CN-122025200-A - Flow regulation data monitoring and early warning decision method and system supporting dynamic risk division

CN122025200ACN 122025200 ACN122025200 ACN 122025200ACN-122025200-A

Abstract

The invention discloses a flow regulation data monitoring and early warning decision method supporting dynamic risk division, which comprises the steps of collecting multi-flow regulation data and carrying out integration processing to obtain integrated flow regulation data, carrying out data standardization and data cleaning processing on the integrated flow regulation data to obtain target flow regulation data, constructing a multi-dimensional dynamic risk assessment index system, calculating a regional risk value according to the target flow regulation data and the multi-dimensional dynamic risk assessment index system by adopting an algorithm combining weighted summation and error correction, optimizing a risk threshold value in real time through a threshold value self-adaptive updating model, carrying out dynamic division of a risk region, carrying out propagation link deduction and close-contact risk classification according to the divided risk region to obtain a processing result, carrying out visual early warning according to the processing result and outputting decision advice. The real-time performance of the flow regulation data monitoring, the accuracy of risk division and the scientificity of prevention and control decisions are conveniently improved.

Inventors

JI YINGKAI
TANG JING
LIANG HAO
LI QI
CHEN XU
ZHANG LICHENG
LIU XIAO
GAO FEI
CHEN XUYAO

Assignees

江苏省疾病预防控制中心(江苏省预防医学科学院)

