Virus transmission tracing method and system driven by spatio-temporal trajectory data

Abstract

The invention discloses a space-time trajectory data driven virus transmission tracing method and system. The method comprises the following steps: firstly, cleaning and preprocessing an obtained patient space-time trajectory data set; then, based on the risk area grading system and the infection theory, a set propagation chain conforming to the infection theory is constructed for the high-risk province; then coding the preprocessed patient space-time trajectory data into patient features by using a suitable coding mode; training a traceability model for high-risk provinces based on the constructed set propagation chain and the encoded patient features; and finally, applying the trained high-risk province traceability model to related high-risk provinces based on a knowledge migration mode, and performing a propagation chain construction task. The virus propagation chain constructed by the method disclosed by the invention has the characteristics of expandability in data volume, diversity in propagation mode and high conformity with the real situation.

Description

Background Art

[0002] In the field of virus transmission prevention and control and public health research, it is often necessary to quickly and accurately understand the transmission mode, key paths and core sources of virus transmission in the province and city, and divide the transmission risk level areas and populations based on the analysis and research results, and formulate targeted prevention and control measures and policies. However, efficiently constructing a virus transmission chain that conforms to the actual situation for massive high-dimensional spatiotemporal trajectory data of patients requires expert manpower and time costs with specific professional knowledge and data analysis capabilities, which is a heavy task for most local governments and health prevention and control departments. Most spatiotemporal trajectory virus transmission chains constructed based on manual screening and analysis usually have problems of poor timeliness and weak scalability, including the problem that the speed of transmission chain construction is difficult to match the speed of virus transmission and the change of transmission mode due to the large amount of time required for manual screening. In addition, expert manpower is difficult to support the exponential growth of the number of patients' transmission and diffusion. Therefore, it is of great significance to develop a transmission chain generation method that can use neural networks to efficiently process provincial-level data scale, mine potential transmission characteristics between patient data, and automatically construct multi-transmission mode virus transmission chains for conducting virus transmission mode research supported by big data, monitoring the development and changes of viruses in high-risk areas, and strengthening health prevention and control work.

[0003] Traditional methods for constructing virus transmission chains mainly rely on expert knowledge or fixed transmission patterns, lacking timeliness and flexibility. These methods are usually unable to adapt to the needs of large-scale virus outbreaks and the spread of the virus, and it is difficult to automatically generate virus transmission chains with multiple transmission patterns. In addition, these methods cannot fully tap the potential data features in the large amount of patient spatiotemporal trajectory data collected by health prevention and control departments at all levels, and cannot capture the correlation and transmission pattern information between patient data, resulting in the constructed virus transmission chain being non-scalable and inconsistent with the actual situation.

[0004] Specifically, there are two main difficulties in the task of constructing the virus's spatiotemporal trajectory transmission chain.

[0005] First, the patient's spatiotemporal trajectory data is large in volume and high in dimension, which makes it difficult to quickly generate a virus transmission chain with a certain timeliness using manual methods.

[0006] Second, the patient's spatiotemporal trajectory data contains a large number of transmission pattern characteristics that are highly correlated with the current virus transmission situation.

[0007] Therefore, how to automatically generate a realistic virus transmission chain with multiple transmission modes based on patient data is a key direction of current research. Summary of the invention

[0008] The technical problem to be solved by the present invention is to provide a virus transmission tracing method and system driven by spatiotemporal trajectory data in response to the deficiencies in the above-mentioned prior art, so as to solve the technical problems of the existing methods such as poor timeliness, weak scalability, inability to cope with large-scale patient data and the difficulty of building multiple transmission modes, realize automatic processing of patient spatiotemporal trajectory data, automatic generation of virus transmission chains, and maximize the matching between the generation speed of the transmission chain and the speed of transmission and diffusion.

[0009] The present invention adopts the following technical solutions:

[0010] A method for tracing the source of virus transmission driven by spatiotemporal trajectory data, comprising the following steps:

[0011] Acquire the patient's spatiotemporal trajectory data and perform cleaning and preprocessing to obtain a preprocessed patient spatiotemporal trajectory dataset;

[0012] Based on the three-level risk classification system of high, medium and low and the contagion theory, a theoretical transmission chain is constructed for high-risk provinces;

[0013] The obtained patient spatiotemporal trajectory datasets were encoded into patient features using a variety of encoding methods;

[0014] Based on the constructed theoretical transmission chain and the encoded patient characteristics, an unsupervised learning method of constructing positive and negative samples was used to train a traceability model for high-risk provinces;

[0015] Load the trained traceability model, transfer the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and use the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

[0016] Preferably, the preprocessed patient spatiotemporal trajectory dataset is specifically:

[0017] Access to patient data recorded by multiple surveillance sites across the country and local health departments, including patient demographic information, movement trajectory information, virus exposure, epidemiological timeline and clinical symptoms;

[0018] The spatiotemporal trajectory data of patients from various provinces across the country were collected and cleaned based on the official designated portal website for a period of one and a half years.

[0019] Preferably, cleaning is specifically:

[0020] Analyze the information meaning of attribute columns and delete attribute columns that convey duplicate information or have the lowest relevance to the task of building a communication chain;

[0021] The values ​​of each attribute column were analyzed, and invalid attribute columns whose total missing values ​​accounted for more than 70% of the total number of patients were removed;

[0022] Analyze data row information, delete invalid data rows containing erroneous information, modify or remove inaccurate data that introduces noise, and filter and delete data rows with incomplete information;

[0023] The cleaned and preprocessed patient spatiotemporal trajectory data are stored.

