Method for processing infectious disease early warning data based on Internet platform search

Abstract

The infectious disease early warning data processing method based on Internet platform search provided by the present invention relates to the technical field of infectious disease early warning. The method includes collecting and preprocessing Internet user search data, extracting infectious feature information, and constructing an infectious disease feature database; determining the users to be observed through matching, determining their locations, and analyzing the spread of infectious diseases; recording the statistical information of the users to be observed, integrating it into disease observation information and storing it as historical disease information; using the historical disease information to construct a training set, training local and neighboring area logistic regression models, adjusting the local prediction value with the neighboring area prediction value, and integrating to obtain a common infectious disease prediction model. This method can capture the epidemic signal before patients seek medical treatment, give early warnings in advance, cover the population who have not sought medical treatment, accurately monitor in combination with IP positioning, eliminate data blind spots, and improve the comprehensiveness and representativeness of monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of infectious disease early warning, and particularly to a method for processing infectious disease early warning data based on Internet platform searches. Background Art

[0002] The spread and prevalence of infectious diseases are one of the major challenges in the global public health field. Their outbreaks often pose a serious threat to human life and health, and at the same time bring huge negative impacts to social and economic development. Timely and accurate early warning of infectious diseases is crucial for taking effective prevention and control measures and reducing the risk of epidemic spread.

[0003] With the rapid development and popularization of Internet technology, the Internet has become an important platform for people to obtain information and communicate. In terms of infectious disease-related information, when people experience physical discomfort or pay attention to the health status around them, they often conduct relevant searches and discussions through Internet platforms such as search engines, health forums, and social media. Behind these search behaviors, there is a large amount of information related to infectious diseases. For example, a user's search for the symptoms of a certain infectious disease may imply that they or someone around them has similar symptoms; a search for preventive measures against infectious diseases may reflect the potential risk of the spread of the infectious disease in the local area, etc.

[0004] Traditional infectious disease monitoring and early warning mainly rely on the case reporting system of medical institutions. This system collects information on confirmed infectious disease cases in medical institutions such as hospitals and clinics, conducts statistical analysis and trend judgment to achieve the monitoring and early warning of infectious diseases.

[0005] However, this traditional method has many limitations. On the one hand, there is a lag in information. From the time when a patient shows symptoms, goes to a medical institution for treatment, undergoes examinations to the final diagnosis, and then to the reporting of case information to relevant departments, the whole process takes a certain amount of time, which causes a significant delay in the acquisition and analysis of epidemic information and makes it difficult to capture the early signs of the spread of infectious diseases in a timely manner. On the other hand, the coverage is limited. Traditional monitoring mainly focuses on patients who have sought medical treatment, while for those in the incubation period, with mild symptoms who have not sought medical treatment or in remote areas lacking medical resources, their infection situations are often difficult to be discovered and counted in a timely manner, resulting in incomplete monitoring data and being unable to comprehensively reflect the true spread situation of infectious diseases. Therefore, it is necessary to provide a method for processing infectious disease early warning data based on Internet platform searches to solve the above technical problems. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for processing infectious disease early warning data based on Internet platform searches, so as to make full use of the search data on the Internet platform, early warn of the outbreak and prevalence of infectious diseases, and provide strong support for public health decision-making.

[0007] To solve the above technical problems, the infectious disease early warning data processing method based on Internet platform search provided by the present invention includes the following steps:

[0008] Collect the search data of users through the Internet platform and preprocess the search data;

[0009] Extract infectious feature information from the preprocessed search data;

[0010] Construct an infectious disease feature database, including the feature information of various common infectious diseases. Each infectious disease record contains the disease name, typical symptoms, transmission routes, high-incidence seasons, and incubation periods;

[0011] Match the user's infectious feature information with the infectious disease feature database to determine whether the user is searching for an infectious disease; record the users who are determined to search for an infectious disease as users to be observed;

[0012] Obtain the IP address of the user to be observed, convert it into geographical location information through the parsing library, and determine the location area; conduct an analysis of the spread of infectious diseases in the location area to obtain the disease search value of the location area;

[0013] Identify the location area of the region, and record the growth volume and growth rate of the users to be observed in the location area as statistical information; integrate the statistical information, infectious feature information, and disease search value of the users to be observed in the corresponding region of the area into disease observation information and store it to form historical disease information;

