Medical event hierarchical early warning method, device and equipment and storage medium

Abstract

The application discloses a medical event hierarchical early warning method and device, equipment and a storage medium, relates to the technical field of public health monitoring and data mining, and comprises the following steps: acquiring self-media webpage information and historical monitoring information; performing keyword matching on corresponding medical articles based on the self-media webpage information, determining medical article matching quantity information and corresponding medical article total quantity information; determining medical event frequency information based on the medical article matching quantity information and the medical article total quantity information; determining time lag parameter verification information based on the medical event frequency information and the historical monitoring information; predicting medical event activity parameters based on the medical event frequency information, the historical monitoring information and the time lag parameter verification information, and determining a medical event activity prediction result; and controlling a system to perform hierarchical early warning based on the medical event activity prediction result. According to the medical event frequency in public opinion, the application can accurately predict the diffusion degree of a medical event in a real environment to realize hierarchical early warning.

Description

Technical Field

[0001] This application relates to the field of public health monitoring and data mining technology, and in particular to methods, devices, equipment and storage media for graded early warning of medical events. Background Technology

[0002] With the acceleration of globalization and the increasing frequency of population movement, the risk of the spread of emerging infectious diseases has increased significantly. In order to achieve the prevention and control goals of early detection, early warning and early treatment, it is necessary to make up for the inherent time lag in the data collection, reporting and summarization of traditional passive reporting systems, so as to maintain stable predictive capabilities in the context of complex epidemics.

[0003] Currently, the existing practice of infectious disease surveillance involves monitoring public search behavior for disease-related keywords to infer epidemic activity trends, identifying posts containing disease symptoms or keywords, and analyzing their posting frequency and geographical distribution to achieve epidemic tracking and early warning. However, this existing approach is susceptible to media hype and public panic. Furthermore, social media content is highly informal, colloquial, and contains a large amount of irrelevant information, resulting in high noise levels and leading to overestimation of the epidemic's scale and poor stability. Therefore, accurately predicting the spread of medical events in the real-world environment to achieve tiered early warning has become an urgent problem to be solved. Summary of the Invention

[0004] The main objective of this application is to provide a method, device, equipment, and storage medium for graded early warning of medical events, aiming to solve the technical problem of how to accurately predict the spread of medical events in the real environment in order to achieve graded early warning.

[0005] To achieve the above objectives, this application proposes a method for graded early warning of medical events, the method comprising: Obtain information from self-media web pages and historical monitoring data; Based on the information from the self-media webpage, the corresponding medical articles are matched with keywords to determine the number of matched medical articles and the total number of corresponding medical articles. The frequency information of medical events is determined based on the number of matched medical articles and the total number of medical articles. Based on the medical event frequency information and the historical monitoring information, time lag parameter verification information is determined; Based on the frequency information of medical events, the historical monitoring information, and the time lag parameter verification information, the parameters of medical event activities are predicted to determine the prediction results of medical event activities. The system implements tiered early warning based on the predicted results of the medical event activities.

[0006] In one embodiment, the step of performing keyword matching on corresponding medical articles based on the self-media webpage information to determine the number of matched medical articles and the total number of corresponding medical articles includes: Data cleaning is performed on the article titles, body content, publication time, publishing platform, and author information in the aforementioned self-media web page information to determine the data cleaning information set; Based on the data cleaning information set, the corresponding medical articles are matched with the preset keyword library to obtain the keyword matching results; Based on the keyword matching results, the number of matching medical articles and the total number of corresponding medical articles are determined.

[0007] In one embodiment, the step of determining the number of matched medical articles and the corresponding total number of medical articles based on the keyword matching results includes: Obtain statistical period information; Based on the keyword matching results, the corresponding medical articles are filtered for interference to determine the medical article interference filtering information. Based on the statistical period information, the number of medical article interference filtering information is counted to determine the number of matching medical articles and the corresponding total number of medical articles.

[0008] In one embodiment, the step of determining the frequency information of medical events based on the number of matched medical articles and the total number of medical articles includes: The ratio information is determined based on the number of matched medical articles and the total number of medical articles. Logarithmic transformation is performed on the ratio information to obtain the frequency information of medical events.

[0009] In one embodiment, the step of determining the time lag parameter verification information based on the medical event frequency information and the historical monitoring information includes: Obtain information on combinations of time lag orders; The historical monitoring information is proportionally transformed to obtain the target variable for prediction; Based on the frequency information of the medical events, the target variable for prediction is fitted to obtain the fitting result and the corresponding autoregressive exogenous model. Based on the time lag order combination information, the prediction error of the autoregressive exogenous model is verified using a preset cross-validation strategy to determine the verification result of the autoregressive exogenous model. Based on the validation results of the autoregressive exogenous model, the information on the combination of time lag orders is filtered to obtain the validation information of the time lag parameters.

[0010] In one embodiment, the step of predicting medical event activity parameters based on the medical event frequency information, the historical monitoring information, and the time lag parameter verification information, and determining the medical event activity prediction result, includes: The autoregressive exogenous model parameter information is determined based on the medical event frequency information, the historical monitoring information, and the time lag parameter validation information. Based on the parameter information of the autoregressive exogenous model, the parameters of medical event activities are predicted to obtain the prediction results of medical event activities.

[0011] In one embodiment, the step of implementing graded early warning based on the medical event activity prediction result control system includes: Obtain historical average level information; When the predicted result of the medical event exceeds a first preset range of the historical average level information, the control system issues a level one warning. When the predicted result of the medical event activity exceeds a second preset interval of the historical average level information, the control system issues a level-two warning. When the predicted result of the medical event exceeds the third preset interval of the historical average level information, the control system issues a level three warning.

[0012] Furthermore, to achieve the above objectives, this application also proposes a medical event classification and early warning device, which includes: The acquisition module is used to acquire information from self-media web pages and historical monitoring information; The processing module is used to perform keyword matching on the corresponding medical articles based on the information of the self-media webpage, and to determine the number of medical articles matched and the total number of corresponding medical articles. The processing module is also used to determine the frequency information of medical events based on the number of medical article matches and the total number of medical articles. The execution module is used to determine time lag parameter verification information based on the medical event frequency information and the historical monitoring information; The execution module is also used to predict the medical event activity parameters based on the medical event frequency information, the historical monitoring information and the time lag parameter verification information, and to determine the medical event activity prediction result; The execution module is also used to control the system for tiered early warning based on the predicted results of the medical event activities.