[0024] Preferably, the theoretical transmission chain for high-risk provinces is constructed as follows:

[0025] For each traffic communication group, the interpersonal communication group relationship between its members is analyzed, all interpersonal communication group members of all interpersonal communication groups contained in the group are merged with the existing traffic communication group members, the existing traffic communication group members are expanded, and the expanded traffic communication group is marked as a communication chain;

[0026] For each event propagation group, the interpersonal propagation group relationship between its members is analyzed, all interpersonal propagation group members of all interpersonal propagation groups contained in the group are merged with the existing event propagation group members, the existing event propagation group members are expanded, and the expanded event propagation group is marked as a propagation chain;

[0027] For patients without constructed transmission chains, if there are patients belonging to the same human-to-human transmission group, they are marked as the same transmission chain.

[0028] Preferably, the transmission group marking of patients in the current high-risk provinces is as follows:

[0029] Patients with the same interpersonal contact relationship and clinical symptoms are marked as a group of interpersonal transmission;

[0030] Patients with the same virus type in the destination, arrival time and clinical symptoms in the patient's movement trajectory information are marked as a traffic transmission group;

[0031] Patients with the same activity locations and virus types in their clinical symptoms are marked as an event transmission group.

[0032] Preferably, for each marked propagation chain, the specific operations are as follows:

[0033] Patient edges are constructed based on the interpersonal contact relationships in the patient's virus contact situation, and the direction of the edge is determined based on the possible virus contact time in the patient's virus contact situation, that is, the possible virus contact time of the starting node is later than that of the ending node.

[0034] Sort the remaining unconnected patient nodes and the connected terminal nodes according to the possible virus exposure time, following the principle that the possible virus exposure time of the starting node should be later than that of the terminal node, and construct directed edges for the patients according to the sorting results.

[0035] Store all directed edges constructed by all propagation chains contained in the current province in txt format.

[0036] Preferably, multiple encoding methods are used to encode the patient's spatiotemporal trajectory data into patient features, specifically:

[0037] Based on the numerical characteristics of different attributes in the patient data, one-hot encoding is used to encode gender in demographic information and virus type and severity attributes in clinical symptoms;

[0038] Use label coding to encode the province and city in the demographic information, the departure and destination in the movement trajectory information, and the activity location attributes in the virus contact information;

[0039] Use hash coding to encode the interpersonal contact relationship attributes in virus contact situations;

[0040] Use timestamp coding to encode the arrival time in the movement trajectory information, the possible virus exposure time in the virus contact situation, the first symptom time and the confirmed time in the epidemiological timeline.

[0041] Preferably, the traceability model is a multi-layer graph neural network, including an autoencoder and a multi-layer graph neural network constructed by multiple SAGEConv graph convolution modules, and a link prediction model with the ability to generate node feature embeddings. The traceability model is trained using the open source deep learning library PyTorch.

[0042] Preferably, the construction of the transmission chain of relevant high-risk provinces is realized by calculating the similarity of nodes as follows:

[0043] The dot product calculation method is used to calculate the node embedding similarity of all patient relationships contained in the current high-risk province, and compared with the pre-set propagation threshold; the propagation threshold θ is set to 0. If the similarity calculation result is lower than the pre-set propagation threshold θ, there is no propagation relationship between the two patients; if the similarity calculation result is higher than the set propagation threshold θ, there is a propagation relationship between the two patients, and the transmission direction between the patients is determined according to the possible virus contact time among the patients' virus contact situations, following the principle that the infected person's possible virus contact time is later than the transmitter.

[0044] In a second aspect, an embodiment of the present invention provides a virus propagation tracing system driven by spatiotemporal trajectory data, including:

[0045] The data module obtains the patient's spatiotemporal trajectory data and performs cleaning and preprocessing to obtain the preprocessed patient spatiotemporal trajectory data set;

[0046] Building modules, based on the three-level risk classification system of high, medium and low and the contagion theory, to build a theoretical transmission chain for high-risk provinces;

[0047] The encoding module encodes the patient spatiotemporal trajectory dataset into patient features using a variety of encoding methods;

[0048] The training module uses an unsupervised learning method of constructing positive and negative samples to train a traceability model for high-risk provinces based on the constructed theoretical transmission chain and encoded patient characteristics;

[0049] The traceability module loads the trained traceability model, transfers the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and uses the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

[0050] In a third aspect, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for tracing the spread of the virus driven by spatiotemporal trajectory data when executing the computer program.

[0051] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned spatiotemporal trajectory data-driven virus propagation tracing method.

[0052] Compared with the prior art, the present invention has at least the following beneficial effects:

[0053] A virus transmission tracing method driven by spatiotemporal trajectory data, firstly, cleans and preprocesses the patient spatiotemporal trajectory data; then, constructs a hypothetical virus transmission chain for high-risk provinces through a three-level classification system of high, medium and low levels and infection theory, and encodes the preprocessed patient spatiotemporal trajectory data into patient characteristics; then, based on the constructed theoretical transmission chain and the encoded patient characteristics, trains a tracing model for high-risk provinces across the country; finally, by using the transmission chain knowledge transfer method, transfers the transmission mode learned from the current high-risk province transmission chain data set to another related high-risk province transmission chain construction task; learns the node feature embedding of the related high-risk province, and constructs a virus transmission chain for it through dot product calculation; the virus transmission chain generated by the present invention is highly consistent with the actual situation, and has scalability and transmission mode diversity.

[0054] Furthermore, the setting of the patient spatiotemporal trajectory dataset is based on the basic principle of big data analysis, that is, the breadth and depth of the data volume determine the learning ability of the model and the accuracy of the analysis results. By collecting patient data from multiple sources, it is possible to cover a variety of different transmission scenarios, making the model more adaptable in responding to different epidemic developments. In addition, the multi-dimensional characteristics of the dataset (such as demographics, movement trajectories, contact history, etc.) can help the model fully capture the complex relationships between patients during the analysis process, thereby more accurately identifying the key nodes and transmission paths in the virus transmission chain.

