A spatio-temporal prediction method, system, terminal and storage medium for urban epidemic situations

Abstract

The present application relates to a method, system, terminal, and storage medium for spatio-temporal prediction of urban epidemic situations. The method includes: collecting individual movement trajectory data and infectious disease case data within a city; processing the individual movement trajectory data, extracting the population movement flow between regions, dividing the population movement flow according to time attributes, and obtaining the population movement relationship under each time attribute based on the proximity relationship between regions; calculating the similarity of infectious disease case data between regions using a histogram-based similarity algorithm, and obtaining the position attention relationship between regions according to the similarity of the infectious disease case data; constructing an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship, and position attention relationship. The present application can model the time dependence and spatial dependence in infectious disease case data, and improve the spatial perception ability and prediction performance of the spatio-temporal prediction model.

Description

Technical Field

[0001] This application belongs to the technical field of infectious disease computing, and particularly relates to an urban epidemic spatio-temporal prediction method, system, terminal, and storage medium. Background Art

[0002] Due to the increased availability of public health surveillance data and the development of sophisticated methods, infectious disease prediction has become increasingly prominent in recent years. However, real-time prediction of disease infections is still hampered by the lack of current estimates of mobility and interaction patterns, which are key drivers of disease transmission. Predicting the incidence trend of infectious diseases is an important issue in the field of public health. Timely detection, tracking, and prediction of key information about infectious diseases, such as peak intensity and outbreak time, are crucial for effectively formulating prevention and control strategies and implementing intervention measures. According to the principles and purposes of modeling, infectious disease models can be classified into two major categories: mechanistic models and non-mechanistic models.

[0003] Taking seasonal infectious diseases as an example, monitoring and predicting the activities of infectious diseases and making corresponding prevention and control preparations in a timely manner are crucial for the prevention and control of seasonal infectious diseases and pandemics. Existing infectious disease prediction modeling methods include those based on Gaussian process models, LSTM (Long Short-Term Memory) neural network models, etc. However, existing infectious disease prediction modeling methods include those based on traditional statistical models and those based on traditional machine learning. Infectious disease prediction methods based on traditional statistical models require data to meet some strict assumptions, such as the stationarity assumption of time series, etc. However, in the real world, these assumptions are not easy to satisfy. Most infectious disease prediction methods based on traditional machine learning still rely on feature engineering and cannot achieve accurate predictions for more difficult prediction problems. In addition, the above methods usually ignore other features from the spatial dimension and thus cannot model the spatio-temporal dependencies contained in spatio-temporal data, which limits the prediction performance of the model.

[0004] The main core of the mechanistic model is the transmission dynamics model, and the most commonly used one is the compartment model. The compartment model divides the population or other hosts within the same unit into corresponding different compartments according to the infection status of the infected individuals, and then simulates the disease development dynamics of different compartments based on the disease transmission characteristics. Its typical representatives are the susceptible-infected-recovered (SIR) and susceptible-exposed-infected-recovered (SEIR) models, etc. Such models are usually used to predict the long-term development trend of diseases and deduce the effects of different intervention measures. However, their modeling is complex, the update is slow, and the uncertainty of model parameters will also greatly affect the accuracy of the model, making it difficult to achieve real-time and accurate prediction of the short-term development trend of infectious diseases.

[0005] Non-mechanistic models include statistical models and machine learning models. Traditional statistical models require that the data meet some strict assumptions (such as the stationarity assumption of time series). However, in the real world, these assumptions are not easy to meet. Although the infectious disease prediction methods based on traditional machine learning are data-driven and do not need to meet strict assumptions about the data, most traditional machine learning methods still rely on feature engineering and cannot achieve accurate predictions for difficult prediction problems. In addition, during the modeling process, whether it is the infectious disease prediction method based on traditional statistical models or the one based on traditional machine learning models, these methods usually ignore other features from the spatial dimension and thus cannot model the spatio-temporal dependencies contained in spatio-temporal data. This modeling method limits the prediction performance of these methods. The infectious disease prediction method based on deep learning can automatically learn non-linear features from spatio-temporal data and has greater performance advantages. Some studies have designed deep learning-based prediction frameworks from multiple perspectives to better model the spatio-temporal dependence relationships contained in spatio-temporal data and improve the prediction performance of the model. However, when modeling the spatial dependence relationships between different regions, most of these studies do not consider the spatial interaction caused by population movement and the spatio-temporal diffusion pattern of infectious diseases within the city. Due to the lack of information, such methods are difficult to accurately predict the epidemic trend of infectious diseases within the city. Summary of the Invention

[0006] This application provides a spatio-temporal prediction method, system, terminal, and storage medium for urban epidemic situations, aiming to solve at least one of the above technical problems in the prior art to a certain extent.