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. An flow regulation data monitoring and early warning decision method supporting dynamic risk division is characterized by comprising the following steps: Acquiring multi-source streaming data and carrying out integration processing to obtain integrated streaming data; performing data standardization and data cleaning treatment on the integrated streaming data to obtain target streaming data; Constructing a multidimensional dynamic risk assessment index system; Calculating a regional risk value by adopting an algorithm combining weighted summation and error correction according to target flow regulation data and a multidimensional dynamic risk assessment index system, optimizing a risk threshold in real time through a threshold self-adaptive updating model, and carrying out dynamic division of a risk region; carrying out propagation link deduction and close-contact risk classification according to the divided risk areas to obtain a processing result; and carrying out visual early warning according to the processing result and outputting decision advice.
2. The method for monitoring and early warning decision-making of streaming data supporting dynamic risk zones according to claim 1, wherein the steps of collecting multi-streaming data and integrating the multi-streaming data comprise: collecting basic health information of cases by adopting RESTfulAPI interfaces through a structured flow regulation form, wherein the basic health information comprises name, age, sex, disease time, diagnosis time, nucleic acid detection result, antibody level, symptom expression and asymptomatic/diagnosis state; The method comprises the steps of carrying out data fusion processing on the current address of a case, LBS positioning data of a communication base station, GPS track data of a transportation trip and a public transportation card swiping record, and collecting space-time track information, wherein the space-time track information comprises an address, longitude and latitude coordinates, an address starting time, a stay time, a trip mode and personnel information of the same person; The method comprises the steps that voice data collected through voice interview equipment are converted into text through an ASR technology based on a transducer architecture, and contact relation information is matched by adopting a data association algorithm in combination with format feedback data of a protocol list, wherein the contact relation information comprises contact person names, identity marks, contact time windows, contact modes and contact time lengths; integrating epidemic place data acquired by an on-site investigation terminal and place information registered by a government system through a data desensitization correlation technology to acquire place characteristic information, wherein the place characteristic information comprises a place type, a building area, a maximum accommodation number, space structure parameters, ventilation times, disinfection frequency and a personnel flow peak; and integrating the basic health information, the space-time track information, the contact relation information and the place characteristic information to obtain integrated flow regulation data.
3. The method for monitoring and early warning decision-making of streaming data supporting dynamic risk zones according to claim 1, wherein the step of performing data normalization and data cleaning on the integrated streaming data to obtain target streaming data comprises the steps of: Formulating a data format standard, and carrying out data standardized conversion on the integrated streaming data to obtain converted data; And eliminating repeated data and missing key information data in the converted data based on the data quality check rule, identifying and correcting unreasonable data in the converted data based on the abnormal value detection algorithm, and supplementing fuzzy data in the converted data to obtain target stream adjustment data.
4. The method for monitoring and early warning decision-making of flow regulation data supporting dynamic risk zones according to claim 1, wherein constructing a multi-dimensional dynamic risk assessment index system comprises: Setting application targets and boundary information of a multi-dimensional dynamic risk assessment index system, wherein the application targets comprise dynamic division of risk areas, close-contact risk classification and prevention and control decision output; Constructing a multidimensional frame based on an epidemic situation transmission causal chain, and screening out a core dimension based on a logic chain of transmission source-transmission carrier-transmission environment-risk intervention, wherein the core dimension comprises transmission capacity, space-time aggregation, site characteristics and prevention and control states; determining key indexes of each core dimension in the risk influence link, and performing index standardization processing on the key indexes to obtain target indexes; Determining the dynamic weight of the target index; and constructing a multi-dimensional dynamic risk assessment index system based on the dynamic weights of the target indexes.
5. The method for monitoring and early warning decision-making with streaming data supporting dynamic risk zones as recited in claim 4, wherein determining key indicators of each core dimension in a risk-affecting link comprises: Transmission capacity, including the duration of the case detoxification period and the infection intensity coefficient; space-time aggregation, including the number of cases in unit area, space-time concomitant density and track coincidence rate; Site characteristics, including personnel density and propagation risk coefficients; And the prevention and control state comprises a control response time length and a close-contact investigation rate.
6. The method for monitoring and early warning decision-making on stream data supporting dynamic risk zones as recited in claim 4, wherein determining the dynamic weights of the target indicators comprises: an expert judgment matrix is constructed through an AHP algorithm, and subjective weights are output based on the fact that epidemiologic experts and prevention and control first-line personnel compare the importance of each core dimension and key index; Carrying out information entropy calculation on the historical flow regulation data by an entropy weight method, determining objective weight according to index distinction, wherein the smaller the entropy value is, the larger the weight is; performing weighted calculation according to subjective weight and objective weight and preset weighting coefficients to obtain initial weight of a target index; setting a trigger condition, and starting weight updating when any trigger condition is satisfied, wherein the number-to-cycle ratio increase of the newly added cases is more than or equal to 30 percent or the cycle ratio decrease is more than or equal to 50 percent, the propagation chain length increase is more than or equal to 2 generations, the propagation force parameter of the virus variant strain is updated, and the risk assessment error is more than or equal to 20 percent; And constructing a weight prediction model by adopting a random forest algorithm, wherein input variables comprise the number of newly added cases, the propagation speed, the implementation completion rate of the control measures and the propagation force coefficient of the virus variant strain, and dynamically adjusting the initial weight to obtain the dynamic weight of the target index.
7. The method for monitoring and early warning decision-making of flow-regulated data supporting dynamic risk zones according to claim 1, wherein the calculation of the zone risk value using an algorithm combining weighted summation and error correction comprises: Calculating a first sum value of each target index and dynamic weight through a summation formula, and taking a second sum value of the first sum value and an error correction term as an initial risk value of a region, wherein the error correction term is determined based on a deviation value of a historical risk assessment accuracy and actual epidemic development, and is optimized in real time by adopting a gradient descent algorithm; And carrying out normalization processing on the initial risk value of the region to obtain the region risk value.
8. The method for monitoring and early warning decision-making of streaming data supporting dynamic risk zones according to claim 1, wherein the step of performing propagation link deduction and close-contact risk classification according to the divided risk zones to obtain a processing result comprises the steps of: Acquiring analysis data corresponding to the risk areas, extracting named entities based on a mixed named entity recognition technology, and constructing a dynamic entity relationship network containing entity attributes, association relationships and time stamp information by an entity relationship extraction technology based on an attention mechanism through constructing a medical field relationship dictionary and combining a time sequence attention mechanism to strengthen extraction capability of time sequence association among entities, and extracting inter-case relationship and case and place visit relationship; the method comprises the steps of taking a dynamic entity relation network as a basis, adopting a bidirectional time sequence association analysis algorithm to obtain a deduction result, inputting a patient toxin expelling period, an infection intensity coefficient, a contact relation type, a track collision result and a place transmission risk coefficient during forward deduction, fitting a transmission direction and intensity through a Bayesian network model, and outputting potential infectious objects, transmission paths, an infection time window and an infection probability of each patient; The method comprises the steps of determining input variables according to deduction results, wherein the input variables comprise contact time length, contact distance and contact mode of a close-contact person and a case, infection intensity coefficient of the case, health state of the close-contact person, overlap degree of close-contact exposure time and a case toxin expelling period, and inputting the input variables into a close-contact person risk classification model constructed by adopting a gradient lifting tree algorithm to obtain a processing result.
9. The method for monitoring and early warning decision-making based on the flow regulation data supporting dynamic risk zones as claimed in claim 1, wherein the steps of performing visual early warning and outputting decision-making advice according to the processing result include: Performing risk area visualization, track and contact relation visualization, close-contact distribution visualization and site risk visualization based on a GIS map according to the processing result to obtain a visualization result; Generating early warning information comprising a risk area range, a risk level, influence groups and potential propagation trends according to different risk levels and scenes by a visual result, and pushing the early warning information to related prevention and control personnel in a short message mode; and carrying out propagation link analysis according to the visual result, and outputting differential decision suggestions aiming at different subjects.
10. A system applying the method for monitoring and early warning decision-making of the flow regulation data supporting dynamic risk zones according to any one of claims 1 to 9, comprising: The acquisition module is used for acquiring the multi-source streaming data and carrying out integration processing to obtain integrated streaming data; The first processing module is used for carrying out data standardization and data cleaning processing on the comprehensive streaming data to obtain target streaming data; the construction module is used for constructing a multi-dimensional dynamic risk assessment index system; The division module is used for calculating a region risk value by adopting an algorithm combining weighted summation and error correction according to the target flow regulation data and a multidimensional dynamic risk assessment index system, optimizing a risk threshold value in real time through a threshold value self-adaptive update model, and carrying out dynamic division on a risk region; the second processing module is used for carrying out propagation link deduction and close-contact risk classification according to the divided risk areas to obtain a processing result; And the early warning module is used for carrying out visual early warning according to the processing result and outputting decision suggestions.

