] The present invention also proposes a regional user health risk assessment device, comprising:

[0049] at least one processor;

[0050] at least one memory for storing at least one program;

[0051] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned regional user health risk assessment method.

[0052] Based on the above scheme, the present invention provides a regional user health risk assessment method, system and device, which combines regional environmental data, overcomes the problem of inaccurate health assessment results of traditional methods that do not consider regional factors, and has the ability to realize personalized health risk warnings based on the user's living habits and residential areas, which helps to provide them with personalized health risk warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flowchart of the steps of a regional user health risk assessment method of the present invention;

[0054] Figure 2 It is a structural block diagram of a regional user health risk assessment system of the present invention. DETAILED DESCRIPTION

[0055] Specific problems of the existing technology: 1) It fails to effectively integrate regional environmental factors and lacks the ability to comprehensively evaluate user health problems. 2) The existing technology mainly relies on medical or traditional health data, but it is difficult to consider the profound impact of user living habits on individual health. The present invention introduces regional environmental factors and living habits for evaluation, which helps to achieve accurate evaluation of user health problems.

[0056] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] It should be noted that, for the convenience of description, only the parts related to the relevant invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0058] It should be understood that the "system", "device", "unit" and / or "module" used in this application is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the word can be replaced by other expressions.

[0059] Unless the context clearly indicates an exception, the words "a", "an", "a kind" and / or "the" do not refer to the singular, but also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements. The elements defined by the sentence "includes a..." do not exclude the existence of other identical elements in the process, method, commodity or device that includes the elements.

[0060] In the description of the embodiments of the present application, "plurality" means two or more than two. The following terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0061] In addition, flow charts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, the various steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0062] Reference Figure 1 , is a flow chart of an optional example of a method for regional user health risk assessment proposed by the present invention, which method can be applied to a computer device. The imaging method proposed in this embodiment may include but is not limited to the following steps:

[0063] Step S1, obtaining physical examination data and collating vulnerability indicators;

[0064] Step S2, obtaining regional environmental data and constructing element combinations;

[0065] Step S3: Associating the vulnerability indicator with the element combination to obtain an environmental label;

[0066] Step S4, associating the vulnerability index with the living habits to obtain a relationship network;

[0067] Step S5: Based on the relationship network, classify the vulnerability indicators and living habits to obtain a data pool;

[0068] Step S6: Obtain the environmental label corresponding to the user to be tested and combine it with the actual living habits as input, match it with the data pool, and obtain the health risk assessment result.

[0069] In some feasible embodiments, the step S1 specifically includes:

[0070] Convert physical examination data into corresponding frailty indicators;

[0071] Vulnerability indicators: 1. Hypertension; 2. Heart disease; 3. Stroke; 4. Neurasthenia; 5. Peptic ulcer; 6. Cholecystitis; 7. Diabetes; 8. Bronchitis; 9. Emphysema; 10. Chronic kidney disease; 11. Cancer; 12. Asthma; 13. Chest tightness / dyspnea; 14. Gastroenteritis; 15. COPD; 16. Prostatitis; 17. Rhinitis; 18. Liver disease; 19. Hyperlipidemia; 20. Tuberculosis; 21. BMI; 22. WHR; 23. Whether the electrocardiogram is abnormal; 24. Whether the chest X-ray is abnormal; 25. Pancreatic disease; 26. Coronary heart disease; 27. Thyroid disease; 28. Stomach disease.

[0072] In this embodiment, there are 28 types of personal vulnerability indicators A=[T 1 ,T 2 ,T 3 ,......,T 28 ], each vulnerability indicator is associated with a specific disease, among which T i (i=1,2,3,...,28) the value is:

[0073]

[0074] In personal physical examination data, specific numerical values ​​can usually be obtained. After being judged by doctors or professional instruments, these numerical values ​​can be converted into binary values, that is, 1 or 0.