[0013] In addition, to achieve the above objectives, this application also proposes a medical event classification and early warning device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the medical event classification and early warning method described above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the medical event classification and early warning method described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: This embodiment proposes a graded early warning method for medical events, which involves acquiring social media webpage information and historical monitoring information; performing keyword matching on corresponding medical articles based on the social media webpage information to determine the number of matched medical articles and the total number of corresponding medical articles; determining the frequency of medical events based on the number of matched medical articles and the total number of medical articles; determining time lag parameter verification information based on the frequency of medical events and the historical monitoring information; predicting the activity parameters of medical events based on the frequency of medical events, the historical monitoring information, and the time lag parameter verification information to determine the prediction results of medical event activities; and controlling the system to perform graded early warning based on the prediction results of medical event activities. This application obtains real-time, low-noise public opinion data sources by acquiring information from self-media web pages and historical monitoring information. It then performs keyword matching on corresponding medical articles based on the self-media web page information to obtain the number of matched medical articles and the total number of medical articles. This allows for the rapid filtering of effective samples related to the target medical event from massive amounts of text, significantly reducing interference from irrelevant information. Furthermore, the frequency information of the medical event can be obtained from the number of matched medical articles and the total number of medical articles, accurately representing the activity level of the medical event in public opinion. This allows for the determination of time lag parameter verification information, achieving dynamic time delay alignment between public opinion signals and the actual epidemic situation. It also allows for the prediction of medical event activity parameters to obtain prediction results, enabling early prediction of the spread trend of medical events, control system implementation of tiered early warning, and differentiated responses under different risk levels. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the medical event classification and early warning method of this application. Figure 2 This is a schematic diagram of the technical route for the medical event classification and early warning method in this application; Figure 3 This is a flowchart illustrating Embodiment 2 of the medical event classification and early warning method of this application; Figure 4 A simplified flowchart illustrating the medical event classification and early warning method provided in this application embodiment; Figure 5 This is a schematic diagram of the module structure of the medical event graded early warning device according to an embodiment of this application; Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the medical event classification and early warning method in the embodiments of this application.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application embodiment is as follows: Obtain self-media webpage information and historical monitoring information; perform keyword matching on corresponding medical articles based on the self-media webpage information to determine the number of matched medical articles and the total number of corresponding medical articles; determine the frequency information of medical events based on the number of matched medical articles and the total number of medical articles; determine time lag parameter verification information based on the frequency information of medical events and the historical monitoring information; predict medical event activity parameters based on the frequency information of medical events, the historical monitoring information, and the time lag parameter verification information to determine the prediction result of medical event activities; and control the system for graded early warning based on the prediction result of medical event activities.

[0023] In this embodiment, for ease of description, the following description will focus on the medical event classification and early warning device as the implementing entity.

[0024] Because existing technologies are easily influenced by media hype and public panic, and because social media content is highly informal, colloquial, and contains a lot of irrelevant information with high noise levels, the scale of the epidemic has been overestimated and its stability is poor.

[0025] This application provides a solution that obtains real-time, low-noise public opinion data sources by acquiring information from self-media web pages and historical monitoring information. It then performs keyword matching on corresponding medical articles based on the self-media web page information to obtain the number of matched medical articles and the total number of medical articles. This allows for rapid filtering of effective samples related to the target medical event from massive amounts of text, significantly reducing interference from irrelevant information. Furthermore, the frequency information of the medical event can be obtained from the number of matched medical articles and the total number of medical articles, accurately representing the activity level of the medical event in public opinion. This allows for the determination of time lag parameter verification information, achieving dynamic time delay alignment between public opinion signals and the actual epidemic situation. It also allows for prediction of medical event activity parameters to obtain prediction results, enabling early prediction of the spread trend of medical events and control systems to implement tiered early warning, achieving differentiated responses under different risk levels.

[0026] Based on this, embodiments of this application provide a method for graded early warning of medical events, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the medical event classification and early warning method of this application.

[0027] In this embodiment, the medical event classification and early warning method includes steps S10 to S40: Step S10: Obtain self-media webpage information and historical monitoring information; It should be noted that the self-media webpage information is raw digital content containing potential public health-related topics collected in real time from Internet self-media platforms. That is, it is a data set consisting of multiple data dimensions such as article title, body content, publication time, publishing platform and author information, which is used to filter effective text related to medical events. The medical events are the occurrence, spread and activity process of epidemic infectious diseases, such as influenza, hand-foot-mouth disease, dengue fever, etc.

[0028] It is understood that the historical monitoring information is authoritative historical epidemic statistics regularly released by official health agencies. It is typically expressed as the percentage of influenza-like cases on a weekly or daily basis, that is, the proportion of patients who visit medical institutions and have influenza-like symptoms. It can characterize the proportion of patients who visit medical institutions and have influenza-like symptoms within a specific time period, and measure the historical activity intensity and trend of medical events.

[0029] In a specific embodiment, information from self-media web pages can be obtained through a distributed web crawler system deployed on a cloud server. However, when collecting self-media data, legal crawling is required. Legal crawling emphasizes adhering to the principles of "authorization priority, reasonable scope, compliant methods, and moderate use." For example, the target website's communication protocol should be followed before collection, and API call authorization should be obtained primarily through official channels. The collection scope should be limited to publicly visible information, avoiding the crawling of backend data that requires login or is explicitly stated as non-collectible. A reasonable request frequency should be controlled to simulate normal user access. During data usage, it should only be used for internal analysis, scientific research, or public health monitoring—non-commercial purposes. The original or processed data must not be used for resale, unfair competition, or infringement of others' legitimate rights. In this case, based on a preset Uniform Resource Locator (URL) and collection strategy, the public web pages of multiple self-media platforms (such as WeChat Official Accounts, Toutiao, Sohu, Baijiahao, etc.) can be accessed periodically (e.g., once a day). The crawler program parses the HTML (Hypertext Markup) of the web pages. The language (Hypertext Markup Language) structure can accurately locate and extract nodes containing information such as article title, body content, publication time, publishing platform, and author. The extracted data is encapsulated into a JSON format message, thus obtaining the information of the self-media webpage.

[0030] When obtaining historical monitoring information, a secure connection can be established with the data release system of the CDC through the network or a preset database interface. According to the preset time period, such as weekly, historical weekly infectious disease monitoring data, such as ILI%, can be automatically retrieved or received from the official data interface. To ensure data quality, outlier detection will be performed after receiving the raw data, such as checking whether the data is within a reasonable range. For example, ILI% should be between 0 and 100. Missing values ​​will be marked or preliminarily imputed to obtain historical monitoring information.

[0031] Step S20: Based on the information of the self-media webpage, perform keyword matching on the corresponding medical articles to determine the number of matched medical articles and the total number of corresponding medical articles; It is understandable that medical articles are text entries extracted from information on social media websites that are related to human health, diseases, symptoms, medical practices, or public health events. However, this does not refer to all medical articles, but specifically to articles whose titles or texts contain keywords related to infectious diseases after initial cleaning. These articles can be a complete tweet, a health science article, or a disease discussion response. Because social media authors often refer to official information or their own experiences when writing, their language is relatively standardized, and their publishing behavior is temporally related to the occurrence, development, and public attention of diseases in the real world.