Description

Flow regulation data monitoring and early warning decision method and system supporting dynamic risk division Technical Field The invention relates to the technical field of data processing, in particular to a method and a system for monitoring and early warning decision-making of flow regulation data supporting dynamic risk zones. Background The key steps of flow regulation data monitoring and risk early warning are public health emergency prevention and control, and the key aims are to accurately identify a risk range, study and judge a propagation trend and provide scientific basis for prevention and control decision through analysis of flow regulation data such as case track, close connection relation, site exposure and the like. Currently, technologies such as data visualization, natural language processing, machine learning and the like are increasingly widely applied to flow adjustment work, and a flow adjustment system has realized basic functions such as data acquisition, form management, track display and the like. However, the dynamics and complexity of epidemic situation spreading have higher requirements on timeliness and accuracy of risk division, namely, on one hand, the virus spreading speed is high, the risk area needs to be dynamically adjusted according to real-time data such as new cases and close-contact tracks, and on the other hand, the spreading risks of different places are affected by factors such as personnel density, space structure and ventilation condition, so that the single risk division standard is difficult to adapt to diversified scenes. The traditional flow regulation data processing and risk early warning method has obvious limitations that 1, the risk area is lack of dynamic adjustment capability, namely, the prior art mostly adopts fixed risk division standards (such as fixed case number threshold values), the risk threshold values cannot be adjusted according to dynamic factors such as epidemic propagation speed, virus variation characteristics, prevention and control measures implementation conditions and the like, so that the risk area division is delayed from epidemic situation development, and accurate prevention and control are difficult to realize. 2. The multi-source data integration analysis capability is insufficient, namely the flow regulation data cover multi-source heterogeneous data such as space-time tracks, contact relations, site characteristics, health states and the like, the prior art processes single type data, and multi-dimensional data cannot be effectively integrated to carry out comprehensive risk assessment. For example, only the number of cases is considered, but key factors such as the density of staff in the place, the contact time length and the like are ignored, so that the risk assessment result is on one side. 3. The pertinence and practicality of risk early warning are insufficient, early warning information in the prior art is mostly generalized prompt, differentiated early warning suggestions cannot be provided by combining propagation characteristics of different scenes (such as schools, hospitals and markets), early warning results are mostly displayed in a thermodynamic diagram or list form, prevention and control decision suggestions are not explicitly output, and connection with actual prevention and control work is not tight enough. 4. The propagation link deduction and risk tracing capability is weak, namely, although the prior art can identify close-connected personnel, the prior art lacks deep excavation of propagation relations among cases, a complete propagation link network is difficult to construct, zero cases and potential propagation paths cannot be accurately positioned, and tracing efficiency and pertinence of prevention and control measures are affected. Disclosure of Invention The present invention aims to solve, at least to some extent, one of the technical problems in the above-described technology. Therefore, the invention aims to provide a method and a system for monitoring and early warning decision-making of flow regulation data supporting dynamic risk regions, which improve the instantaneity of flow regulation data monitoring, the accuracy of risk regions and the scientificity of prevention and control decision-making. In order to achieve the above objective, an embodiment of the present invention provides a method for monitoring and early warning decision-making of streaming data supporting dynamic risk zones, including: Acquiring multi-source streaming data and carrying out integration processing to obtain integrated streaming data; performing data standardization and data cleaning treatment on the integrated streaming data to obtain target streaming data; Constructing a multidimensional dynamic risk assessment index system; Calculating a regional risk value by adopting an algorithm combining weighted summation and error correction according to target flow regulation data and a multidimensional dynamic ris