[0075] In some feasible embodiments, the step S2 specifically includes:

[0076] S2.1. The regional environmental data include atmospheric particulate metal concentrations, drinking water metal concentrations and soil metal concentrations;

[0077] The regional environmental data measured by relevant instruments mainly include atmospheric particulate metal concentration (M 1 ), drinking water metal concentration (M 2), soil metal concentration (M 3 )+three major categories. Each category contains 16 chemical elements, including V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Mo, Cd, Sn, Sb, Tl, and Pb. Therefore, three environmental data matrices can be obtained:

[0078] M 1 =[m 1 ,m 2 ,m 3 ,...,m 16 ]

[0079] M 2 =[m 1 ,m 2 ,m 3 ,...,m 16 ]

[0080] M 3 =[m 1 ,m 2 ,m 3 ,...,m 16 ]

[0081] Where m t , t=1,..,16 are the specific values ​​of the concentrations of different chemical elements in the measured environment.

[0082] S2.2. Taking a preset number of chemical elements from the environmental data of different categories to construct an element combination.

[0083] On this basis, in M 1 、M 2 、M 3 Take one chemical element from each as a combination (total method).

[0084] The subsequent determination is used to determine the value of each combination pair A=[T 1 ,T 2 ,T 3 ,......,T 28 ] was used to determine which indicator had the most significant impact (i.e., the greatest importance). Finally, 28 combinations were selected, each corresponding to a personal vulnerability indicator.

[0085] In some feasible embodiments, the step S3 specifically includes:

[0086] S3.1, merging the regional environmental data to obtain a first environmental matrix;

[0087] First, M 1 、M 2 、M 3Merge into a matrix M:

[0088]

[0089] in Respectively represent M 1 、M 2 、M 3 The corresponding data.

[0090] S3.2, combining the physical examination data and the fragility index to obtain a first index matrix;

[0091] A 1 To A n Merge into a matrix A:

[0092]

[0093] in It represents the value of the nth row and jth column, corresponding to the sequence number of the person and the sequence number of the vulnerability indicator.

[0094] S3.3, based on the first environment matrix and the first indicator matrix, using machine learning to establish a relationship model;

[0095] Use machine learning technology to establish a relationship network between environmental data and personal vulnerability indicators, and obtain the relevant training model N, where A n Represents the overall vulnerability index data of the nth person:

[0096] S3.4, shuffling the first environment matrix to obtain a second environment matrix;

[0097] M' is obtained by randomly shuffling the data in the tth (t=1, 2, 3, ..., 16) column of M up and down (shuffling horizontally will give various combinations).

[0098] S3.5, inputting the second environment matrix into the relationship model to obtain a second indicator matrix;

[0099] Input M' into the previously obtained model N to obtain the new A'.

[0100] S3.6. Subtract the second indicator matrix from the first indicator matrix to obtain a subtraction result;

[0101] The subtraction calculation is expressed as follows:

[0102] D = A'-A;

[0103]

[0104] in Represents the difference between the data before and after the operation on the nth row and jth column.

[0105] S3.7. Based on the subtraction result, calculate the influence degree of the corresponding element combination and the vulnerability index, screen the combination, and obtain the environmental label.

[0106] Then calculate To represent chemical element combinations The impact degree of the jth vulnerability indicator is finally compared with 28 B j Get the maximum value B j , which can represent the combination of chemical elements The jth vulnerable indicator has the greatest impact, and the combination The corresponding j-th vulnerability indicator bundle is defined as the environmental label E j .

[0107] In some feasible embodiments, the step S4 specifically includes:

[0108] Establish a personal vulnerability index A = [T 1 ,T 2 ,T 3 ,......,T 28 ] and the relationship network W of corresponding living habits.

[0109] Among them, living habits can be learned from questionnaires, and the questionnaire options (such as smoking frequency, sleep quality, etc.) can be quantified into life characteristics K = (d 1 ,d 2 ,d 3 ,...,d n )(n∈N + ));

[0110] The personal vulnerability index values ​​in physical examination data often correspond to the individual's living habits, and it has been medically proven that there is a significant correlation between the two. A person's unhealthy living habits are often the factors that lead to his or her illness.