[0032] Additionally, it should be noted that keyword matching is a rule-based or statistical text filtering technique used to quickly identify potentially relevant entries for a target infectious disease from a vast amount of medical articles. It can be combined with natural language processing techniques such as word segmentation, synonym expansion, and negative word filtering to improve the accuracy and recall of the matching. For example, for the word "fever," keyword matching can simultaneously identify synonyms or near-synonyms such as "fever," "high fever," and "elevated body temperature," while excluding negative contexts such as "no fever" and "no fever."

[0033] In a specific embodiment, the article titles, main content, publication time, publishing platform, and author information in the self-media web page information are cleaned to determine a cleaned data information set. Based on the cleaned data information set, the corresponding medical articles are matched with a preset keyword library to obtain keyword matching results. Based on the keyword matching results, the number of matched medical articles and the total number of corresponding medical articles are determined. That is, regular expressions can be used to remove irrelevant content such as HTML tags, special symbols, emoticons, and URL links from the articles. The SimHash algorithm is used to calculate the binary fingerprint of each article. Articles with a similarity greater than 85% are considered duplicates and are removed. Articles with too few words (such as less than 100 words) are filtered out to exclude invalid information. At the same time, advertising and promotional articles are removed by keywords (such as "advertisement", "promotion", "click to receive", etc.) to form a cleaned data information set, thereby significantly reducing noise and redundancy in the original self-media data and making the retained article content more standardized and pure. At this time, the cleaned article data can be stored in a distributed database (such as Elasticsearch) and indexed by time, platform, region, and other dimensions to facilitate retrieval and analysis and significantly improve the retrieval efficiency of keyword matching and feature extraction.

[0034] Step S30: Determine the frequency information of medical events based on the number of matched medical articles and the total number of medical articles; It should be noted that the medical event frequency information is a quantitative indicator used to characterize the relative frequency of occurrence of self-media articles related to the target infectious disease among all medical articles within a specific time period, and to characterize the real-time activity of the target infectious disease in the public opinion field.

[0035] In a specific embodiment, a ratio is determined based on the number of matched medical articles and the total number of medical articles, that is, the number of infectious disease-related articles is counted weekly (or daily). and the total number of articles collected in the same period The ratio information is calculated and expressed as follows:

[0036] Obtaining the ratio information can effectively eliminate the interference of fluctuations in the total number of self-media articles published in different time periods on the absolute number, make the public opinion heat across cycles comparable, and characterize the true proportion of infectious disease-related topics in all self-media content.

[0037] Logarithmic transformation is performed on the ratio information to obtain medical event frequency information; that is, logarithmic transformation is performed on the ratio information to calculate characteristic indicators. This refers to the frequency of medical events, represented as:

[0038] Logarithmic transformation can compress the scale of data, stabilize variance, and make the ratio information, which may originally be skewed, approach the normal distribution assumption, thereby satisfying the homoscedasticity requirement of autoregressive exogenous models for exogenous input variables.

[0039] In one feasible implementation, step S30 may include steps A11-A12: Step A11: Determine the ratio information based on the number of matched medical articles and the total number of medical articles; It should be noted that the ratio information represents the relative frequency of articles related to the target infectious disease among all collected self-media articles. It is used to eliminate the impact of fluctuations in the total number of self-media articles in different periods on the absolute number, so that the public opinion heat across cycles is comparable.

[0040] Step A12: Perform a logarithmic transformation on the ratio information to obtain medical event frequency information.

[0041] Understandably, logarithmic transformation is a mathematical mapping operation that transforms the original numerical values ​​into new values ​​by taking the logarithm (usually the natural logarithm or the common logarithm). This can compress the scale range of the data, stabilize the variance, and make the ratio information, which may originally be skewed, tend to be normally distributed. This satisfies the statistical assumption of homoscedasticity of input variables in autoregressive exogenous models. At the same time, logarithmic transformation can enhance the model's sensitivity to small ratio changes when the early signals of an epidemic are weak, thereby improving the accuracy of predictions.

[0042] Step S40: Determine time lag parameter verification information based on the medical event frequency information and the historical monitoring information; It should be noted that the time lag parameter verification information is the optimal combination of time lag orders determined through cross-validation. It is used to quantify the time lead of self-media public opinion signals relative to official epidemic statistics, and may include the time lag order of official monitoring data. Time lag order of self-media article feature indicators ,in, This indicates how many weeks of official historical data are used to predict the current week's COVID-19 activity. This indicates how many weeks of self-media article features are used to assist in prediction.

[0043] In a specific embodiment, the time lag order combination information is obtained, which is the set of all candidate value pairs to be evaluated. For example, the search range can be set based on prior knowledge and data length. The value range is set to 1 to 4 weeks. The value range is set to 0 to 4 weeks, where =0 indicates that no self-media features are introduced and only the autoregressive part is used. In this case, all possible combinations form a candidate set, totaling 4×5=20 combinations. This ensures that the optimal combination that minimizes the prediction error can be selected from multiple candidates, thereby avoiding the model underfitting or overfitting problems caused by the fixed lag order and improving the model's adaptability.

[0044] The historical monitoring information is proportionally transformed to obtain the target variable for prediction. This historical monitoring information can be historical weekly infectious disease surveillance data obtained from official health institutions, such as the percentage of epidemiologically similar cases. ,right Perform a logit transformation to obtain the target variable for prediction. , is represented as:

[0045] At this time, the target variable The official monitoring data for week t (after logit transformation) is used to make the transformed target variable approximately follow a normal distribution, satisfying the requirements of autoregressive exogenous models for homoscedasticity and linearity of the dependent variable, while enhancing the model's predictive stability and goodness of fit when the epidemic is at extremely low or high levels.

[0046] Based on the frequency information of the medical events, the prediction target variable is fitted to obtain the fitting result and the corresponding autoregressive exogenous model. Based on the combination information of time lag orders, the prediction error of the autoregressive exogenous model is verified using a preset cross-validation strategy to determine the verification result of the autoregressive exogenous model. Based on the verification result of the autoregressive exogenous model, the combination information of time lag orders is filtered to obtain the verification information of time lag parameters, that is, for each group of time lag orders (… , The time-series rolling cross-validation method is employed: historical data is divided into training and validation sets. The training set is used to fit model parameters, and the validation set is used to calculate prediction errors. The validation results of the autoregressive exogenous model are then filtered to obtain time-lag parameter validation information. For example, for historical data with a total length of T=52 weeks (one year), the initial training window can be set to 30 weeks, and the validation window to 1 week. Then, the data is rolled forward sequentially to obtain a total of 52 weeks of data. 30 = 22 validation errors. The final prediction error of each candidate combination is the average of all validation set errors (e.g., root mean square error RMSE). This yields the validation results of the autoregressive exogenous model. By iterating through all candidate combinations of the autoregressive exogenous model validation results, the average prediction error of their cross-validation is calculated. This provides validation information for the time lag parameters, enabling an objective evaluation of the generalization performance of different lag combinations on data not used in training. This effectively avoids overfitting to historical data and significantly improves the model's real-time prediction accuracy of epidemic activity trends, thus achieving precise early warning.