[0055] Furthermore, the principle of data cleaning is based on the core concept of data processing, that is, "data quality determines the quality of analysis results". In the process of cleaning spatiotemporal trajectory data, firstly, by deleting redundant and low-correlation attribute columns, reducing data dimensions and reducing model complexity, the model training process is made more efficient and stable. Secondly, by removing attributes with too many missing values ​​and invalid data rows, data noise can be significantly reduced, the sensitivity of the model to key features can be improved, and the risk of misjudgment can be reduced. In addition, cleaning data also involves the correction of erroneous information and the deletion of noisy data. If erroneous information is not handled, it may cause deviations in the model in the construction of the transmission chain, thereby affecting the accuracy of the transmission chain and the scientific nature of prevention and control measures. At the same time, data cleaning can also improve the operability and comprehensibility of the data set, so that the data can play a more effective role in the subsequent coding and modeling process.

[0056] Furthermore, the principle of constructing a theoretical transmission chain is based on the basic epidemiological theory of virus transmission and the particularity of high-risk provinces. In areas with high incidence of the epidemic, the transmission paths of the virus are often more complex and diverse, and it may be difficult to capture key transmission chains in a timely manner by relying solely on data-driven methods. The construction of the theoretical transmission chain utilizes the relationship analysis of the traffic transmission group, the event transmission group, and the interpersonal transmission group to systematically integrate the relationships within and between these groups. By analyzing known transmission patterns, potential transmission paths are inferred, thereby reducing the computational burden of the model in actual data analysis. In addition, the theoretical transmission chain also provides an initial reference framework for treating patients for whom the transmission chain has not been constructed, making the subsequent automatic generation of transmission chains more accurate and efficient.

[0057] Furthermore, the principle of labeling patients with transmission groups is based on the contact transmission theory in epidemiology and the diversity of virus transmission paths. Different transmission groups reflect different ways of virus transmission, such as interpersonal transmission, spatial movement transmission, and specific event transmission. By classifying and labeling patient data by transmission group, data can be divided and organized more effectively, reducing the burden on the model when processing data. The labeled data makes the construction of the transmission chain more modular and systematic, and the model can more accurately identify and analyze key nodes in different transmission pathways. This not only improves the speed of generating virus transmission chains, but also enhances the applicability and flexibility of the model in different scenarios. In addition, the transmission group labeling also provides a preliminary understanding of the virus transmission path in high-risk provinces, so that data processing and analysis can be more targeted in the subsequent transmission chain generation and model training process. This method is particularly suitable for the rapid tracing needs in areas with high incidence of the epidemic, and helps to formulate corresponding prevention and control measures in a timely manner.

[0058] Furthermore, the principle of setting up the marked transmission chain is based on the concept of directed graphs in graph theory. In the construction of the virus transmission chain, the transmission relationship between patients can be regarded as an edge in a directed graph, and the direction of the edge represents the process of the virus spreading from the infector to the infected. By representing the transmission relationship between patients as directed edges and sorting and connecting them according to the time order, the flow process of the virus in the entire transmission chain can be clearly described. This method helps to visualize the complex virus transmission process, making the structure of the transmission chain more intuitive and easy to analyze. At the same time, recording and storing these directed edge data facilitates further data sharing and analysis, and also provides a detailed reference of key nodes and transmission paths for prevention and control decisions.

[0059] Furthermore, the principle of encoding the spatiotemporal trajectory data of patients into features is based on the feature representation and data preprocessing methods in feature engineering. Different encoding methods are suitable for different types of data and can help the model learn and understand the information in the data more effectively. For example, one-hot encoding can avoid the misleading caused by the ordering relationship between category attributes, while label encoding helps to retain the order information. Hash encoding is to reduce the dimension of high-dimensional data, avoid the dimensionality disaster, and maintain the validity of the data. Timestamp encoding can help the model retain the importance of time information, so that the model can make reasonable inferences in the time dimension. By using a variety of encoding methods, various attributes in patient data are converted into numerical forms suitable for model processing, which not only improves the learning efficiency of the model, but also reduces the possibility of information loss. The encoded data can more accurately reflect the characteristics of patient behavior and the potential path of virus transmission, and improve the prediction and generalization capabilities of the model when constructing the transmission chain.

[0060] Furthermore, the setting of the traceability model is based on the theory of graph neural network (GNN), autoencoder theory and link prediction model theory. GNN can process and learn the node embedding representation of graph structure data, and is particularly suitable for capturing and analyzing the complex relationship between nodes. In the virus transmission traceability, the transmission relationship between patients naturally conforms to the data representation method of the graph structure. Through the graph convolution (SAGEConv) module, the model can extract node features layer by layer, while taking into account the influence of neighboring nodes, so as to learn the key features in the transmission path more comprehensively. The role of the autoencoder in the model is to reduce noise and extract features from the input data, further improving the representation ability of the model, especially when processing high-dimensional spatiotemporal trajectory data. The link prediction model uses the learned node feature embedding to predict the edge, that is, to identify which patients may have a transmission relationship and infer the transmission direction. The advantage of this method is that it can automatically generate a transmission chain based on data, rather than relying on manual rules or prior knowledge, which greatly improves the automation and efficiency of traceability.

[0061] Furthermore, the principle of constructing a propagation chain by calculating node similarity is based on the node embedding representation and similarity measurement of graph neural networks. In graph neural networks, the features of each patient node are learned by the model to obtain a high-dimensional embedding representation, which can summarize the patient's spatiotemporal trajectory, contact history, symptoms and other multi-dimensional information. By calculating the similarity of these embedded representations (for example, by dot product calculation), the strength of the propagation relationship between patients can be quantified. The purpose of setting a propagation threshold is to filter out node pairs that are unlikely to have a propagation relationship, ensuring that the model only identifies highly correlated propagation chains. This method can effectively reduce the error rate when constructing a propagation chain and improve the authenticity and accuracy of the propagation chain.

