A method and system for identifying epidemic risk factors based on association rules

Through the method based on association rules, the correlation between multi-dimensional factors is identified, which solves the problems of inefficiency and dependence on prior knowledge of traditional statistical analysis methods, and effectively recognizes epidemic risk factors, improves the efficiency and accuracy of data analysis. sex.

CN114548646BActive Publication Date: 2024-11-26CENT SOUTH UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111532410.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-15
Publication Date
2024-11-26
Estimated Expiration
2041-12-15

AI Technical Summary

Technical Problem

In the existing research on epidemic factors, traditional statistical analysis methods are inefficient, difficult to identify potential risk factors in multi-dimensional data, and rely on prior knowledge, which has great limitations.

Method used

Using a method based on association rules, through data extraction, feature hierarchy and labeling, frequent pattern tree construction and association rule generation, the correlation between multi-dimensional factors is identified, efficiency is improved, and limitations of data scale and dimension are overcome.

Benefits of technology

It has achieved efficient identification of potential epidemic transmission risk factors, improved data robustness and ductility, avoided the generation of meaningless rules, and obtained valuable correlation rules easily and quickly, greatly improved efficiency, and overcome the limitations of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114548646B_ABST
    Figure CN114548646B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for identifying epidemic risk factors based on association rules, comprising the steps of: extracting data from official data sources; transforming the data form of non-discrete data in the extracted data; performing data mining on the data after the data form transformation, generating association rules, and identifying risk factors; first analyzing and sorting the association rules, and then analyzing the risk factors to obtain the risk degree of the risk factors; also discloses an epidemic risk factor identification system based on association rules, the system comprising a data extraction module, a data processing module, a data mining module and a comprehensive analysis module; the method realizes the conversion of the form of epidemic-related data by performing feature classification and labeling on epidemic-related data, thereby improving the robustness and scalability of the data, so that association rules can be used to perform data mining and analysis on large-scale, multi-dimensional data, and are no longer limited by the scale and dimension of the data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of epidemic risk factor identification, and in particular, relates to an epidemic risk factor identification method and system based on association rules. Background Art

[0002] Most of the existing studies on epidemic factors are based on traditional statistical analysis methods for quantitative analysis. Such methods usually analyze and test the relationship between two factors one by one. When there are too many factor dimensions and the data scale is large, the efficiency will be extremely low. At the same time, such methods require a certain amount of manual intervention, and often have certain limitations because they are based on prior knowledge and it is difficult to identify implicit potential risk factors. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art that are limited by the scale and dimension of the data and the low inspection efficiency, and to provide a method for identifying epidemic risk factors that is no longer limited by the scale and dimension of the data and can simultaneously explore the correlation between multiple dimensional factors. The method is a highly efficient method for identifying epidemic risk factors based on association rules.

[0004] Another object of the present invention is to provide an epidemic risk factor identification system based on association rules according to an epidemic risk factor identification method based on association rules.

[0005] The present invention provides a method for identifying epidemic risk factors based on association rules, comprising the steps of:

[0006] S1: extract data from official data sources;

[0007] S2: Transform the data format of the non-discrete data in the extracted data; first classify the features of the non-discrete data to obtain the feature value; then label it, assign a feature number to each extracted data, and combine the feature value with the feature number to form a combined feature value;

[0008] S3: Perform data mining on the transformed data, generate association rules, and identify risk factors;

[0009] S4: First analyze and sort the association rules, then analyze the risk factors to obtain the risk level of the risk factors.

[0010] Preferably, in S1, the official data source includes infection case data, meteorological data, and population flow data, the infection case data includes the number of infections, and the age and gender of the infected persons; the meteorological data includes temperature, rainfall and air quality index; the population flow data includes a population in-migration index and a population out-migration index, and data is extracted from the official data source every day to construct a data set, and the data extracted every day includes date, gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index and population out-migration index.

[0011] Preferably, S2 comprises the following specific steps:

[0012] S2.1: Feature classification; gender, age, number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index are all feature items. For age, classification and feature value selection are performed according to age groups to obtain feature values; for gender, the feature value corresponding to male is "1" and the feature value corresponding to female is "2"; for the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index, the mean and standard deviation are calculated respectively, and classification and feature value selection are performed according to the numerical range of the corresponding mean and standard deviation to obtain the feature values ​​corresponding to the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index respectively;

[0013] S2.2: Labeling: assign a feature number to gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index, and population out-migration index respectively;

[0014] S2.3: Combine each eigenvalue obtained in S2.1 with the corresponding eigennumber in S2.2 to form a combined eigenvalue.