[0014] Use the historical disease information corresponding to the area to construct a training data set, train a logistic regression model, and use the gradient descent method to minimize the loss function; at the same time, obtain the historical disease information of the neighboring area to train the neighboring area logistic regression model;

[0015] At the same time, obtain the historical disease information of the neighboring area corresponding to the area and retrain the neighboring area logistic regression model; input the disease observation information into the local and neighboring area logistic regression models respectively to obtain the corresponding predicted infection values; use the predicted infection values of the neighboring area to adaptively adjust and update the local predicted infection values; then integrate all the logistic regression models of infectious diseases to obtain a common infectious disease prediction model.

[0016] Preferably, for the preprocessing of the search data, the preprocessing operations include removing duplicate, invalid data, desensitization processing, and standardizing the search keywords to unify synonyms and near-synonyms; the specific steps of its preprocessing operations include:

[0017] Removing duplicate data: Denote all the search data of users as the search data set , and obtain the hash value set after hash encoding , and obtain a new search data set after removing the data with the same hash value ;

[0018] Removing invalid data: Define the rules for invalid data, including meaningless characters and too short length. For each search data , check whether it conforms to the invalid data rules. If it conforms, remove it from the search data set to obtain a data set ;

[0019] Desensitization processing: Identify sensitive information in the search data, use regular expressions to match sensitive information, and replace sensitive information with specific characters; Sensitive information includes user names, ID numbers, and contact information;

[0020] Keyword standardization: Establish a dictionary of synonyms and near-synonyms, and uniformly replace synonyms and near-synonyms in the search keywords with standard words; Denote the keyword as k and the dictionary as S. If k has synonyms or near-synonyms in S, then replace k with .

[0021] Preferably, extract infectious feature information from the preprocessed search data. The infectious feature information includes the name of the disease, symptom keywords, transmission route keywords, and high-incidence season keywords; The specific extraction operations are as follows:

[0022] Use the named entity recognition algorithm to process the search data and identify the name of the disease in it. Denote the search data as d, and obtain the set of disease names after processing by the named entity recognition algorithm ;

[0023] Construct a symptom keyword dictionary , and extract symptom keywords from the search data by string matching. For each search data d, which contains the keyword s in it, then extract it as a symptom keyword to obtain the set of symptom keywords ;

[0024] Then construct a transmission route keyword dictionary , and use the method of string matching to extract transmission route keywords. The set of extracted transmission route keywords is ;

[0025] Then extract high-incidence season keywords by keyword matching and semantic analysis. Set the high-incidence season keyword dictionary as , and the set of extracted high-incidence season keywords is .

[0026] Preferably, match the infectious feature information searched by the user with the infectious disease feature database. The specific matching method is as follows:

[0027] Record the user's infection characteristic information as , and the characteristic information of one infectious disease in the infectious disease characteristic data is , where is the set of names of an infectious disease, is the set of typical symptoms of an infectious disease, is the set of transmission routes of an infectious disease, is the set of information on the high-incidence seasons of an infectious disease; calculate the matching degree M, and the formula is: , where , , , are weight coefficients, and + + + = 1. The weight coefficients are determined by expert experience, historical data training or machine learning algorithms. When the matching degree M is greater than the set search matching threshold , it is determined that the user searches for an infectious disease, and the user is recorded as a user to be observed.

[0028] Preferably, perform an analysis of the spread of infectious diseases in the area where the user is located. The specific method is as follows:

[0029] Denote the neighborhood of the area R where the user is located as , within the time range , count the number of users searching for this infectious disease in the neighborhood and record it as the total number of disease searches. Divide the time range into several time partitions, and count the number of users searching for this infectious disease in the time partition and record it as the number of disease searches sf in the partition. Calculate the disease search trend value for the number of disease searches sf in the time partition in the time range using the standard deviation formula. The standard deviation formula is: , where sf is the number of disease searches sf in each time partition f, represents the average value of the number of disease searches, and r is the number of time partitions; perform a weighted process on the total number of disease searches and the disease search trend value to obtain the disease search value; if the disease search value is greater than the set spread threshold, it is considered that there is a risk of spread of the infectious disease.

[0030] Preferably, the formula of the logistic regression model is:

[0031] , where represents the probability of the occurrence of this infectious disease in this area under the feature vector , and w0, w1,..., wn represent the model parameters.