[0062] It can be understood that the beneficial effects of the second aspect mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0063] In summary, the present invention realizes the automation of the whole process from data cleaning, feature coding to transmission chain construction by utilizing the virus transmission tracing method driven by spatiotemporal trajectory data. This greatly reduces the dependence on manual screening and analysis, improves the speed of virus transmission chain construction, and enables the tracing process to adapt more quickly to the speed and pattern changes of virus transmission. In addition, by using a multi-layer graph neural network (GNN) and a variety of feature encoding methods, the present invention can accurately capture the complex features and potential transmission patterns in the patient's spatiotemporal trajectory data; by using node similarity calculation and threshold filtering, it ensures that the construction of the transmission chain conforms to the actual transmission situation, greatly improving the accuracy of the tracing analysis; at the same time, by introducing a multi-layer graph neural network including an autoencoder and a graph convolution module, the tracing model of the present invention can automatically learn and extract key features in patient data, and generate a transmission chain that conforms to the actual situation. This intelligent model significantly reduces labor costs while improving the ability to process large-scale, high-dimensional data. In terms of social value, the present invention can effectively utilize a large amount of patient spatiotemporal trajectory data to automatically construct the virus transmission chain and provide scientific, data-driven decision-making support for public health departments. It not only helps to timely identify high-risk areas and transmission paths, but also provides a basis for precise epidemic prevention and control measures.

[0064] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A flow chart of the virus propagation and tracing method driven by spatiotemporal trajectory data of the present invention;

[0066] Figure 2 This is the overall framework diagram of the virus propagation and tracing method driven by spatiotemporal trajectory data of the present invention;

[0067] Figure 3 This is a framework diagram of the autoencoder of the present invention;

[0068] Figure 4 A schematic diagram of a computer device provided by an embodiment of the present invention;

[0069] Figure 5 The present invention is a block diagram of an electronic device provided according to an embodiment. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0072] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include plural forms unless the context clearly indicates otherwise.

[0073] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects are in an "or" relationship.

[0074] It should be understood that, although the terms first, second, third, etc. may be used to describe preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0075] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0076] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0077] The present invention provides a virus transmission tracing method driven by spatiotemporal trajectory data, firstly, the obtained patient spatiotemporal trajectory data set is cleaned and preprocessed; then, based on the risk zone classification system and the theory of infection, a set transmission chain that conforms to the theory of infection is constructed for high-risk provinces; then, the preprocessed patient spatiotemporal trajectory data is encoded into patient characteristics using a suitable encoding method; then, based on the constructed set transmission chain and the encoded patient characteristics, a tracing model is trained for high-risk provinces; finally, the trained high-risk province tracing model is applied to relevant high-risk provinces based on the knowledge transfer method to perform the transmission chain construction task. The virus transmission chain constructed by the method of the present invention has the characteristics of scalability in data volume, diversity in transmission mode and high consistency with the actual situation.

[0078] See also Figure 1 and Figure 2 The present invention provides a virus propagation tracing method driven by spatiotemporal trajectory data, comprising the following steps:

[0079] S1. Obtain patient spatiotemporal trajectory data, and clean and preprocess the patient data set to obtain a preprocessed patient spatiotemporal trajectory data set;

[0080] S101. Obtain patient data recorded by multiple surveillance sites across the country and local health departments, including patient demographic information, movement trajectory information, virus exposure, epidemiological timeline and clinical symptoms;

[0081] S102. The spatiotemporal trajectory data of patients in various provinces across the country for a period of one and a half years were collected from the official designated portal website, as shown in Table 1. The data set was cleaned to remove erroneous data, inaccurate data, and incomplete data; the quality and credibility of the data set were ensured as follows:

[0082] S1021. Analyze the information meaning of attribute columns and delete attribute columns that convey duplicate information or have a low relevance to the task of building a communication chain;

[0083] S1022, analyzing the values ​​of each attribute column, and removing invalid attribute columns where the total number of missing values ​​exceeds 70% of the total number of patients;

[0084] S1023. Analyze data row information, delete invalid data rows containing error information, modify or remove inaccurate data that introduces noise, and screen and delete data rows with incomplete information;

[0085] S1024. Store the cleaned and preprocessed patient spatiotemporal trajectory data for subsequent analysis and application.

[0086] The cleaning process includes removing the epidemiological information (Original_Text_CN and Original_Text_EN) attribute columns and symptom description (Symptom) attribute columns with more than 70% missing values ​​in the patient trajectory data. Delete patient data containing incorrect information and modify patient data containing inaccurate information to facilitate subsequent risk area classification and transmission chain construction.

[0087] Table 1: Patient's spatiotemporal trajectory data table

[0088]

[0089]

[0090] The processed dataset contains 6431 data in total, each of which mainly includes the patient's demographic information (5 items), movement trajectory information (3 items), virus contact information (3 items), epidemiological timeline (2 items) and clinical symptoms (2 items), with a total of 15 columns of attributes.

[0091] S2. Divide different provinces into risk zones, and construct a theoretical transmission chain for high-risk provinces based on the three-level risk classification system of high, medium and low risks and the theory of contagion;

[0092] The provinces are divided into risk zones using statistical methods. The transmission chain set based on infectious theory can use the patient's movement trajectory information, virus contact and clinical symptoms in the patient's spatiotemporal trajectory data to construct the transmission chain from the perspectives of interpersonal transmission, traffic transmission and event transmission.

[0093] The present invention uses a three-level classification method of high, medium and low to divide the risk areas of each province in the country, and uses statistical methods to calculate and sort the number of patients in each province according to the province attribute column in the patient demographic information. Provinces with more than 400 patients are defined as high-risk provinces, and the patient data of each high-risk province is stored separately for subsequent use, as shown in Table 2.