[0015] Preferably, in S3, the specific steps of performing data mining, generating association rules, and identifying risk factors are:

[0016] Find frequent itemsets.

[0017] S3.1: Use the frequent pattern algorithm to traverse the data set and sort it in descending order according to the frequency of occurrence. Remove the feature items whose frequency is less than the minimum support to obtain the frequent item set, which is recorded as F1;

[0018] S3.2: For the data extracted daily, after removing the feature items whose frequency is less than the minimum support, the remaining feature items are re-sorted in descending order according to the frequency of occurrence;

[0019] S3.3: Initialize the root node of the frequent pattern tree, use the frequent pattern algorithm to traverse the data set again, insert the feature items sorted in descending order in S3.2 into the frequent pattern tree, and obtain a new frequent pattern tree. Each node in the new frequent pattern tree represents a feature item and the count corresponding to the feature item;

[0020] S3.4: According to the new frequent pattern tree, traverse each feature item in the frequent item set F1 from bottom to top to the ancestor node of the feature item, obtain the feature items contained in the corresponding path, and form a feature item set. The count of the end node of the path is the count of the feature item set. The feature item set with a count greater than or equal to the minimum support is the final frequent item set.

[0021] Generate association rules,

[0022] S3.5: In each final frequent item set, the attributes of the feature items are judged to define association rules. The association rules are recorded as X→Y, where the former item X represents the feature item with the risk possibility factor, and the latter item Y represents the feature item with the infection factor. Each final frequent item set includes multiple association rules. The confidence of each association rule X→Y is calculated. If the confidence is greater than or equal to the minimum confidence, X→Y is a strong association rule, otherwise it is a weak association rule.

[0023] S3.6: Calculate the lift for each strong association rule. If the lift is greater than or equal to 1, the strong association rule is a valid strong association rule, and the factor of the antecedent X of the strong association rule is the risk factor.

[0024] Preferably, in S4, the specific steps of obtaining the risk level of the risk factor are:

[0025] S4.1: Conduct in-depth analysis on each association rule according to the lift degree, and sort the association rules from high to low according to the lift degree;

[0026] S4.2: For the sorted association rules, analyze the degree of influence of the risk factors contained in the antecedent of the association rules on the infection factors contained in the latter according to the confidence level; the higher the confidence level, the higher the risk of the risk factor in the spread of the epidemic.

[0027] Preferably, in S2.2, the feature numbers are numbered in ascending order of gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index, and population out-migration index.

[0028] Preferably, S3.3 further includes that when data extracted from one day is inserted each time, if the corresponding feature item already exists in the frequent pattern tree, the count is increased by 1; if not, a new branch is added to the frequent pattern tree.

[0029] Preferably, the calculation formula of support is:

[0030] Support(X→Y)=P(X∪Y),Support(X)=P(X)

[0031] Where P(X) represents the proportion of X in the data set, and P(X) = Count(X) / Count(D), Count(X) represents the count of X in the data set, X represents the antecedent of the association rule, and Count(D) represents the total count in the data set;

[0032] The confidence calculation formula is:

[0033]

[0034] The calculation formula for lift is:

[0035]

[0036] Among them, P(X∪Y) represents the proportion of X∪Y in the data set, P(Y) represents the proportion of Y in the data set, and Y represents the latter term of the association rule.

[0037] Preferably, in S3.3, the method for obtaining a new frequent pattern tree is: inserting the feature items sorted in descending order in S3.2 into the frequent pattern tree from top to bottom in sequence, connected by edges, to obtain a new frequent pattern tree.

[0038] The present invention also provides an epidemic risk factor identification system based on association rules, which is applied to the above-mentioned epidemic risk factor identification method based on association rules, and includes a data extraction module, a data processing module, a data mining module and a comprehensive analysis module. The data extraction module is used to extract data from official data sources; the data processing module is used to transform the data format of non-discrete data in the extracted data; the data mining module is used to perform data mining on the data after the data format transformation, generate association rules, and identify risk factors; the comprehensive analysis module is used to analyze and sort the association rules, and then analyze the risk factors to obtain the risk level of the risk factors.