[0047] In one feasible implementation, step S40 may include steps B11-B15: Step B11: Obtain information on the combination of time lag orders; It should be noted that the time lag order combination information is a candidate set used to describe the values ​​of the two lag parameters in the autoregressive exogenous model, including the autoregressive lag order of official historical monitoring data and the exogenous lag order of the self-media article feature indicators.

[0048] Step B12: Perform a scaling transformation on the historical monitoring information to obtain the target variable for prediction; It should be noted that the target variable for prediction is a new variable obtained by transforming the historical weekly infectious disease monitoring data released by official health agencies through a logit scaling transformation. This transforms the original scaling data from the bounded interval (0,1) to the entire real number domain, eliminating statistical problems caused by boundary constraints and making the variable closer to a normal distribution, thereby satisfying the homoscedasticity and linearity assumptions of the autoregressive exogenous model for the dependent variable.

[0049] Step B13: Fit the predicted target variable based on the medical event frequency information to obtain the fitting result and the corresponding autoregressive exogenous model; It should be noted that the fitting results can characterize the model's interpretability of historical epidemic data, facilitate cross-validation, effectively avoid overfitting of the model to the training data, and ensure that the selected parameters have good generalization ability and stability.

[0050] Step B14: Based on the time lag order combination information, the prediction error of the autoregressive exogenous model is verified using a preset cross-validation strategy to determine the verification result of the autoregressive exogenous model. It is understandable that the prediction error of an autoregressive exogenous model is the difference between the predicted value and the true value when the parameters estimated by the model on the training set are applied to the validation set. The smaller the prediction error, the stronger the model's generalization ability on unseen data.

[0051] Step B15: Based on the validation results of the autoregressive exogenous model, the information on the combination of time lag orders is filtered to obtain the validation information of the time lag parameters.

[0052] It should be noted that the validation results of the autoregressive exogenous model are calculated for each candidate combination of time lag orders using a preset cross-validation strategy (such as rolling time series cross-validation or k-fold cross-validation) to obtain the model prediction error value (such as root mean square error). The prediction performance of each lag combination is quantified in the form of error values ​​and used to screen the optimal parameters. Usually, the combination that minimizes the prediction error is selected as the time lag parameter validation information. Alternatively, when the errors are similar, the combination with the smaller order can be selected according to the principle of simplicity.

[0053] Step S50: Based on the medical event frequency information, the historical monitoring information, and the time lag parameter verification information, predict the medical event activity parameters and determine the medical event activity prediction result; It should be noted that the predicted results of the medical event activity are numerical values ​​obtained by predicting the level of infectious disease activity at a certain point in the future (such as the next week or day). The predicted target variable output by the model can be restored to an actual interpretable proportion value through inverse transformation (such as logit inverse transformation), and a confidence interval is attached, which is equivalent to quantifying the future development trend of the epidemic.

[0054] In a specific embodiment, the autoregressive exogenous model parameter information is determined based on the medical event frequency information, the historical monitoring information, and the time lag parameter validation information; the medical event activity parameters are predicted based on the autoregressive exogenous model parameter information to obtain the medical event activity prediction result. That is, the autoregressive exogenous model parameter information is constructed based on the medical event frequency information, the historical monitoring information, and the time lag parameter validation information, wherein the autoregressive exogenous model can be represented as:

[0055] in, This is the official monitoring data for week t (after logit transformation). The metric represents the characteristic indicators of self-media articles in week t. and These represent the time lag order of official monitoring data and the characteristics of self-media articles, respectively. and Here are the parameters to be estimated in the model, and c is a constant term. This is the random error term.

[0056] The parameters to be estimated in the predicted model at this time are the parameters of the autoregressive exogenous model, which are used to predict the parameters of medical event activities and obtain the prediction results of medical event activities. These results can be directly used to judge the future trend of epidemic activities. Among them, the parameters of medical event activities are measurable indicators used to quantitatively describe the actual epidemic intensity or activity level of a specific infectious disease within a certain time period. For example, the proportion of influenza-like cases is the proportion of patients with influenza-like symptoms among the outpatients and emergency patients of medical institutions to the total number of patients during the same period, which represents the real dynamic trend of disease changes.

[0057] In one feasible implementation, step S10 may include steps C11-C12: Step C11: Determine the autoregressive exogenous model parameter information based on the medical event frequency information, the historical monitoring information, and the time lag parameter verification information; It should be noted that the parameter information of the autoregressive exogenous model is the set of parameter estimates obtained by fitting the autoregressive exogenous model with the least squares method after setting the optimal time lag parameter verification information as the model structure, using the historical weekly prediction target variable and medical event frequency information, and after setting the model structure. This fully describes the quantitative relationship between public opinion characteristics and epidemic dynamics.

[0058] Step C12: Based on the autoregressive exogenous model parameter information, predict the medical event activity parameters to obtain the medical event activity prediction results.

[0059] Understandably, medical event activity parameters are quantifiable indicators used to characterize the actual prevalence intensity of infectious diseases at a specific point in time or period. These can be routine monitoring data released by official health agencies, such as the proportion of influenza-like cases, which is the proportion of patients with influenza-like symptoms among those visiting medical institutions, characterizing the true level and trend of disease activity.

[0060] Step S60: Based on the medical event activity prediction result control system, a graded early warning is issued.

[0061] Understandably, a tiered early warning system is a risk management strategy that divides early warning signals into different levels and triggers corresponding response mechanisms. This allows for differentiated public health responses. For low-risk epidemic fluctuations, it is only recommended to strengthen monitoring; for medium-risk epidemic spreads, it is suggested to activate emergency preparedness; and for high-risk epidemic outbreaks, it is mandatory to take intervention measures, thereby achieving precise allocation and efficient utilization of public health resources.

[0062] In a specific embodiment, historical average level information can be obtained by extracting medical event activity parameters from the historical monitoring database of official health institutions in the same period in the past (such as the same week or the same month), such as the proportion of influenza-like cases, and calculating their mean or median as historical average level information. The normal fluctuation range can be determined by combining the standard deviation, avoiding the deviation caused by subjectively setting thresholds, so that the graded early warning can effectively distinguish between normal fluctuations and abnormal risks.