[0094] Table 2: Statistics of population in high-risk provinces

[0095]

[0096]

[0097] By analyzing all the attribute information contained in the patient's spatiotemporal trajectory data, the patients in the current high-risk provinces are marked as transmission groups, as follows:

[0098] First, the patients with the same interpersonal contact relationship (Contact_ID_Relationship) and the same virus type (Virus_type) in their clinical symptoms were marked as a group of "interpersonal-relation transmission groups" (Interpersonal-relation Transmission Group;

[0099] Then, patients with the same destination (Destination), arrival time (Arrival_Date) and virus type (Virus_type) in their movement trajectory information are marked as a group of "Transportation Transmission Group";

[0100] Finally, patients with the same activity location (Venue) in their virus contact and virus type (Virus_type) in their clinical symptoms are marked as a group of "Event Transmission Groups".

[0101] In this way, in the subsequent process of constructing the set transmission chain, the constructed set transmission chain can be based on the actual transmission status of the virus, and can be developed from three transmission modes: interpersonal contact, spatiotemporal overlap of traffic, and spatiotemporal overlap of activity events, making it authentic and explainable, as shown in Table 3.

[0102] In the stage of setting up the transmission chain construction, first for each traffic transmission group, the interpersonal communication group relationship between its members is analyzed, and all interpersonal communication group members of all interpersonal communication groups contained in the group are merged with the existing traffic transmission group members, that is, the existing traffic transmission group members are expanded, and the expanded traffic transmission group is marked as a transmission chain.

[0103] Similarly, for each event propagation group, the interpersonal propagation group relationship between its members is analyzed, and all interpersonal propagation group members of all interpersonal propagation groups contained in the group are merged with the existing event propagation group members, that is, the existing event propagation group members are expanded, and the expanded event propagation group is marked as a propagation chain.

[0104] Finally, for patients without constructed transmission chains, if there are patients belonging to the same human-to-human transmission group, they are marked as the same transmission chain.

[0105] The transmission chain is a list of edges consisting of directed edges with patients as nodes. Each directed edge in the list represents the transmission relationship from the starting node (the infected person) to the ending node (the infector). For each marked transmission chain, the specific operations are as follows:

[0106] First, the patient edge is constructed according to the interpersonal contact relationship (Contact_ID_Relationship) in the patient's virus contact situation, and the edge direction is determined according to the possible virus contact time (Possible_Date) in the patient's virus contact situation, that is, the possible virus contact time of the starting node (infected person) is later than that of the ending node (infector).

[0107] Afterwards, the remaining unconnected patient nodes and the connected terminal nodes are sorted according to the possible virus exposure time, following the principle that the possible virus exposure time of the starting node (infected person) should be later than that of the terminal node (infector), and directed edges are constructed for the patients according to the sorting results.

[0108] Finally, all directed edges constructed by all propagation chains contained in the current province are stored in txt format.

[0109] Table 3: Table of the number of transmission chain edges set for high-risk provinces

[0110]

[0111] S3, using multiple encoding methods to encode the patient's spatiotemporal trajectory data into patient characteristics;

[0112] S301. According to the numerical characteristics of different attributes in the patient data, the gender (Gender) in the demographic information, the virus type (Virus_type) and the severity (Severity_level) attributes in the clinical symptoms are encoded using a one-hot encoding method;

[0113] S302, using label coding to encode the province (Province) and city (City) in the demographic information, the departure (Departure) and destination (Destination) in the movement trajectory information, and the activity location (Venue) in the virus contact situation;

[0114] S303, using hash coding to encode the interpersonal contact relationship (Contact_ID_Relationship) attribute in the virus contact situation;

[0115] S304. Use timestamp coding to encode the arrival time (Arrival_Date) in the movement trajectory information, the possible virus exposure time (Possible_Date) in the virus exposure situation, the first symptom time (Date_Symptom_Onset) and the confirmed time (Date_Confirmation) in the epidemiological timeline.

[0116] S4. Based on the theoretical transmission chain constructed in step S2 and the patient characteristics encoded in step S3, an unsupervised learning method of constructing positive and negative samples is used to train a traceability model for high-risk provinces;

[0117] The traceability model is a link prediction model that combines an autoencoder and a multi-layer graph neural network built by multiple SAGEConv graph convolution modules and has the ability to generate node feature embeddings.

[0118] See also Figure 3 ,Autoencoder is used for feature representation learning in unsupervised learning. Its core is to characterize the input information by taking the input information as the learning target. It can automatically remove and filter the noise and abnormal data in the patient characteristics, which helps to extract the key information in the patient characteristics, reduce the complexity of the patient characteristics, and generate more representative patient characteristics.

[0119] In the present invention, the autoencoder is used to reduce the dimension of high-dimensional patient features to achieve reconstruction of patient features. In addition, the autoencoder learns and mines the potential features and variables in the patient features, automatically removing and filtering the noise and abnormal data in the patient features. Such processing helps to extract key information in the patient features, reduce the complexity of the patient features, and generate a more representative patient feature representation.

[0120] SAGEConv is a variant of graph convolutional network. The multi-layer graph neural network constructed by multiple SAGEConv graph convolution modules can update the node representation by sampling neighbor nodes layer by layer, aggregating their features and combining them with the current node features. This update method can capture different levels of features in graph structure data layer by layer, improve the node representation ability of the traceability model, and optimize the performance of the traceability model in link prediction tasks.

[0121] Specifically, at a lower level, the network can capture the local neighborhood information of the node, while at a higher level, the network can capture more extensive graph structure information, thereby improving the node representation ability of the traceability model and enhancing the performance of the traceability model in link prediction tasks.