[0039] Beneficial effects:

[0040] 1. This method can identify potential epidemic transmission risk factors based on association rules, overcoming the limitation of general statistical analysis methods that rely too much on prior knowledge and fail to identify potential transmission risk factors.

[0041] 2. This method achieves the conversion of epidemic-related data by classifying and labeling its features, thereby improving the robustness and scalability of the data. It can then use association rules to conduct data mining and analysis on large-scale, multi-dimensional data, and is no longer limited by the scale and dimension of the data.

[0042] 3. This method is a non-purposeful unsupervised learning method. By traversing the entire data and generating association rules, it can simultaneously explore the correlation between multiple dimensional factors. Compared with general statistical methods that analyze the correlation between two factors one by one, the efficiency is greatly improved.

[0043] 4. This method can be used to identify the factor attributes of the preceding and following items when generating association rules, avoid the generation of meaningless rules, and obtain valuable association rules simply and quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 It is a schematic diagram of the framework of the epidemic risk factor identification method and system based on association rules in the implementation of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, this embodiment provides an epidemic risk factor identification system based on association rules and an epidemic risk factor identification method based on association rules. The system is applied to the method and includes a data extraction module, a data processing module, a data mining module and a comprehensive analysis module. The data extraction module is used to extract data from official data sources; the data processing module is used to transform the data format of non-discrete data in the extracted data; the data mining module is used to perform data mining on the data after the data format transformation, generate association rules, and identify risk factors; the comprehensive analysis module is used to analyze and sort the association rules, and then analyze the risk factors to obtain the risk level of the risk factors.

[0048] A method for identifying epidemic risk factors based on association rules, comprising the steps of:

[0049] S1: extract data from official data sources;

[0050] Specifically, in the process of epidemic spread, the related factors involved are complex and diverse. Therefore, when studying the risk factors of epidemic spread, comprehensive considerations should be made from multiple dimensions. This method starts with data. In order to ensure the authenticity and reliability of the data, the data are all from authoritative official websites. The official data sources include infection case data, meteorological data, and population mobility data. The infection case data comes from the website of Nanjing Municipal Health Commission, the meteorological data comes from the National Meteorological Science Data Center and the International Air Quality Website, and the population mobility data comes from Baidu Map Migration Big Data. The infection case data includes the number of infections, as well as the age and gender of the infected. Subsequent cases that are converted from asymptomatic to confirmed are no longer counted repeatedly; meteorological data include temperature, rainfall and air quality index; population mobility data include population in-migration index and population out-migration index. Data are extracted from official data sources every day to construct a data set. The data extracted daily include date, gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index and population out-migration index.

[0051] Compared with previous studies on epidemic factors, which rarely studied the impact of multidimensional factors on the spread of the epidemic, or studied the risk factors that affect treatment, or studied the impact of a single factor on the spread of the epidemic, this method collects relevant data from many aspects to consider the risk of multidimensional factors in the process of epidemic spread.

[0052] S2: Transform the data format of the non-discrete data in the extracted data; first classify the features of the non-discrete data to obtain the feature value; then label it, assign a feature number to each extracted data, and combine the feature value with the feature number to form a combined feature value;

[0053] Specifically, the data format transformation is performed for the non-discrete data in the extracted data, mainly by first performing feature classification and then labeling, two steps, the steps are:

[0054] S2.1: Feature classification; gender, age, number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index are all feature items. For age, classification and feature value selection are performed according to age groups to obtain feature values; for gender, the feature value corresponding to male is "1" and the feature value corresponding to female is "2"; for the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index, the mean and standard deviation are calculated respectively, and classification and feature value selection are performed according to the numerical range of the corresponding mean and standard deviation to obtain the feature values ​​corresponding to the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index respectively;

[0055] Specific examples of feature classification are shown in Tables 1, 2 and 3 below;

[0056] Table 1 Age characteristics classification;

[0057] age Eigenvalue x<18 1 18≤x≤60 2 60<x 3

[0058] As can be seen from Table 1, age is classified and characterized according to age groups to obtain the characteristic value of age; when the age is less than 18 years old, the characteristic value of age is "1"; when the age is greater than or equal to 18 years old and less than or equal to 60 years old, the characteristic value of age is "2"; when the age is greater than 60 years old, the characteristic value of age is "3".