[0063] When the predicted result of the medical event activity exceeds the first preset interval of the historical average level information, the control system issues a first-level warning. That is, the first preset interval can be set as the predicted result exceeding the historical average level by 1.5 times but less than 2 times (i.e., between 1.5 and 2 times the historical average level). When the predicted result of the medical event activity output by the model falls into this interval, the system automatically triggers a first-level warning (such as a yellow warning), and notifies the monitoring personnel through an interface pop-up window, SMS or email to "suggest strengthening monitoring". This enables early identification of a slight abnormal rise in the epidemic, issues a warning when the risk is still under control, and buys time for public health departments to deploy monitoring resources in advance and verify the epidemic dynamics. At the same time, it avoids frequent false alarms due to the threshold being too low.

[0064] When the predicted result of the medical event activity exceeds the second preset interval of the historical average level information, the control system issues a level 2 warning. That is, the second preset interval can be set as the predicted result exceeding the historical average level by 2 times but less than 3 times (i.e., between 2 and 3 times the historical average level). When the predicted result reaches this interval, the system automatically triggers a level 2 warning (such as an orange warning), outputs "It is recommended to start an emergency response" along with a predicted trend graph and confidence interval. At this time, it indicates that the epidemic has entered a medium-risk level and substantial prevention and control actions need to be taken. Through the automated warning mechanism, the decision-making delay from risk identification to emergency response is significantly shortened, which helps to quickly allocate resources, strengthen case screening and public information dissemination, and prevent the further spread of the epidemic.

[0065] When the predicted result of the medical event exceeds the third preset interval of the historical average level, the control system issues a three-level warning. The third preset interval can be set as the predicted result exceeding three times or more of the historical average level. When the predicted result meets this condition, the system automatically triggers a three-level warning (such as a red warning), which is simultaneously pushed to the decision-making end of the CDC. It can also be linked with the emergency plan system to initiate intervention measures. At this time, in high-risk scenarios corresponding to an outbreak or large-scale epidemic, the highest level of warning prompts decision-makers to immediately take strong intervention measures (such as restricting gatherings, activating the emergency command system, etc.), effectively reducing the impact of the epidemic on public health, achieving precise prevention and control with risk classification and differentiated response, and improving the intelligence level and emergency management efficiency of the public health monitoring system.

[0066] In one feasible implementation, step S60 may include steps D11-D14: Step D11: Obtain historical average level information; It should be noted that the historical average level information is the statistical center value of medical event activity parameters (such as the proportion of influenza-like cases) extracted from the historical monitoring database of official health institutions that are in the same period (such as the same week or the same month) as the current prediction time point. It is usually expressed as the mean or median, and can be combined with the standard deviation to reflect the normal fluctuation range. It can be used as a baseline reference for graded early warning to determine whether the current prediction result significantly exceeds the normal level and avoid triggering alarms due to normal fluctuations.

[0067] Step D12: When the predicted result of the medical event activity exceeds the first preset interval of the historical average level information, the control system issues a level one warning. Understandably, Level 1 warning is the lowest risk level warning in the system, usually corresponding to a yellow warning. When the predicted result of medical event activity exceeds the first preset range of the historical average (e.g., exceeding 1.5 times but less than 2 times the historical average), the system automatically triggers Level 1 warning, prompting public health monitoring personnel to "recommend strengthening monitoring" in order to capture early signals of a mild abnormal rise in the epidemic, issue warnings while the risk is still under control, and avoid frequent false alarms due to a low threshold.

[0068] Step D13: When the predicted result of the medical event activity exceeds the second preset interval of the historical average level information, the control system issues a level two warning; Understandably, a Level 2 warning is a medium-risk warning level in the system, usually corresponding to an orange warning. When the predicted result of medical event activity exceeds the second preset interval of the historical average (e.g., more than twice but less than three times the historical average), the system automatically triggers a Level 2 warning, outputting "It is recommended to activate an emergency response" along with a predicted trend chart and confidence interval. This indicates that the epidemic has entered a medium-risk level and requires substantial prevention and control actions. It helps decision-makers to quickly allocate medical resources, strengthen case screening and public communication, and prevent the further spread of the epidemic.

[0069] Step D14: When the predicted result of the medical event activity exceeds the third preset interval of the historical average level information, the control system issues a level 3 warning.

[0070] Understandably, Level 3 warning is the highest risk level in the system, usually corresponding to a red alert. When the predicted result of a medical event exceeds the third preset interval of the historical average (e.g., reaching or exceeding three times the historical average), the system automatically triggers a Level 3 warning, which is simultaneously pushed to the decision-making end of the CDC. It can also be linked with the emergency response system to initiate intervention measures (such as restricting gatherings and activating the emergency command system). This corresponds to a high-risk scenario of an outbreak or large-scale epidemic. The highest level of warning prompts decision-makers to take strong intervention measures immediately to minimize the impact of the epidemic on public health and achieve precise prevention and control with risk grading and differentiated responses.

[0071] This embodiment proposes a graded early warning method for medical events, which involves acquiring social media webpage information and historical monitoring information; performing keyword matching on corresponding medical articles based on the social media webpage information to determine the number of matched medical articles and the total number of corresponding medical articles; determining the frequency of medical events based on the number of matched medical articles and the total number of medical articles; determining time lag parameter verification information based on the frequency of medical events and the historical monitoring information; predicting the activity parameters of medical events based on the frequency of medical events, the historical monitoring information, and the time lag parameter verification information to determine the prediction results of medical event activities; and controlling the system to perform graded early warning based on the prediction results of medical event activities. This invention addresses the technical challenge of accurately predicting the spread of medical events in real-world environments to achieve tiered early warning systems. Compared to existing technologies, this application obtains real-time, low-noise public opinion data sources by acquiring information from social media web pages and historical monitoring data. It then performs keyword matching on corresponding medical articles based on the social media web page information to obtain the number of matched medical articles and the total number of medical articles. This allows for rapid filtering of effective samples related to the target medical event from massive amounts of text, significantly reducing interference from irrelevant information. Furthermore, the frequency of medical events can be obtained from the number of matched medical articles and the total number of medical articles, accurately representing the activity level of medical events in public opinion. This allows for the determination of time lag parameters to verify information, achieving dynamic time delay alignment between public opinion signals and the actual epidemic situation. It also predicts medical event activity parameters to obtain prediction results, enabling early prediction of the spread trend of medical events and control systems to implement tiered early warning systems, achieving differentiated responses under different risk levels.