[0032] Preferably, use the predicted infection value of the adjacent area to adaptively adjust and update the predicted infection value of the local area. Specifically:

[0033] Set the adjacent area adjustment coefficient , use the formula , where represents the predicted infection value of the local model origin, is the predicted infection value of the adjacent area model;

[0034] Use the predicted infection value of the adjacent area combined with the adjacent area adjustment coefficient Adaptively adjust and update the predicted infection value of the local area to obtain .

[0035] Preferably, after generating the predicted infection value of the local area, set several color mapping intervals, and match the predicted infection value of the local area with the color mapping intervals to obtain the corresponding mapping color of the area;

[0036] And establish a two-dimensional display image. The two-dimensional display image takes the geographical area as the horizontal axis and time as the vertical axis, and displays the mapping colors matched by the infectious disease prediction values of each area at different times at the corresponding positions in the two-dimensional display image to form an intuitive spatio-temporal distribution map of infectious disease risks.

[0037] Preferably, it also includes regularly updating the two-dimensional display image; obtaining the new predicted infection value of the local area according to the set update interval, and timely adjusting the mapping color display of each area in the two-dimensional display image according to the new predicted infection value of the local area.

[0038] Compared with the related technology, the infectious disease early warning data processing method provided by the present invention has the following beneficial effects:

[0039] 1. By collecting and analyzing the search data of Internet platform users in real time, the present invention can capture potential epidemic signals before patients seek medical treatment. Compared with the traditional passive monitoring mode that relies on medical institutions to report, this method significantly advances the early warning time, especially suitable for early risk identification in the incubation period or the initial stage of symptoms, and strives for a critical window period for prevention and control decisions; at the same time, relying on the universality of Internet search data, it can cover the unmedicalized population, break through the dependence on medical resources in traditional monitoring, and combine IP address parsing technology to accurately locate potential epidemic areas, achieve full-domain dynamic monitoring, eliminate the data blind spots of traditional methods, and improve the comprehensiveness and representativeness of monitoring results.

[0040] 2. Based on the standardized matching algorithm of multi-dimensional infection feature information, combined with the logical regression model and the neighborhood data collaborative training mechanism, the present invention effectively integrates the local and neighborhood epidemic trends, dynamically adjusts the predicted infection value, quantifies the regional transmission situation through disease investigation value calculation and diffusion risk analysis, reduces the deviation of a single data source, and significantly improves the scientificity and reliability of the early warning results.

[0041] 3. Through the spatio-temporal distribution map of infectious disease risks, the present invention maps the predicted infection values into intuitive color identifications, and displays the spatio-temporal evolution trend of infectious disease risks in real time. Combined with a regular update mechanism, it provides a dynamic and visual decision-making basis for public health departments, facilitating the accurate implementation of resource allocation and prevention and control measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of a method for processing infectious disease early warning data based on Internet platform search provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0044] The terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit this disclosure. The singular forms of "group", "class" and "the" used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0045] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to a determination".

[0046] Please refer to Figure 1 An infectious disease early warning data processing method based on Internet platform search includes the following steps:

[0047] Collect the search data of users through the Internet platform and preprocess the search data; the preprocessing operations include removing duplicate and invalid data, desensitization processing, and standardizing the search keywords to unify synonyms and near-synonyms; it should be noted that the Internet platform specifically includes but is not limited to search engine websites, health forums, social media, etc.;

[0048] Extract infectious feature information from the preprocessed search data; the infectious feature information includes the name of the disease, symptom keywords (such as fever, cough, diarrhea, etc.), transmission route keywords (such as airborne transmission, contact transmission, etc.), and high-incidence season keywords;

[0049] Construct an infectious disease feature database, including the feature information of various common infectious diseases. Each infectious disease record contains the name of the disease, typical symptoms, transmission routes, high-incidence seasons, and incubation periods;

[0050] Match the user's infectious feature information with the infectious disease feature database to determine whether the user has determined to search for an infectious disease; mark the users who have determined to search for an infectious disease as users to be observed;

[0051] Obtain the IP address of the user to be observed to determine the region where the Internet search is located. Convert the IP address into geographical location information through an IP address resolution library to determine the region R;

[0052] Conduct an analysis of the spread of infectious diseases in the region R to obtain the disease search value of the region;