[0122] Use the open source deep learning library PyTorch to train the traceability model. Configure the parameters of the autoencoder and graph neural network as needed, such as the number of hidden units of the autoencoder, the number of layers, number of hidden units, aggregation method, learning rate, and number of training steps of the graph neural network model.

[0123] In the model training phase of the traceability model, an unsupervised learning method is used to construct positive and negative samples on the graph. The unsupervised learning process is usually based on the intuitive assumption that "the embeddings of nodes that are close to each other on the graph are similar, while the embedding distances of unrelated and non-adjacent nodes are far." It usually refers to learning the node embedding of the graph by inputting the structural information and feature information of the graph itself without prior knowledge of node labels (or edge labels). This learning method is suitable for link prediction tasks on graphs.

[0124] The unsupervised learning process of constructing positive and negative samples on the graph means that for a batch of batch nodes on the graph, its neighboring nodes are used as its positive sample nodes, and the nodes that are not connected to the node are used as its negative sample nodes. In order to balance the computational efficiency and avoid overfitting of the model, the number of negative samples of each seed node is set to k = 10. By randomly sampling the positive sample nodes and negative sample nodes of the node, the positive subgraph samples and negative subgraph samples on the graph are constructed; then the total loss of the traceability model is calculated based on the positive subgraph samples and negative subgraph samples using the interval loss method, and the gradient descent (GD) algorithm is used to optimize the model parameters of the traceability model along the direction of reducing the total loss, thereby completing the model training. .

[0125] The core idea of ​​the model optimization method of margin loss is to make the score of positive sub-image samples as high as possible and the score of negative sub-image samples as low as possible, thereby guiding the model to better distinguish between positive and negative samples.

[0126] For the sampled positive and negative sub-graph samples, the model is optimized by calculating the interval loss combined with gradient descent. The loss function expression of calculating the interval loss is as follows:

[0127]

[0128] Among them, E pos is the set of edges in the positive subgraph sample, e is an edge in the positive subgraph sample, e ′ is the edge in the negative subgraph sample corresponding to edge e, score(e) is the score of edge e, score(e ′ ) is the edge e ′ score.

[0129] For each edge e in the graph, there are two hidden vectors h at its endpoints u and h v (u and v represent the starting node and ending node of the edge respectively); by calculating h u and h v The dot product of the edge uv is used to calculate the score uv The calculation expression of edge score is as follows:

[0130] score uv =h u ·h v

[0131] During the optimization process, the scores of positively sampled sub-graph samples need to be as close to 1 as possible, and the scores of negatively sampled sub-graph samples need to be as close to 0 as possible.

[0132] S5. Load the traceability model trained in step S4, transfer the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and use the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

[0133] According to the trained high-risk province traceability model, the set transmission chain and patient characteristics of another related high-risk province are used as input. The trained traceability model can mine the potential transmission chain pattern in the patient characteristics based on the theoretical transmission of interpersonal transmission relationships, traffic transmission relationships, and event transmission relationships shown by patients; and calculate the ability to embed similar nodes for patients with potential transmission relationships.

[0134] In the present invention, the dot product calculation method is used to calculate the node embedding similarity of all patient relationships contained in the current high-risk province, and compare it with the preset propagation threshold. The propagation threshold θ is set to 0. If the similarity calculation result is lower than the preset propagation threshold, it means that there is no propagation relationship between the two patients;

[0135] On the contrary, if the similarity is higher than the set transmission threshold, it means that there is a transmission relationship between the two patients, and the transmission direction between the patients is determined according to the possible virus contact time (Possible_Date) among the patients' virus contact situations, following the principle that the possible virus contact time of the starting node (the infected person) should be later than that of the ending node (the infectious person).

[0136] It will be appreciated by those skilled in the art that various aspects of the present invention may be implemented as systems, methods or program products. Therefore, various aspects of the present invention may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be collectively referred to herein as "circuits", "modules" or "platforms".

[0137] In another embodiment of the present invention, a virus propagation tracing system driven by spatiotemporal trajectory data is provided. The system can be used to implement the above-mentioned virus propagation tracing method driven by spatiotemporal trajectory data. Specifically, the virus propagation tracing system driven by spatiotemporal trajectory data includes a data module, a construction module, an encoding module, a training module and a tracing module.

[0138] Among them, the data module obtains the patient's spatiotemporal trajectory data and cleans and preprocesses it to obtain the preprocessed patient spatiotemporal trajectory data set;

[0139] Building modules, based on the three-level risk classification system of high, medium and low and the contagion theory, to build a theoretical transmission chain for high-risk provinces;

[0140] The encoding module encodes the patient spatiotemporal trajectory dataset into patient features using a variety of encoding methods;

[0141] The training module uses an unsupervised learning method of constructing positive and negative samples to train a traceability model for high-risk provinces based on the constructed theoretical transmission chain and encoded patient characteristics;

[0142] The traceability module loads the trained traceability model, transfers the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and uses the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

[0143] In another embodiment of the present invention, a terminal device is provided, the terminal device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding functions; the processor described in the embodiment of the present invention can be used for the operation of the virus propagation tracing method driven by spatiotemporal trajectory data, including:

[0144] The spatiotemporal trajectory data of patients are obtained and cleaned and preprocessed to obtain the preprocessed spatiotemporal trajectory dataset of patients; based on the three-level risk classification system of high, medium and low and the infectious theory, a theoretical transmission chain is constructed for high-risk provinces; a variety of encoding methods are used to encode the spatiotemporal trajectory dataset of patients into patient characteristics; based on the constructed theoretical transmission chain and the encoded patient characteristics, an unsupervised learning method of constructing positive and negative samples is used to train a traceability model for high-risk provinces; the trained traceability model is loaded, and the node representation capability containing the transmission mode information learned by the traceability model is transferred to another relevant high-risk province, and the transmission chain of the relevant high-risk provinces is constructed by calculating the node similarity.