[0072] It should be understood that, such as Figure 2 As shown, Figure 2This diagram illustrates the technical roadmap of the medical event tiered early warning method proposed in this application. The method can be applied to a medical event tiered early warning system, which consists of a data acquisition layer, a feature extraction layer, a model building layer, and an output early warning layer. In the data acquisition layer, articles are collected from self-media platforms using distributed crawlers, and then deduplicated, filtered for invalidity, and standardized for data cleaning to obtain a high-quality cleaned data set. In the feature extraction layer, the cleaned articles are filtered based on a pre-defined keyword library, the proportion of relevant articles is calculated, and self-media feature indicators (i.e., medical event frequency information) are constructed. In the model building layer, the self-media feature indicators and official historical data (after proportional transformation) are input into an autoregressive exogenous model. The optimal time lag parameter is determined through cross-validation, completing model construction and parameter estimation. In the output early warning layer, the trained model is used to predict the epidemic trend, and a tiered early warning is triggered based on threshold judgments. Simultaneously, a feedback optimization loop is formed through periodic model updates, achieving full automation from data acquisition and feature extraction to model prediction and tiered early warning, significantly improving the timeliness and accuracy of monitoring infectious diseases.

[0073] Based on the first embodiment of this application, in the second embodiment of this application, the same or similar content as the first embodiment can be referred to the above description, and will not be repeated hereafter.

[0074] In this embodiment, refer to Figure 3 , Figure 3 This is a flowchart illustrating Embodiment 2 of the medical event classification and early warning method of this application. Step S20 specifically includes steps S21 to S23: Step S21: Perform data cleaning on the article title, main content, publication time, publishing platform and author information in the self-media web page information to determine the data cleaning information set; It should be noted that the data cleaning information set is a standardized data set obtained by performing a series of normalization, noise reduction, deduplication, and filtering operations on the original self-media web page information. It retains the core semantic content (such as title and body text) in the original information, while removing irrelevant noise data (such as HTML tags, advertisements, and emoticons) and unifying the data format (such as time format and encoding format). This can directly affect the accuracy and recall rate of medical article recognition.

[0075] Understandably, the article title is the main title text of a self-media article, usually located in a webpage tag or a specific title node. It highly summarizes the core content of the article and is an important basis for keyword matching. During the cleaning process, it is necessary to remove any special symbols (such as decorative symbols), extra spaces, and platform-added identifiers that may exist in the title. The body text is the main text of the self-media article, containing the author's description of the event, opinions, data, and other detailed information. It is usually located in webpage nodes and needs to remove non-text elements such as HTML tags, CSS styles, JavaScript scripts, image links, and video embedding code, while retaining paragraph structure and punctuation. Symbols are used to maintain semantic integrity. The publication time is the date and time the article was published, which determines the article's position in the time series. During cleaning, time strings of different formats need to be uniformly converted into standard timestamps (such as Unix timestamps or datetime objects). The publishing platform is the name of the self-media platform from which the article originated, used to distinguish the characteristics of different data sources. During cleaning, the platform name needs to be standardized into a unified enumeration value. The author information is the name or account ID of the article's publisher, such as "Health Science Popularization Jun" or "Official Account of the CDC," used to identify authoritative sources and help judge the credibility of the article. During cleaning, special symbols and irrelevant suffixes in the author information need to be removed.

[0076] In specific embodiments, regular expressions can be used to remove non-text content such as HTML tags, special symbols, emoticons, and URL links from self-media web page information. The SimHash algorithm is used to calculate article similarity and eliminate duplicate articles (similarity greater than 85% is considered duplicate). SimHash is a locality-sensitive hashing algorithm that can be used for large-scale text similarity detection and deduplication. It maps text of any length to a fixed-length binary fingerprint (e.g., 64 bits). Fingerprints generated by similar texts are also close in Hamming distance. At the same time, invalid information with too few words (e.g., less than 100 words) or containing obvious advertising keywords (e.g., "advertisement", "promotion", "click to receive") needs to be filtered out, thereby obtaining a cleaned data information set. Through systematic cleaning and deduplication, the noise and redundancy in the original self-media data are greatly reduced, making the retained article content more standardized and pure, and significantly improving the accuracy and efficiency of identifying articles related to infectious diseases.

[0077] Step S22: Based on the data cleaning information set, perform keyword matching between the corresponding medical articles and the preset keyword library to obtain keyword matching results; It should be noted that the keyword matching result is a record of the determination of the relevance to the target infectious disease obtained by comparing the title and body content of each medical article in the data cleaning information set with the preset keyword database one by one.

[0078] Understandably, the preset keyword library is a pre-built collection of terms or phrases closely related to the target infectious disease. It typically uses a term frequency-inverse document frequency (TF-IDF) algorithm to select high-frequency representative words from relevant historical articles, and undergoes manual review and revision by epidemiological experts to remove ambiguous or irrelevant words. It can be customized according to different infectious diseases (such as influenza, hand-foot-mouth disease, dengue fever, etc.) and supports dynamic updates and optimization during model operation.

[0079] In a specific implementation, a keyword library covering terms related to the target infectious disease (such as "fever," "cough," "vaccine," and "infection") can be pre-constructed using a term frequency-inverse document frequency (TF-IDF) algorithm combined with epidemiological expert review. For each article in the data cleansing information set, the title and body text are segmented and matched with keywords. If at least one keyword is matched, it is marked as an "infectious disease related article." Semantic filtering rules are used to filter the text, such as excluding confusing words that clearly belong to other similar epidemics, thereby obtaining keyword matching results. This enables the rapid and automatic screening of articles related to the target infectious disease from massive amounts of self-media text, significantly improving processing speed compared to purely manual review. Furthermore, through dynamic optimization of the keyword library and semantic filtering, the false positive and false negative rates are effectively reduced.

[0080] Step S23: Determine the number of matched medical articles and the total number of corresponding medical articles based on the keyword matching results.

[0081] It should be noted that the medical article matching quantity information is the number of medical articles that are truly related to the target infectious disease within a preset statistical period, used to characterize the relative popularity of the infectious disease in the public opinion field. The total number of medical articles information is the total number of all medical articles obtained within the same statistical period, used to eliminate the impact of fluctuations in the overall number of self-media articles published in different time periods.

[0082] In a specific embodiment, statistical period information is obtained; based on the keyword matching results, the corresponding medical articles are filtered for interference to determine the medical article interference filtering information; based on the statistical period information, the number of medical article interference filtering information is counted to determine the medical article matching quantity information and the corresponding total number of medical articles. That is, interference filtering is performed on the basis of keyword matching results, such as excluding forwarded content and articles that are obviously related to other epidemics in the same period. The number of articles marked as infectious disease-related in each period is counted as the medical article matching quantity information, which can directly represent the public opinion heat related to the disease. At the same time, the total number of all articles after cleaning in the same period is counted as the total number of medical articles to eliminate the impact of fluctuations in the total number of articles in different periods on the indicator.

[0083] In one feasible implementation, step S23 may include steps E11 to E13: Step E11: Obtain statistical period information; It should be noted that the statistical period information is a time unit or interval used to aggregate self-media articles in terms of time dimension, such as a week, a day, or a month. It can be preset according to the monitoring needs of the target infectious disease and the frequency of official data release, or it can be dynamically adjusted to adapt to the spread speed of different diseases, thereby unifying the time granularity and ensuring that the number of medical articles matched and the total number are comparable within the same time window, that is, providing a standardized time benchmark.