[0111] In some feasible embodiments, the step S5 specifically includes:

[0112] The 28 personal vulnerability indicators are divided into four categories (7 items in each category): high risk R1, medium-high risk R2, medium-low risk R3, and low risk R4. Combined with the relationship network W established above, life characteristics K = (d 1 ,d 2 ,d 3 ,...,d n )(n∈N + ) Following the individual vulnerability indicators, they are also divided into four corresponding categories, that is, four data pools V are established. 1 、V2 、V 3 、V 4 .

[0113] After completing the above steps, in order to determine the main categories to which the living habits belong, a model is used for analysis. The model can determine which category the living habits are most likely to belong to and classify them into the corresponding category. Then, based on the specific environmental labels corresponding to the user's environment, it is further determined which vulnerability indicator in this category is most likely to correspond. The system will recommend that users go to the hospital for targeted examinations to achieve health risk warning effects. The implementation of this process helps save users' time and money, while effectively avoiding the waste of hospital medical resources.

[0114] In some feasible embodiments, the step S6 specifically includes:

[0115] S6.1. Determine anchor points, positive samples, and negative samples according to the data pool;

[0116] Randomly select a sample from each data pool as the anchor point X ahr , and then randomly select another sample from the same data pool as the positive sample X pos , and then randomly select a sample from different data pools as a negative sample X neg .

[0117] S6.2, define twin network;

[0118] Design the structure of the twin network, which contains two identical sub-networks that share weight parameters.

[0119] S6.3, respectively inputting the positive sample and the negative sample into the twin network and generating a low-dimensional feature representation, and training them in combination with the anchor point to obtain a trained twin network;

[0120] The positive and negative samples input respectively eventually produce a low-dimensional representation, and the anchor points and loss functions are trained to obtain a trained twin network model.

[0121] In the twin network model, it is used as the ContrastiveLOSS loss function for calculation:

[0122]

[0123] Among them, Y represents the same category label:

[0124]

[0125] The above formula shows that the anchor point and the positive sample are in V i(i=1,2,3,4) The same data pool takes 1, and the anchor point and negative sample are in the same data pool V i (i=1,2,3,4) then take 0.

[0126] D represents the low-dimensional representation of samples and anchor points F(X) and F(X ahr ), the following is calculated using the Mahalanobis distance:

[0127]

[0128] Among them, ∑ -1 is the covariance matrix N is the size of the data pool, X k is the kth data point in the data pool, and μ is the inverse matrix of the mean vector of the data pool.

[0129] m represents the edge threshold in ContrastiveLOSS and is a predetermined hyperparameter.

[0130] S6.4. Processing the data pool based on a clustering algorithm to obtain corresponding category center points;

[0131] Using clustering algorithm, we get four data pools V 1 、V 2 、V 3 、V 4 The corresponding four category center points C i (i=1,2,3,4).

[0132] S6.5, obtain the environment label corresponding to the user to be tested and construct data points based on actual living habits;

[0133] S6.6, inputting the data points into the trained twin network to obtain a low-dimensional representation to be tested;

[0134] At this time, for a new data point X new , input it into the trained twin network model to obtain the corresponding low-dimensional representation F(X new ), this low-dimensional representation is a data representation with good features obtained through network learning, which contains information about the input data X new Related important feature information can be used to calculate similarity

[0135] S6.7, calculating the cosine similarity between the low-dimensional representation to be tested and the center point of the category, and obtaining the similarity of the data point for each category;

[0136] The following uses cosine similarity to judge X new Which category does it belong to:

[0137]

[0138] Calculate the new data point X new For each category, the similarity S i (i=1,2,3,4).

[0139] S6.8. Obtain a health risk assessment result of the user to be tested based on the similarity of the data points for each category.