[0059] Table 2 Classification of air quality index characteristics;

[0060] Air Quality Index Eigenvalue 21.46≤x≤50.7 1 x<21.46 2 x>50.7 3

[0061] After calculation, the average value of the air quality index is 36.08, and the standard deviation is 14.62. It can be seen from Table 2 that the characteristic value of the air quality index is obtained by grading and selecting the characteristic value according to the numerical range of the average value and the standard deviation; when the air quality index is greater than or equal to 21.46 and less than or equal to 50.7, the characteristic value of the air quality index is "1"; when the air quality index is less than 21.46, the characteristic value of the air quality index is "2"; when the air quality index is greater than 50.7, the characteristic value of the air quality index is "3".

[0062] Table 3 Classification of characteristics of population out-migration index;

[0063] Population emigration index Eigenvalue 0.37≤x≤2.11 1 x<0.37 2 x>2.11 3

[0064] After calculation, the average value of the population out-migration index is 1.24, and the standard deviation is 0.87. It can be seen from Table 3 that the characteristic value of the population out-migration index is obtained by grading and selecting characteristics according to the numerical range of the average value and standard deviation; when the population out-migration index is greater than or equal to 0.37 and less than or equal to 2.11, the characteristic value of the population out-migration index is "1"; when the population out-migration index is less than 0.37, the characteristic value of the population out-migration index is "2"; when the population out-migration index is greater than 2.11, the characteristic value of the population out-migration index is "3".

[0065] S2.2: Labeling; assign a feature number to each of gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index, and population out-migration index.

[0066] S2.3: Combine each eigenvalue obtained in S2.1 with the corresponding eigennumber in S2.2 to form a combined eigenvalue.

[0067] The feature numbers are numbered in ascending order according to gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index, and population out-migration index; for example, the feature number of gender is "01", the feature number of age is "02", and so on; the feature value is combined with the corresponding feature number to form a combined feature value, for example: "021" means the feature with a feature value of "1" under the feature number "02", which means the age is under 18 years old.

[0068] Compared with previous data forms, the original data features of this method are mostly in continuous numerical form. Since continuous features have poor robustness and will interfere with the algorithm model, this method realizes the conversion of data form through feature classification and labeling in the data processing link, improves the robustness and scalability of the data, and is conducive to further mining and analysis of the data.

[0069] S3: Perform data mining on the transformed data, generate association rules, and identify risk factors;

[0070] Specifically, the steps for conducting data mining, generating association rules, and identifying risk factors are as follows:

[0071] Find frequent item sets, use the frequent pattern algorithm (fp-growth algorithm) to traverse the data set, construct a frequent pattern tree (fp-tree) to obtain frequent item sets.

[0072] S3.1: Use the frequent pattern algorithm (fp-growth algorithm) to traverse the data set, and sort it in descending order according to the frequency of occurrence, remove the feature items whose frequency is less than the minimum support, and obtain the frequent item set, which is denoted as F1.

[0073] S3.2: For the data extracted daily, after removing the feature items whose frequency is less than the minimum support, the remaining feature items are re-sorted in descending order according to the frequency of occurrence.

[0074] S3.3: Initialize the root node of the frequent pattern tree (fp-tree), use the frequent pattern algorithm (fp-growth algorithm) to traverse the data set again, and insert the feature items sorted in descending order in S3.2 into the frequent pattern tree (fp-tree) from top to bottom, connected by edges, to obtain a new frequent pattern tree (fp-tree). Each node in the new frequent pattern tree (fp-tree) represents a feature item and the corresponding count of the feature item; when inserting the data extracted from one day, if the corresponding feature item already exists in the frequent pattern tree (fp-tree), the count is increased by 1. If it does not exist, a new branch is added to the frequent pattern tree (fp-tree).

[0075] S3.4: According to the new frequent pattern tree (fp-tree), each feature item in the frequent item set F1 is traversed from bottom to top to the ancestor node of the feature item, and the feature items contained in the corresponding path are obtained to form a feature item set. The count of the end node of the path is the count of the feature item set. The feature item set whose count is greater than or equal to the minimum support is the final frequent item set.

[0076] Generate association rules, calculate confidence based on frequent item sets, and generate association rules when the minimum support threshold is met.