[0084] Step E12: Based on the keyword matching results, perform interference filtering on the corresponding medical articles to determine the medical article interference filtering information; It should be noted that the medical article interference filtering information refers to the final set of clean articles and their corresponding tag information. After removing forwarded content, articles that clearly belong to other epidemics during the same period, and noise such as advertising and promotion, only articles that truly represent the activity of the target disease are retained, thereby improving the accuracy of statistics.

[0085] Understandably, interference filtering is a post-processing operation that can eliminate false alarms or irrelevant information that may be introduced during the keyword matching stage. It can significantly improve the accuracy of the matching results and make the final statistical count of medical article matches more accurately represent the actual intensity of public opinion.

[0086] Step E13: Based on the statistical period information, the number of medical article interference filtering information is statistically analyzed to determine the number of matching medical articles and the corresponding total number of medical articles.

[0087] It is understandable that by using the information on the number of matching medical articles and the total number of medical articles, the interference of fluctuations in the total number of articles in different periods on public opinion heat can be eliminated, making the characteristic indicators across periods comparable and statistically stable.

[0088] This embodiment proposes a graded early warning method for medical events. The method involves data cleaning of article titles, content, publication time, publishing platform, and author information from self-media web pages to determine a cleaned data information set. Based on this cleaned data information set, corresponding medical articles are matched with a preset keyword library to obtain keyword matching results. Based on these results, the number of matched medical articles and the total number of medical articles are determined. This method solves the technical problem of accurately predicting the spread of medical events in the real-world environment to achieve graded early warning. Compared to existing technologies, this application, by cleaning the article titles, content, publication time, publishing platform, and author information from self-media web pages to determine a cleaned data information set and matching the corresponding medical articles with a preset keyword library, can quickly filter out effective samples related to the target medical event from massive amounts of text, significantly reducing interference from irrelevant information. Furthermore, by determining the number of matched medical articles and the total number of medical articles based on the keyword matching results, it can accurately represent the activity level of medical events in public opinion and achieve accurate prediction between public opinion signals and the actual epidemic situation.

[0089] For example, to help understand the implementation process of the medical event classification and early warning method obtained in this embodiment combined with the above embodiment one, please refer to... Figure 4 , Figure 4 A simplified flowchart of a medical event tiered early warning method is provided, specifically: Referring to Example 1, information from self-media web pages and historical monitoring information are obtained; based on the self-media web page information, keyword matching is performed on the corresponding medical articles to determine the number of matched medical articles and the total number of corresponding medical articles; based on the number of matched medical articles and the total number of medical articles, the frequency of medical events is determined; based on the frequency of medical events and the historical monitoring information, time lag parameter verification information is determined; based on the frequency of medical events, the historical monitoring information, and the time lag parameter verification information, the activity parameters of medical events are predicted to determine the prediction results of medical event activities; based on the prediction results of medical event activities, the system performs graded early warning. Referring to Example 2, data cleaning is performed on the article titles, body content, publication time, publication platform, and author information in the self-media web page information to determine a data cleaning information set; based on the data cleaning information set, the corresponding medical articles are matched with a preset keyword library to obtain keyword matching results; based on the keyword matching results, the number of matched medical articles and the total number of corresponding medical articles are determined. By acquiring information from social media websites and historical monitoring data, and identifying and extracting features from social media articles related to infectious diseases, an infectious disease prediction model is constructed and optimized. The prediction results are output and tiered early warnings are implemented, such as Level 1 to Level 3 warnings. The model is updated regularly based on actual monitoring feedback, forming a closed-loop optimization. This achieves automated monitoring from multi-source data collection, feature extraction, model prediction to early warning output, effectively improving the timeliness and accuracy of early warnings for infectious diseases.

[0090] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the medical event classification and early warning method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0091] This application also provides a medical event grading and early warning device; please refer to... Figure 5 The medical event grading and early warning device includes: Module 10 is used to acquire information from self-media web pages and historical monitoring information; Processing module 20 is used to perform keyword matching on the corresponding medical articles based on the information of the self-media webpage, and to determine the number of matched medical articles and the total number of corresponding medical articles. Processing module 20 is also used to determine medical event frequency information based on the medical article matching quantity information and the total number of medical articles; Execution module 30 is used to determine time lag parameter verification information based on the medical event frequency information and the historical monitoring information; The execution module 30 is also used to predict the medical event activity parameters based on the medical event frequency information, the historical monitoring information and the time lag parameter verification information, and to determine the medical event activity prediction result; The execution module 30 is also used to control the system for graded early warning based on the prediction results of the medical event activities.

[0092] The processing module 20 is also used to perform data cleaning on the article title, main content, publication time, publication platform and author information in the self-media web page information, and determine the data cleaning information set; Based on the data cleaning information set, the corresponding medical articles are matched with the preset keyword library to obtain the keyword matching results; Based on the keyword matching results, the number of matching medical articles and the total number of corresponding medical articles are determined.

[0093] The processing module 20 is also used to obtain statistical period information; Based on the keyword matching results, the corresponding medical articles are filtered for interference to determine the medical article interference filtering information. Based on the statistical period information, the number of medical article interference filtering information is counted to determine the number of matching medical articles and the corresponding total number of medical articles.

[0094] The processing module 20 is further configured to determine ratio information based on the medical article matching quantity information and the total number of medical articles; Logarithmic transformation is performed on the ratio information to obtain the frequency information of medical events.

[0095] The execution module 30 is also used to obtain information on the combination of time lag orders; The historical monitoring information is proportionally transformed to obtain the target variable for prediction; Based on the frequency information of the medical events, the target variable for prediction is fitted to obtain the fitting result and the corresponding autoregressive exogenous model. Based on the time lag order combination information, the prediction error of the autoregressive exogenous model is verified using a preset cross-validation strategy to determine the verification result of the autoregressive exogenous model. Based on the validation results of the autoregressive exogenous model, the information on the combination of time lag orders is filtered to obtain the validation information of the time lag parameters.

[0096] The execution module 30 is further configured to determine the autoregressive exogenous model parameter information based on the medical event frequency information, the historical monitoring information, and the time lag parameter verification information; Based on the parameter information of the autoregressive exogenous model, the parameters of medical event activities are predicted to obtain the prediction results of medical event activities.

[0097] The execution module 30 is also used to obtain historical average level information; When the predicted result of the medical event exceeds a first preset range of the historical average level information, the control system issues a level one warning. When the predicted result of the medical event activity exceeds a second preset interval of the historical average level information, the control system issues a level-two warning. When the predicted result of the medical event exceeds the third preset interval of the historical average level information, the control system issues a level three warning.