[0053] Identify the region where the region is located, and record the statistical information of all users to be observed in this region, including the growth amount of users to be observed (the growth amount is the difference in the number of users to be observed in adjacent statistical periods) and the growth rate (the growth amount is the ratio of the number of users to be observed in the current period to the number of users to be observed in the previous period);

[0054] Integrate the statistical information, infectious feature information, and disease search value of the users to be observed in the region into the disease observation information corresponding to the region; store the disease observation information, and all the stored disease information is recorded as historical disease information;

[0055] Use the historical disease information corresponding to the region to construct a training data set. By training a logistic regression model, obtain the model parameters; use the gradient descent method to train the model parameters to minimize the loss function of the model on the training data set. The loss function formula is ; where m is the number of samples, yi is the actual label, and pi is the predicted probability;

[0056] At the same time, obtain the historical disease information of the regions corresponding to the neighborhood of the region and retrain the logistic regression model of the neighboring regions;

[0057] Input the disease observation information into the logistic regression models of the local and neighboring regions respectively to obtain the corresponding predicted infection values; use the predicted infection values of the neighboring regions to adaptively adjust and update the predicted infection values of the local region; integrate the logistic regression models of all types of infectious diseases to obtain a common infectious disease prediction model.

[0058] In this application, the specific steps of the preprocessing search data include:

[0059] Removing duplicate data: Denote all the search data of the user as the search data set , and obtain the hash value set after hash encoding . After removing the data with the same hash value, obtain the new search data set ;

[0060] Removing invalid data: Define the rules for invalid data, including meaningless characters, too short length, etc. For each search data , check whether it conforms to the invalid data rules. If it does, remove it from the search data set to obtain the data set ;

[0061] Desensitization processing: Identify the sensitive information in the search data, use regular expressions to match the sensitive information, and replace the sensitive information with specific characters; the sensitive information includes user name, ID number, contact information, etc.;

[0062] Keyword standardization: Establish a synonym and near-synonym dictionary, and uniformly replace the synonyms and near-synonyms in the search keywords with standard words; Denote the keyword as k and the dictionary as S. If k has synonyms or near-synonyms in S, then replace k with .

[0063] In this application, extract the infectious feature information from the preprocessed search data, specifically including:

[0064] Use the named entity recognition (NER) algorithm (such as the BERT - BiLSTM - CRF model based on deep learning) to process the search data and identify the disease names in it. Denote the search data as d, and obtain the disease name set after being processed by the NER algorithm ;

[0065] Construct a symptom keyword dictionary , and extract the symptom keywords from the search data by string matching. For each search data d, which contains the keyword s in it, then extract it as the symptom keyword to obtain the symptom keyword set ;

[0066] Then construct a transmission route keyword dictionary , and use the method of string matching to extract the transmission route keywords. The set of the extracted transmission route keywords is ;

[0067] Then extract the high-incidence season keywords by keyword matching and semantic analysis methods. Set the high-incidence season keyword dictionary as , the obtained set of keywords for the high-incidence season is .

[0068] In this application, the infectious disease characteristic information searched by the user is matched with the infectious disease characteristic database. The specific matching method is as follows:

[0069] Record the infectious disease characteristic information of the user as , and the characteristic information of one infectious disease in the infectious disease characteristic data is , where is the set of disease names of an infectious disease, is the set of typical symptoms of an infectious disease, is the set of transmission routes of an infectious disease, is the set of high-incidence season information of an infectious disease; calculate the matching degree M, and the formula is: , where , , , are weight coefficients, and + + + = 1. The weight coefficients are determined by expert experience, historical data training or machine learning algorithms. When the matching degree M is greater than the set search matching threshold , it is determined that the user is searching for an infectious disease, and the user is recorded as a user to be observed.

[0070] In this application, the analysis of the spread of infectious diseases in the area R is carried out. The specific method is as follows:

[0071] Record the neighborhood of the area R as , within the time range , count the number of users searching for this infectious disease in the neighborhood as the total number of disease searches. Divide the time range into several time partitions, and count the number of users searching for this infectious disease in the time partition as the number of disease searches sf in the partition. Calculate the disease search trend value for the number of disease searches sf in the time partition in the time range using the standard deviation formula. The standard deviation formula is: , where sf is the number of disease searches sf in each time partition f, represents the average value of the number of disease searches, and r is the number of time partitions; perform weighted processing on the total number of disease searches and the disease search trend value to obtain the disease search value; if the disease search value is greater than the set spread threshold, it is considered that there is a risk of infectious disease spread.