[0077] S3.5: In each final frequent item set, the attributes of the feature items are judged to define association rules. The association rules are recorded as X→Y, where the former item X represents the feature item with risk possibility factors, and the latter item Y represents the feature item with infection factors. Each final frequent item set includes multiple association rules. The confidence (Confidence) of each association rule X→Y is calculated. If the confidence (Confidence) is greater than or equal to the minimum confidence, X→Y is a strong association rule, otherwise it is a weak association rule.

[0078] S3.6: For each strong association rule, the lift (Lift) is calculated respectively. If the lift (Lift) is greater than or equal to 1, the strong association rule is a valid strong association rule, and the factor of the antecedent X of the strong association rule is the risk factor.

[0079] The calculation formula for support is:

[0080] Support(X→Y)=P(X∪Y),Support(X)=P(X)

[0081] Where P(X) represents the proportion of X in the data set, and P(X) = Count(X) / Count(D), Count(X) represents the count of X in the data set, X represents the antecedent of the association rule, and Count(D) represents the total count in the data set;

[0082] The confidence calculation formula is:

[0083]

[0084] The calculation formula for lift is:

[0085]

[0086] Among them, P(X∪Y) represents the proportion of X∪Y in the data set, P(Y) represents the proportion of Y in the data set, and Y represents the latter term of the association rule.

[0087] In this embodiment, the FP-Growth algorithm is used to analyze multi-dimensional factors, and the frequent item sets are obtained by traversing the entire data set twice. When generating association rules, the factor attributes of the preceding and following items of the rules are judged to avoid the generation of meaningless rules and to obtain valuable association rules simply and quickly. In this link, the correlation between factors of multiple dimensions can be explored at the same time, and risk factors can be quickly and easily identified based on the obtained association rules, which greatly improves efficiency.

[0088] S4: First analyze and sort the association rules, then analyze the risk factors to obtain the risk level of the risk factors;

[0089] Through the data mining in the previous module, we can mine practical association rules from the epidemic data and identify epidemic risk factors; according to the support (Support), confidence (Confidence) and lift (Lift) of each association rule, we can conduct in-depth analysis of the corresponding association rules.

[0090] Specifically, the steps to obtain the risk level of the risk factor are:

[0091] S4.1: Support is a macro-statistical analysis of association rules, that is, the proportion of association rules in the entire data set; lift reflects the effectiveness of association rules. The higher the lift, the stronger the positive correlation. Therefore, each association rule is deeply analyzed according to the lift, and the association rules are sorted from high to low according to the lift.

[0092] S4.2: For the sorted association rules, analyze the degree of influence of the risk factors contained in the antecedent of the association rules on the infection factors contained in the latter according to the confidence level. The higher the confidence level, the higher the risk of the risk factor in the spread of the epidemic. Corresponding countermeasures and suggestions can be further proposed for the high-risk factors identified in the analysis, thereby facilitating epidemic prevention and control work.

[0093] The epidemic risk factor identification method and system based on association rules provided in this embodiment have the following beneficial effects:

[0094] 1. This method can identify potential epidemic transmission risk factors based on association rules, overcoming the limitation of general statistical analysis methods that rely too much on prior knowledge and fail to identify potential transmission risk factors.

[0095] 2. This method achieves the conversion of epidemic-related data by classifying and labeling its features, thereby improving the robustness and scalability of the data. It can then use association rules to conduct data mining and analysis on large-scale, multi-dimensional data, and is no longer limited by the scale and dimension of the data.

[0096] 3. This method is a non-purposeful unsupervised learning method. By traversing the entire data and generating association rules, it can simultaneously explore the correlation between multiple dimensional factors. Compared with general statistical methods that analyze the correlation between two factors one by one, the efficiency is greatly improved.

[0097] 4. This method can be used to identify the factor attributes of the preceding and following items when generating association rules, avoid the generation of meaningless rules, and obtain valuable association rules simply and quickly.

[0098] 5. The experimental results show that this method can accurately and effectively identify transmission risk factors from epidemic data, providing certain help and reference for epidemic prevention and control work.