[0098] The medical event grading and early warning device provided in this application, employing the medical event grading and early warning method in the above embodiments, can solve the technical problem of how to accurately predict the spread of medical events in the real environment to achieve grading and early warning. Compared with the prior art, the beneficial effects of the medical event grading and early warning device provided in this application are the same as those of the medical event grading and early warning method provided in the above embodiments, and other technical features in the medical event grading and early warning device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0099] This application provides a medical event classification and early warning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the medical event classification and early warning method in the above embodiment 1.

[0100] The following is for reference. Figure 6 The diagram illustrates a structural schematic suitable for implementing a medical event grading and early warning device according to embodiments of this application. The medical event grading and early warning device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The medical event classification and early warning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0101] like Figure 6 As shown, the medical event grading and early warning device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the medical event grading and early warning device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touch screens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the medical event triage and early warning device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows medical event triage and early warning devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0102] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0103] The medical event grading and early warning device provided in this application, employing the medical event grading and early warning method in the above embodiments, can solve the technical problem of how to accurately predict the spread of medical events in the real environment to achieve grading and early warning. Compared with the prior art, the beneficial effects of the medical event grading and early warning device provided in this application are the same as those of the medical event grading and early warning method provided in the above embodiments, and other technical features in this medical event grading and early warning device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0104] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0105] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0106] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the medical event classification and early warning method in the above embodiments.

[0107] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0108] The aforementioned computer-readable storage medium may be included in the medical event classification and early warning device; or it may exist independently and not be assembled into the medical event classification and early warning device.

[0109] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the medical event grading and early warning device, the medical event grading and early warning device causes the following to occur: acquire social media webpage information and historical monitoring information; perform keyword matching on corresponding medical articles based on the social media webpage information to determine the number of matched medical articles and the total number of corresponding medical articles; determine the frequency of medical events based on the number of matched medical articles and the total number of medical articles; determine time lag parameter verification information based on the frequency of medical events and the historical monitoring information; predict the activity parameters of medical events based on the frequency of medical events, the historical monitoring information, and the time lag parameter verification information to determine the prediction result of medical event activity; and control the system to perform grading and early warning based on the prediction result of medical event activity.

[0110] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0111] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0112] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0113] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described medical event tiered early warning method. This solves the technical problem of how to accurately predict the extent of the spread of medical events in the real-world environment to achieve tiered early warning. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the medical event tiered early warning method provided in the above embodiments, and will not be repeated here.

[0114] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for graded early warning of medical events, characterized in that, The method includes: Obtain information from self-media web pages and historical monitoring data; Based on the information from the self-media webpage, the corresponding medical articles are matched with keywords to determine the number of matched medical articles and the total number of corresponding medical articles. The frequency information of medical events is determined based on the number of matched medical articles and the total number of medical articles. Based on the medical event frequency information and the historical monitoring information, time lag parameter verification information is determined; Based on the frequency information of medical events, the historical monitoring information, and the time lag parameter verification information, the parameters of medical event activities are predicted to determine the prediction results of medical event activities. The system implements tiered early warning based on the predicted results of the medical event activities.

2. The method as described in claim 1, characterized in that, The step of performing keyword matching on corresponding medical articles based on the information from the self-media webpages to determine the number of matched medical articles and the total number of corresponding medical articles includes: Data cleaning is performed on the article titles, body content, publication time, publishing platform, and author information in the aforementioned self-media web page information to determine the data cleaning information set; Based on the data cleaning information set, the corresponding medical articles are matched with the preset keyword library to obtain the keyword matching results; Based on the keyword matching results, the number of matching medical articles and the total number of corresponding medical articles are determined.

3. The method as described in claim 2, characterized in that, The steps of determining the number of matched medical articles and the total number of corresponding medical articles based on the keyword matching results include: Obtain statistical period information; Based on the keyword matching results, the corresponding medical articles are filtered for interference to determine the medical article interference filtering information. Based on the statistical period information, the number of medical article interference filtering information is counted to determine the number of matching medical articles and the corresponding total number of medical articles.

4. The method as described in claim 1, characterized in that, The step of determining the frequency information of medical events based on the matching quantity information of medical articles and the total number of medical articles includes: The ratio information is determined based on the number of matched medical articles and the total number of medical articles. Logarithmic transformation is performed on the ratio information to obtain the frequency information of medical events.

5. The method as described in claim 1, characterized in that, The step of determining the time lag parameter verification information based on the medical event frequency information and the historical monitoring information includes: Obtain information on combinations of time lag orders; The historical monitoring information is proportionally transformed to obtain the target variable for prediction; Based on the frequency information of the medical events, the target variable for prediction is fitted to obtain the fitting result and the corresponding autoregressive exogenous model. Based on the time lag order combination information, the prediction error of the autoregressive exogenous model is verified using a preset cross-validation strategy to determine the verification result of the autoregressive exogenous model. Based on the validation results of the autoregressive exogenous model, the information on the combination of time lag orders is filtered to obtain the validation information of the time lag parameters.

6. The method as described in claim 1, characterized in that, The step of predicting medical event activity parameters based on the medical event frequency information, the historical monitoring information, and the time lag parameter verification information, and determining the medical event activity prediction result, includes: The autoregressive exogenous model parameter information is determined based on the medical event frequency information, the historical monitoring information, and the time lag parameter validation information. Based on the parameter information of the autoregressive exogenous model, the parameters of medical event activities are predicted to obtain the prediction results of medical event activities.

7. The method as described in claim 1, characterized in that, The steps of the hierarchical early warning system based on the medical event activity prediction results include: Obtain historical average level information; When the predicted result of the medical event exceeds a first preset range of the historical average level information, the control system issues a level one warning. When the predicted result of the medical event activity exceeds a second preset interval of the historical average level information, the control system issues a level-two warning. When the predicted result of the medical event exceeds the third preset interval of the historical average level information, the control system issues a level three warning.

8. A medical event grading and early warning device, characterized in that, The device includes: The acquisition module is used to acquire information from self-media web pages and historical monitoring information; The processing module is used to perform keyword matching on the corresponding medical articles based on the information of the self-media webpage, and to determine the number of medical articles matched and the total number of corresponding medical articles. The processing module is also used to determine the frequency information of medical events based on the number of medical article matches and the total number of medical articles. The execution module is used to determine time lag parameter verification information based on the medical event frequency information and the historical monitoring information; The execution module is also used to predict the medical event activity parameters based on the medical event frequency information, the historical monitoring information and the time lag parameter verification information, and to determine the medical event activity prediction result; The execution module is also used to control the system for tiered early warning based on the predicted results of the medical event activities.

9. A medical event grading and early warning device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the medical event grading and early warning method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the medical event classification and early warning method as described in any one of claims 1 to 7.

Images