[0072] In this application, the formula of the logistic regression model is:

[0073] , where represents at the feature vector The probability of the occurrence of the infectious disease in the area, where \(w_0, w_1, \cdots, w_n\) represent the model parameters.

[0074] In this application, the predicted infection value of the local area is adaptively adjusted and updated by using the predicted infection value of the adjacent areas. Specifically:

[0075] Set the adjacent area adjustment coefficient , and use the formula , where represents the original predicted infection value of the local model, is the predicted infection value of the adjacent area model;

[0076] Use the predicted infection value of the adjacent area combined with the adjacent area adjustment coefficient to adaptively adjust and update the predicted infection value of the local area to obtain .

[0077] In this application, after generating the predicted infection value of the local area, several color mapping intervals are set, and the predicted infection value of the local area is matched with the color mapping intervals to obtain the corresponding mapping color of the area;

[0078] And a two-dimensional display image is established. The two-dimensional display image takes the geographical area as the horizontal axis and the time as the vertical axis, and displays the mapping colors corresponding to the predicted infection values of each area at different times at the corresponding positions in the two-dimensional display image, forming an intuitive spatio-temporal distribution map of the infectious disease risk.

[0079] In this application, it also includes regularly updating the two-dimensional display image; obtaining the new predicted infection value of the local area according to the set update interval, and timely adjusting the mapping color display of each area in the two-dimensional display image according to the new predicted infection value of the local area.

[0080] Those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include known common knowledge or conventional technical means in the technical field not disclosed in this disclosure. The specification and examples are only regarded as exemplary, and the true scope and spirit of the present invention are pointed out by the following claims.

[0081] It should be understood that the present invention is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A method for processing infectious disease early warning data based on Internet platform search, characterized in that, It includes the following steps: Collect the search data of users through the Internet platform and preprocess the search data; Extract the infectious feature information from the preprocessed search data; Construct an infectious disease feature database, including the feature information of various common infectious diseases. Each infectious disease record contains the disease name, typical symptoms, transmission routes, high-incidence seasons, and incubation periods; Match the infectious feature information of the user with the infectious disease feature database to determine whether the user is searching for an infectious disease; mark the users who are determined to search for an infectious disease as users to be observed; Obtain the IP address of the user to be observed, convert it into geographical location information through the parsing library, and determine the location area; conduct an analysis of the spread of infectious diseases in the location area to obtain the disease search value of the location area. The specific method is as follows: Denote the neighborhood of the area R as , within the time region , count the number of users searching for the infectious disease in the neighborhood and denote it as the total disease search volume. Divide the time region into several time partitions, and count the number of users searching for the infectious disease within the time partition and denote it as the disease search number sf in the partition. Calculate the disease search trend value for the disease search number sf in the time partition in the time region using the standard deviation formula. The standard deviation formula is: , where sf is the disease search number in each time partition f, represents the average value of the disease search numbers, and r is the number of time partitions; perform weighted processing on the total disease search volume and the disease search trend value to obtain the disease search value; if the disease search value is greater than the set diffusion threshold, it is considered that there is a risk of spread of the infectious disease; Identify the location area of the region, and record the growth amount and growth rate of the users to be observed in the location area as statistical information; integrate the statistical information, infectious feature information, and disease search value of the users to be observed corresponding to the region into the disease observation information and store it to form historical disease information; Use the historical disease information corresponding to the region to construct a training data set, train a logistic regression model, and use the gradient descent method to minimize the loss function; at the same time, obtain the historical disease information of the neighboring area to train the neighboring area logistic regression model; At the same time, obtain the historical disease information of the neighboring area corresponding to the region and train the neighboring area logistic regression model again; input the disease observation information into the local and neighboring area logistic regression models respectively to obtain the corresponding predicted infection values; Use the predicted infection value of the neighboring area to adaptively adjust and update the predicted infection value of the local area; then integrate all the logistic regression models of infectious diseases to obtain a common infectious disease prediction model.