[0099] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions or improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying epidemic risk factors based on association rules. It is characterized in that Includes steps: S1: Extract data from official data sources; The official data source includes infection case data, meteorological data, and population flow data. The infection case data includes the number of infected persons, as well as the age and gender of the infected persons; the meteorological data includes temperature, rainfall, and air quality index; the population flow data includes a population in-migration index and a population out-migration index. Data is extracted from the official data source every day to construct a data set. The data extracted every day includes date, gender, age, number of infected persons, temperature, rainfall, air quality index, population in-migration index, and population out-migration index; S2: Transform the data format of the non-discrete data in the extracted data; first classify the features of the non-discrete data to obtain the feature value; then label it, assign a feature number to each extracted data, and combine the feature value with the feature number to form a combined feature value; S2.1: Feature classification of non-discrete data in the extracted data; the gender, age, number of infected people, temperature, rainfall, air quality index, population in-migration index and population out-migration index are all feature items. For age, classification and feature value selection are performed according to age groups to obtain feature values; for gender, the feature value corresponding to male is "1" and the feature value corresponding to female is "2"; for the number of infected people, temperature, rainfall, air quality index, population in-migration index and population out-migration index, the mean and standard deviation are calculated respectively, and classification and feature value selection are performed according to the numerical range of the corresponding mean and standard deviation to obtain the feature values ​​corresponding to the number of infected people, temperature, rainfall, air quality index, population in-migration index and population out-migration index respectively; S2.2: Label the non-discrete data in the extracted data; assign a feature number to gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index, and population out-migration index respectively; S2.3: Combine each eigenvalue obtained in S2.1 with the corresponding eigennumber in S2.2 to form a combined eigenvalue; S3: Perform data mining on the transformed data, generate association rules, and identify risk factors; The specific steps to conduct data mining, generate association rules, and identify risk factors are: Find frequent itemsets. S3.1: Use the frequent pattern algorithm to traverse the data set and sort it in descending order according to the frequency of occurrence, remove the feature items whose frequency is less than the minimum support, and obtain the frequent item set, which is denoted as F 1 ; S3.2: For the data extracted daily, after removing the feature items whose frequency is less than the minimum support, the remaining feature items are re-sorted in descending order according to the frequency of occurrence; S3.3: Initialize the root node of the frequent pattern tree, use the frequent pattern algorithm to traverse the data set again, insert the feature items sorted in descending order in S3.2 into the frequent pattern tree, and obtain a new frequent pattern tree. Each node in the new frequent pattern tree represents a feature item and the count corresponding to the feature item; S3.4: According to the new frequent pattern tree, the frequent item sets F 1 Each feature item in is traversed from bottom to top to the ancestor node of the feature item, and the feature items contained in the corresponding path are obtained to form a feature item set. The count of the end node of the path is the count of the feature item set, and the feature item set with a count greater than or equal to the minimum support is the final frequent item set; Generate association rules, S3.5: In each final frequent item set, the attributes of the feature items are judged to define association rules. The association rules are recorded as X→Y, where the former item X represents the feature item with the risk possibility factor, and the latter item Y represents the feature item with the infection factor. Each final frequent item set includes multiple association rules. The confidence of each association rule X→Y is calculated. If the confidence is greater than or equal to the minimum confidence, X→Y is a strong association rule, otherwise it is a weak association rule. S3.6: Calculate the lift for each strong association rule. If the lift is greater than or equal to 1, the strong association rule is a valid strong association rule, and the factor of the antecedent X of the strong association rule is the risk factor. S4: First analyze and sort the association rules, then analyze the risk factors to obtain the risk level of the risk factors.

2. According to the method for identifying epidemic risk factors based on association rules in claim 1, It is characterized in that In S4, the specific steps for obtaining the risk level of the risk factor are: S4.1: Conduct in-depth analysis on each association rule based on the lift, and sort the association rules from high to low according to the lift; S4.2: For the sorted association rules, analyze the degree of influence of the risk factors contained in the antecedent of the association rules on the infection factors contained in the latter according to the confidence level; the higher the confidence level, the higher the risk of the risk factor in the spread of the epidemic.

3. According to the method for identifying epidemic risk factors based on association rules in claim 1, It is characterized in that In S2.2, the feature numbers are numbered in ascending order of gender, age, number of infections, temperature, rainfall, air quality index, population in-migration index and population out-migration index.

4. According to the method for identifying epidemic risk factors based on association rules in claim 1, It is characterized in that S3.3 also includes that when the data extracted from one day is inserted, if the corresponding feature item already exists in the frequent pattern tree, the count is increased by 1; if not, a new branch is added to the frequent pattern tree.