[0140] In this embodiment, a specific application scenario example is also given: the user fills in the living habits questionnaire in the system, and the system quantifies the user's living habits into specific numbers that can be recognized by the computer according to the relationship network W. In the first step, the user is preliminarily classified into high-risk R1 according to the quantified living habits characteristics; in the second step, the system obtains the environmental label of the user's region. There are 7 kinds of vulnerable indicators in the high-risk R1 (corresponding to 7 kinds of environmental labels respectively). The system determines the similarity between the user's environmental label (such as [V, Cr, Mn]) and each of the 7 environmental labels (the low-dimensional representation obtained by the twin network model and the cluster center point obtained by machine learning technology are used for similarity calculation), and 7 similarities are obtained and arranged from high to low to be displayed to the user. The vulnerable indicator corresponding to the environmental label with the largest similarity is taken (such as [V, Cr, Mn] has the largest similarity with [V, Cr, Fe], and [V, Cr, Fe] corresponds to chronic kidney disease), and the user is reminded to go to the hospital for examinations related to the vulnerable indicator (chronic kidney disease).

[0141] In some feasible embodiments, the step S6 further includes:

[0142] A correction compensation is introduced to correct the similarity of the data points for each category.

[0143] As age increases, the body's resistance will gradually decline, that is, the probability of illness will continue to increase, and women's resistance is slightly lower than that of men. At the same time, a person's BMI can also more intuitively reflect the strength of their resistance. Therefore, age, gender and BMI occupy an indispensable position in health risk assessment and early warning. For this reason, the present invention introduces the age correction amount P age , Gender correction supplement P gender , BMI correction supplement P bmi To obtain more accurate results, the introduction process is as follows:

[0144] The age correction is defined as follows:

[0145]

[0146] Where α is the age coefficient, which is set by the expert’s prior knowledge; X age is the age of the new data point (0-100 years old); Xstd is the base age, the most obvious turning point of personal physical changes announced by the National Health Commission, i.e. 35 years old; σ is the standard deviation of the Gaussian distribution of the population age.

[0147] The following formula defines the gender correction amount:

[0148] P gender =β·X gender

[0149] Where β is the weak coefficient of gender, which is set after further weakening the expert’s prior knowledge; X gender Gender is a binary quantity:

[0150]

[0151] The following formula defines the BMI correction amount:

[0152]

[0153] Among them, γ, a, and b are BMI strength coefficients, which are set after further strengthening the prior knowledge of experts; X bmi is the BMI value of the new data point.

[0154] At the same time, we further define the interaction compensation I 1 ,I 2 :

[0155] I 1 =P age ·P bmi

[0156]

[0157] From the above, we can get the final correction compensation P final The definition is as follows:

[0158]

[0159] Therefore, a more accurate new data point X can be obtained. new The corresponding membership degree of each category

[0160] Affiliation i (i=1,2,3,4):

[0161] Affiliation i =max i (S i -P final )

[0162] Finally, the category with the highest membership is selected as the new data point X newclassification or corresponding results.

[0163] Based on the above health risk assessment results, targeted early warnings are implemented to optimize users' time and money costs and avoid unnecessary waste of medical resources.

[0164] like Figure 2 As shown, a regional user health risk assessment system includes:

[0165] A first data sorting module is used to obtain physical examination data and sort out vulnerability indicators;

[0166] The second data sorting module is used to obtain regional environmental data and construct element combinations;

[0167] A first association module, used to associate the vulnerability indicator with the element combination to obtain an environmental label;

[0168] A second association module is used to associate the vulnerability indicators with living habits to obtain a relationship network;

[0169] A classification module, based on the relationship network, classifies the vulnerability indicators and living habits to obtain a data pool;

[0170] The matching module is used to obtain the environmental label corresponding to the user to be tested and combine it with the actual living habits as input, match it with the data pool, and obtain the health risk assessment result.

[0171] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0172] A regional user health risk assessment device:

[0173] at least one processor;

[0174] at least one memory for storing at least one program;

[0175] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned regional user health risk assessment method.

[0176] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0177] A storage medium stores processor-executable instructions, which are used to implement a regional user health risk assessment method as described above when executed by the processor.