[0145] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It is understandable that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device, and can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that more specific examples (non-exhaustive list) of the computer-readable storage medium here include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0146] Computer readable storage media also include data signals propagated in baseband or as part of a carrier wave, which carry readable program codes. Such propagated data signals can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium can also be any readable medium other than a readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the readable storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0147] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0148] The processor may load and execute one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the method for tracing the source of virus transmission driven by spatiotemporal trajectory data in the above embodiment; the processor may load and execute the following steps:

[0149] The spatiotemporal trajectory data of patients are obtained and cleaned and preprocessed to obtain the preprocessed spatiotemporal trajectory dataset of patients; based on the three-level risk classification system of high, medium and low and the infectious theory, a theoretical transmission chain is constructed for high-risk provinces; a variety of encoding methods are used to encode the spatiotemporal trajectory dataset of patients into patient characteristics; based on the constructed theoretical transmission chain and the encoded patient characteristics, an unsupervised learning method of constructing positive and negative samples is used to train a traceability model for high-risk provinces; the trained traceability model is loaded, and the node representation capability containing the transmission mode information learned by the traceability model is transferred to another relevant high-risk province, and the transmission chain of the relevant high-risk provinces is constructed by calculating the node similarity.

[0150] See also Figure 4 , the terminal device is a computer device. The computer device 60 of this embodiment includes: a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When the computer program 63 is executed by the processor 61, the method for tracing the virus propagation driven by the spatiotemporal trajectory data in the embodiment is implemented. To avoid repetition, it is not described one by one here. Alternatively, when the computer program 63 is executed by the processor 61, the functions of each model / unit in the virus propagation tracing system driven by the spatiotemporal trajectory data in the embodiment are implemented. To avoid repetition, it is not described one by one here.

[0151] The computer device 60 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will appreciate that Figure 4 This is only an example of the computer device 60 and does not constitute a limitation of the computer device 60. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0152] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, central processing units, graphics processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, data processing logic based on quantum computing, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0153] The memory 62 may be an internal storage unit of the computer device 60, such as a hard disk or memory of the computer device 60. The memory 62 may also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the computer device 60.

[0154] Furthermore, the memory 62 may include both an internal storage unit of the computer device 60 and an external storage device. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 may also be used to temporarily store data that has been output or is to be output.

[0155] Any reference to memory, database or other media used in the embodiments provided in the present application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0156] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0157] See also Figure 5 The terminal device 600 is an electronic device, and the electronic device is in the form of a general computing device. The components of the electronic device may include but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, etc.

[0158] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present invention described in the above method section of this specification. For example, the processing unit 610 can perform the following steps: Figure 1 Follow the steps shown in .

[0159] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0160] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0161] Bus 630 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0162] The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 650. Furthermore, the electronic device 600 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 660. The network adapter 660 may communicate with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, etc.

[0163] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0164] Experimental environment and parameter settings:

[0165] Written in Python. Runs on Windows 11, CUDA 11.6, Pytorch 2.2.0, Python 3.9, and NVIDIA GeForce RTX 3050Ti.

[0166] The detection method was trained for 200 epochs using the Adam optimizer with a learning rate of 0.01, 128 hidden layer units and 8 output dimensions for the autoencoder, and 100 hidden layer units and 8 output dimensions for the graph neural network.

[0167] Evaluation Metrics:

[0168] The effect of model fitting is judged by the loss. The lower the loss, the better the model training effect. When the loss tends to be stable, the model training can be ended.

[0169] Table 4: Local experimental results

[0170]

[0171]

[0172] Table 5: Local information of sub-chain patients

[0173]

[0174]

[0175] Experimental results:

[0176] Based on the spatiotemporal trajectory data of patients, the present invention marks the transmission group, trains the traceability model, constructs the transmission chain, and automatically generates a virus transmission chain with more diverse patterns. Table 4 shows the local subchain of the experimental results generated by the traceability model in Henan Province, and Table 5 shows the local information of the patients in the subchain. The experiment shows the effectiveness of the virus transmission and traceability method driven by spatiotemporal trajectory data.

[0177] In summary, the present invention provides a virus transmission source tracing method and system driven by spatiotemporal trajectory data, and proposes a virus transmission chain automatic construction model with scalability and diversity of transmission modes for large-scale patient data; the model uses the movement trajectory information, virus contact conditions and clinical symptoms in the patient's spatiotemporal trajectory data to construct a set transmission chain that conforms to the theory of infection from the perspectives of interpersonal transmission, traffic transmission and event transmission. Based on the encoded patient characteristics, a tracing model is trained for high-risk provinces, which first uses an autoencoder to reduce the dimension of patient characteristics into more representative feature embedding, and then uses the SAGEConv graph convolution module to mine the potential transmission mode of the data; further, the relevant high-risk provinces use the trained tracing model to transfer the transmission mode and node feature representation capabilities learned by the model to the local provinces for application, and automatically generate a virus transmission chain with multiple transmission modes based on the theory of infection.

[0178] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0179] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0181] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0182] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0183] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A virus transmission tracing method driven by spatiotemporal trajectory data, It is characterized in that The following steps are involved: Acquire the patient's spatiotemporal trajectory data and perform cleaning and preprocessing to obtain a preprocessed patient spatiotemporal trajectory dataset; Based on the three-level risk classification system of high, medium and low and the contagion theory, a theoretical transmission chain is constructed for high-risk provinces; The obtained patient spatiotemporal trajectory datasets were encoded into patient features using a variety of encoding methods; Based on the constructed theoretical transmission chain and the encoded patient characteristics, an unsupervised learning method of constructing positive and negative samples was used to train a traceability model for high-risk provinces; Load the trained traceability model, transfer the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and use the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

2. According to the spatiotemporal trajectory data-driven virus transmission tracing method of claim 1, It is characterized in that The preprocessed patient spatiotemporal trajectory dataset is as follows: Access to patient data recorded by multiple surveillance sites across the country and local health departments, including patient demographic information, movement trajectory information, virus exposure, epidemiological timeline and clinical symptoms; The spatiotemporal trajectory data of patients from various provinces across the country were collected and cleaned based on the official designated portal website for a period of one and a half years.