5. According to the method for identifying epidemic risk factors based on association rules in claim 1, It is characterized in that The calculation formula for support is: Support(X→Y)=P(X∪Y),Support(X)=P(X) Where P(X) represents the proportion of X in the data set, and P(X) = Count(X) / Count(D), Count(X) represents the count of X in the data set, X represents the antecedent of the association rule; Count(D) represents the total count in the data set; The confidence calculation formula is: The calculation formula for lift is: Among them, P(X∪Y) represents the proportion of X∪Y in the data set, P(Y) represents the proportion of Y in the data set, and Y represents the latter term of the association rule.

6. According to the method for identifying epidemic risk factors based on association rules in claim 1, It is characterized in that In S3.3, the method for obtaining a new frequent pattern tree is: insert the feature items sorted in descending order in S3.2 into the frequent pattern tree from top to bottom in sequence, connected by edges, to obtain a new frequent pattern tree.

7. An epidemic risk factor identification system based on association rules, It is characterized in that It includes data extraction module, data processing module, data mining module and comprehensive analysis module. The data extraction module is used to extract data from official data sources; The official data source includes infection case data, meteorological data, and population flow data. The infection case data includes the number of infected persons, as well as the age and gender of the infected persons; the meteorological data includes temperature, rainfall, and air quality index; the population flow data includes a population in-migration index and a population out-migration index. Data is extracted from the official data source every day to construct a data set. The data extracted every day includes date, gender, age, number of infected persons, temperature, rainfall, air quality index, population in-migration index, and population out-migration index; The data processing module is used to transform the non-discrete data in the extracted data into a data form, including: The non-discrete data in the extracted data are characterized by classification; the gender, age, number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index are all characteristic items. For age, classification and characteristic value selection are performed according to age groups to obtain characteristic values; for gender, the characteristic value corresponding to male is "1", and the characteristic value corresponding to female is "2"; for the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index, the average value and standard deviation are calculated respectively, and the classification and characteristic value selection are performed according to the numerical range of the corresponding average value and standard deviation to obtain the characteristic values ​​corresponding to the number of infected persons, temperature, rainfall, air quality index, population in-migration index and population out-migration index respectively; Label the non-discrete data in the extracted data; assign a feature number to gender, age, number of infected people, temperature, rainfall, air quality index, population in-migration index, and population out-migration index respectively; Each eigenvalue is combined with the corresponding eigennumber to form a combined eigenvalue; The data mining module is used to perform data mining on the data after the data form is transformed, generate association rules, and identify risk factors; The specific steps to conduct data mining, generate association rules, and identify risk factors are: Find frequent itemsets. The frequent pattern algorithm is used to traverse the data set and sort it in descending order according to the frequency of occurrence. The feature items with a frequency less than the minimum support are eliminated to obtain the frequent item set, which is denoted as F. 1 ; For the data extracted daily, after removing the feature items whose frequency is less than the minimum support, the remaining feature items are re-sorted in descending order according to the frequency of occurrence; Initialize the root node of the frequent pattern tree, use the frequent pattern algorithm to traverse the data set again, insert the descending sorted feature items into the frequent pattern tree, and obtain a new frequent pattern tree, in which each node represents a feature item and the count corresponding to the feature item; According to the new frequent pattern tree, the frequent item sets F 1 Each feature item in is traversed from bottom to top to the ancestor node of the feature item, and the feature items contained in the corresponding path are obtained to form a feature item set. The count of the end node of the path is the count of the feature item set, and the feature item set with a count greater than or equal to the minimum support is the final frequent item set; Generate association rules, In each final frequent item set, the attributes of the feature items are judged to define the association rules, which are recorded as X→Y, where the former X represents the feature item with the risk possibility factor, and the latter Y represents the feature item with the infection factor. Each final frequent item set includes multiple association rules, and the confidence of each association rule X→Y is calculated. If the confidence is greater than or equal to the minimum confidence, X→Y is a strong association rule, otherwise it is a weak association rule; The lift is calculated for each strong association rule. If the lift is greater than or equal to 1, the strong association rule is a valid strong association rule, and the factor of the antecedent X of the strong association rule is the risk factor. The comprehensive analysis module is used to analyze and sort the association rules, and then analyze the risk factors to obtain the risk level of the risk factors.