[0178] The contents of the above method embodiments are all applicable to the present storage medium embodiments. The functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0179] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A regional user health risk assessment method, characterized in that: The following steps are involved: Obtain physical examination data and collate frailty indicators; Vulnerability indicators include: high blood pressure; heart disease; stroke; neurasthenia; Obtain regional environmental data and construct element combinations; Associating the vulnerability indicator with the element combination to obtain an environmental label; Associating the vulnerability indicators with the living habits to obtain a relationship network; Lifestyle habits were obtained by questionnaire, including smoking frequency and sleep quality; Based on the relationship network, the vulnerability indicators and living habits are classified to obtain a data pool; Obtain the environmental label corresponding to the user to be tested and combine it with the actual living habits as input, match it with the data pool, and obtain the health risk assessment result; The step of obtaining regional environmental data and constructing element combinations specifically includes: The regional environmental data include atmospheric particulate metal concentrations, drinking water metal concentrations, and soil metal concentrations; Taking a preset number of chemical elements from the environmental data of different categories to construct an element combination; The step of associating the vulnerability indicator with the element combination to obtain the environmental label specifically includes: According to the regional environmental data, a first environmental matrix is ​​obtained by merging; Combining the physical examination data and the fragility index to obtain a first index matrix; Based on the first environment matrix and the first indicator matrix, establishing a relationship model by using machine learning; Scrambling the first environment matrix to obtain a second environment matrix; Inputting the second environment matrix into the relationship model to obtain a second indicator matrix; Subtracting the second indicator matrix from the first indicator matrix to obtain a subtraction result; Based on the subtraction result, the influence degree of the corresponding element combination and the vulnerability index is calculated, and the combination is screened to obtain an environmental label; The step of obtaining the environmental label corresponding to the user to be tested and combining it with the actual living habits as input, matching it with the data pool, and obtaining the health risk assessment result specifically includes: According to the data pool, determining anchor points, positive samples and negative samples; Define the twin network and design the structure of the twin network, which contains two identical sub-networks that share weight parameters; Inputting the positive sample and the negative sample into the twin network respectively and generating low-dimensional feature representation, and training them in combination with the anchor point to obtain a trained twin network; The loss function of the twin network training process is expressed as follows: Among them, Y represents the label of the same category, D represents the distance between the low-dimensional feature representation of the sample and the anchor point, and m represents the preset edge threshold; Processing the data pool based on a clustering algorithm to obtain corresponding category center points; Obtain the environmental labels corresponding to the users to be tested and construct data points based on their actual living habits; Input the data points into the trained twin network to obtain a low-dimensional representation to be tested; Calculating the cosine similarity between the low-dimensional representation to be tested and the center point of the category to obtain the similarity of the data point for each category; The health risk assessment result of the user to be tested is obtained according to the similarity of the data points to each category.

2. A regional user health risk assessment method according to claim 1, characterized in that: The calculation formula of the impact degree is as follows: in, Indicates the difference between the data before and after the operation in the nth row and jth column of the subtraction result, N represents the total number of rows, and B j Indicates the impact degree of the corresponding vulnerability indicator.

3. A regional user health risk assessment method according to claim 2, characterized in that: Also includes: A correction compensation is introduced to correct the similarity of the data points for each category.

4. A regional user health risk assessment system, characterized in that: Used to perform the user health risk assessment method as claimed in claim 1, comprising: A first data sorting module is used to obtain physical examination data and sort out vulnerability indicators; The second data sorting module is used to obtain regional environmental data and construct element combinations; A first association module, used to associate the vulnerability indicator with the element combination to obtain an environmental label; A second association module is used to associate the vulnerability indicators with living habits to obtain a relationship network; A classification module, based on the relationship network, classifies the vulnerability indicators and living habits to obtain a data pool; The matching module is used to obtain the environmental label corresponding to the user to be tested and combine it with the actual living habits as input, match it with the data pool, and obtain the health risk assessment result.

5. A regional user health risk assessment device, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements a regional user health risk assessment method as described in any one of claims 1-3.

Images