3. According to the spatiotemporal trajectory data-driven virus transmission tracing method of claim 2, It is characterized in that Cleaning is as follows: Analyze the information meaning of attribute columns and delete attribute columns that convey duplicate information or have the lowest relevance to the task of building a communication chain; The values ​​of each attribute column were analyzed, and invalid attribute columns whose total missing values ​​accounted for more than 70% of the total number of patients were removed; Analyze data row information, delete invalid data rows containing erroneous information, modify or remove inaccurate data that introduces noise, and filter and delete data rows with incomplete information; The cleaned and preprocessed patient spatiotemporal trajectory data are stored.

4. The method for tracing the source of virus transmission driven by spatiotemporal trajectory data according to claim 1, It is characterized in that The specific construction of the theoretical transmission chain for high-risk provinces is as follows: For each traffic communication group, the interpersonal communication group relationship between its members is analyzed, all interpersonal communication group members of all interpersonal communication groups contained in the group are merged with the existing traffic communication group members, the existing traffic communication group members are expanded, and the expanded traffic communication group is marked as a communication chain; For each event propagation group, the interpersonal propagation group relationship between its members is analyzed, all interpersonal propagation group members of all interpersonal propagation groups contained in the group are merged with the existing event propagation group members, the existing event propagation group members are expanded, and the expanded event propagation group is marked as a propagation chain; For patients without constructed transmission chains, if there are patients belonging to the same human-to-human transmission group, they are marked as the same transmission chain.

5. According to the spatiotemporal trajectory data-driven virus transmission tracing method of claim 4, It is characterized in that The transmission group markings for patients in the current high-risk provinces are as follows: Patients with the same interpersonal contact relationship and clinical symptoms are marked as a group of interpersonal transmission; Patients with the same virus type in the destination, arrival time and clinical symptoms in the patient's movement trajectory information are marked as a traffic transmission group; Patients with the same activity locations and virus types in their clinical symptoms are marked as an event transmission group.

6. The method for tracing the source of virus transmission driven by spatiotemporal trajectory data according to claim 4, It is characterized in that For each marked propagation chain, the specific operations are as follows: Construct patient edges based on the interpersonal contact relationships in the patient's virus contact situation, and determine the direction of the edge based on the possible virus contact time in the patient's virus contact situation, that is, the possible virus contact time of the starting node is later than that of the ending node; Sort the remaining unconnected patient nodes and the connected terminal nodes according to the possible virus exposure time, following the principle that the possible virus exposure time of the starting node should be later than that of the terminal node (, and construct directed edges for the patients according to the sorting results; Store all directed edges constructed by all propagation chains contained in the current province in txt format.

7. The method for tracing the source of virus transmission driven by spatiotemporal trajectory data according to claim 1, It is characterized in that The patient's spatiotemporal trajectory data is encoded into patient characteristics using a variety of encoding methods, specifically: Based on the numerical characteristics of different attributes in the patient data, one-hot encoding is used to encode gender in demographic information and virus type and severity attributes in clinical symptoms; Use label coding to encode the province and city in the demographic information, the departure and destination in the movement trajectory information, and the activity location attributes in the virus contact information; Use hash coding to encode the interpersonal contact relationship attributes in virus contact situations; Use timestamp coding to encode the arrival time in the movement trajectory information, the possible virus exposure time in the virus contact situation, the first symptom time and the confirmed time in the epidemiological timeline.

8. The method for tracing the source of virus transmission driven by spatiotemporal trajectory data according to claim 1, It is characterized in that The traceability model is a multi-layer graph neural network, including an autoencoder and a multi-layer graph neural network built by multiple SAGEConv graph convolution modules, and a link prediction model with the ability to generate node feature embeddings. The traceability model is trained using the open source deep learning library PyTorch.

9. The method for tracing the source of virus transmission driven by spatiotemporal trajectory data according to claim 1, It is characterized in that The construction of the transmission chain of relevant high-risk provinces by calculating node similarity is as follows: The dot product calculation method is used to calculate the node embedding similarity of all patient relationships contained in the current high-risk provinces and compare them with the pre-set propagation threshold; Set the propagation threshold θ = 0. If the similarity calculation result is lower than the preset propagation threshold θ, there is no propagation relationship between the two patients. If the similarity calculation result is higher than the set transmission threshold θ, there is a transmission relationship between the two patients, and the transmission direction between the patients is determined based on the possible virus contact time among the patients' virus contact situations, following the principle that the infected person's possible virus contact time is later than the transmitter.

10. A virus transmission tracing system driven by spatiotemporal trajectory data, It is characterized in that include: The data module obtains the patient's spatiotemporal trajectory data and performs cleaning and preprocessing to obtain the preprocessed patient spatiotemporal trajectory data set; Building modules, based on the three-level risk classification system of high, medium and low and the contagion theory, to build a theoretical transmission chain for high-risk provinces; The encoding module encodes the patient spatiotemporal trajectory dataset into patient features using a variety of encoding methods; The training module uses an unsupervised learning method of constructing positive and negative samples to train a traceability model for high-risk provinces based on the constructed theoretical transmission chain and encoded patient characteristics; The traceability module loads the trained traceability model, transfers the node representation capability containing the propagation pattern information learned by the traceability model to another relevant high-risk province, and uses the calculation of node similarity to realize the construction of the propagation chain of the relevant high-risk province.

Images