2. The method for processing infectious disease early warning data based on Internet platform search according to claim 1, wherein For the preprocessing of the search data, the preprocessing operations include removing duplicate and invalid data, desensitization processing, and standardizing the search keywords to unify synonyms and near-synonyms; The specific steps of its preprocessing operation include: Removing duplicate data: Denote all the search data of the user as the search data set , and obtain the hash value set after hash encoding , and obtain a new search data set after removing the data with the same hash value ; Removing invalid data: Define the rules for invalid data, including meaningless characters and too short length. For each search data , check whether it conforms to the invalid data rules. If it does, remove it from the search data set to obtain a data set ; Desensitization processing: Identify the sensitive information in the search data, use regular expressions to match the sensitive information, and replace the sensitive information with specific characters; the sensitive information includes user names, ID numbers, and contact information; Keyword standardization: Establish a thesaurus of synonyms and near-synonyms, and uniformly replace the synonyms and near-synonyms in the search keywords with standard words; Denote the keyword as k and the thesaurus as S. If k has synonyms or near-synonyms in S, then replace k with .

3. The method for processing infectious disease early warning data based on Internet platform search according to claim 1, wherein Extract the infectious feature information from the preprocessed search data. The infectious feature information includes the name of the infectious disease, symptom keywords, transmission route keywords, and high-incidence season keywords. The specific extraction operations specifically include: Process the search data using a named entity recognition algorithm to identify the names of diseases in it. Denote the search data as d, and obtain the set of disease names after processing by the named entity recognition algorithm ; Construct a symptom keyword dictionary , extract symptom keywords from the search data by string matching. For each search data d, which contains the keyword s in, it will be extracted as a symptom keyword to obtain the symptom keyword set ; Reconstruct the keyword dictionary for the dissemination channels , and use the string matching method to extract the keywords for the dissemination channels. The set of the extracted keywords for the dissemination channels is ; Then, high-incidence season keywords are extracted through keyword matching and semantic analysis methods, and the high-incidence season keyword dictionary is set as , and the set of high-incidence season keywords obtained by extraction is .

4. The method for processing infectious disease early warning data based on Internet platform search according to claim 3, wherein Match the infectious feature information searched by the user with the infectious disease feature database. The specific matching method is as follows: Record the user's infection characteristic information as , and the characteristic information of one infectious disease in the infectious disease characteristic data is , where is the set of names of an infectious disease, is the set of typical symptoms of an infectious disease, is the set of transmission routes of an infectious disease, is the set of information on the high-incidence seasons of an infectious disease; Calculate the matching degree M, and the formula is: , where , , , are weight coefficients, and + + + = 1. The weight coefficients are determined by expert experience, historical data training or machine learning algorithms. When the matching degree M is greater than the set search matching threshold , it is determined that the user has determined to search for infectious diseases, and the user is recorded as an observed user.

5. The method for processing infectious disease early warning data based on Internet platform search according to claim 1, wherein The formula of the logistic regression model is: , where represents the probability of the occurrence of this infectious disease in this region under the feature vector , and w0, w1,..., wn represent the model parameters.

6. The method for processing infectious disease early warning data based on Internet platform search according to claim 1, wherein Use the predicted infection value of the neighboring area to adaptively adjust and update the predicted infection value of the local area. Specifically: Set the adjacent land adjustment coefficient , and use the formula , where represents the predicted infection value of the local model, is the predicted infection value of the adjacent land model; Using the predicted infection value of adjacent land combined with the adjacent land adjustment coefficient Adaptively adjust and update the predicted infection value of the local area to obtain .

7. The method for processing infectious disease early warning data based on Internet platform search according to claim 1, wherein After generating the predicted infection value of the local area, set several color mapping intervals, match the predicted infection value of the local area with the color mapping intervals to obtain the mapped color corresponding to the region; And establish a two-dimensional display image. The two-dimensional display image takes the geographical area as the horizontal axis and time as the vertical axis, and displays the mapped colors corresponding to the infectious disease prediction values of each region at different times at the corresponding positions in the two-dimensional display image to form an intuitive spatio-temporal distribution map of infectious disease risks.

8. The method for processing infectious disease early warning data based on Internet platform search according to claim 7, wherein It also includes regularly updating the two-dimensional display image; obtaining a new local predicted infection value according to a set update interval, and timely adjusting the mapped color display of each region in the two-dimensional display image according to the new local predicted infection value.

Images