[0007] To solve the above problems, this application provides the following technical solutions:

[0008] A spatio-temporal prediction method for urban epidemic situations, comprising:

[0009] Collect individual movement trajectory data and infectious disease case data within the city;

[0010] Process the individual movement trajectory data through a data-driven method, extract the population movement flow between regions, divide the population movement flow according to time attributes, and obtain the population movement relationship under each time attribute based on the proximity relationship between regions;

[0011] Use a histogram-based similarity algorithm to calculate the similarity of infectious disease case data between regions, and obtain the location attention relationship between regions according to the similarity of the infectious disease case data; construct an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship, and location attention relationship;

[0012] Input the infectious disease case data within the city into the infectious disease spatio-temporal prediction model, and obtain the infectious disease prediction result within the city through the infectious disease spatio-temporal prediction model.

[0013] The technical solution adopted in the embodiment of this application further includes: The specific implementation of dividing the population movement flow according to time attributes and obtaining the population movement relationship under each time attribute based on the proximity relationship between regions is as follows:

[0014] The time attributes include weekdays, weekends, or holidays; convert the population movement flow of each time attribute into a weighted directed graph G=(V,E), where V represents the set of nodes, E represents the set of edges, the vertices represent the regions within the city, and the edges are used to capture the movement patterns; and based on the proximity relationship between regions, determine whether there is an edge between two nodes in the directed graph according to whether the two regions are in contact with each other, and obtain the adjacency matrix of the population movement relationship corresponding to the proximity relationship.

[0015] The technical solution adopted in the embodiment of this application further includes: The implementation of using a histogram-based similarity algorithm to calculate the similarity of infectious disease case data between regions and obtaining the location attention relationship between regions according to the similarity of the infectious disease case data includes:

[0016] Construct corresponding histograms for the infectious disease case data of each region respectively;

[0017] Calculate the similarity of infectious disease case data between two regions. If the similarity of infectious disease case data between two regions is higher than the set threshold, it is considered that the infectious disease outbreak trends of the two regions are similar and correlated, and an edge on the graph is constructed between the two regions in the directed graph, generating an adjacency matrix representing the location attention relationship between all regions.

[0018] The technical solution adopted in the embodiment of the present application further includes: The calculation formula of the position attention relationship is as follows:

[0019]

[0020] where θ is the threshold for establishing edges based on the histogram similarity algorithm; when the similarity w between region i and region j i,j is higher than the threshold θ, an edge is created between region i and region j, and finally an adjacency matrix of the position attention relationship corresponding to each region in the city is obtained.

[0021] The technical solution adopted in the embodiment of the present application further includes: The construction of the infectious disease spatio-temporal prediction model based on the graph neural network and the long short-term memory network according to the population movement relationship, the proximity relationship, and the position attention relationship includes:

[0022] Input the graph structure into the graph neural network, and the graph neural network normalizes the directed graph using the neighborhood aggregation method so that the weighted in-edges of each node are equal to 1:

[0023]

[0024] where H i is a matrix containing the node representations of the previous layer, and the initial H 0 is set to the historical data representing the changes in the number of infectious disease cases in each region, W i represents the trainable parameter matrix of the i-th layer, and f is a non-linear activation function.

[0025] The technical solution adopted in the embodiment of the present application further includes: The construction of the infectious disease spatio-temporal prediction model based on the graph neural network and the long short-term memory network according to the population movement relationship, the proximity relationship, and the position attention relationship further includes:

[0026] At each time step, a message passing neural network is used to obtain a representation sequence h i,t-n , h i,t-n+1 ,..., h i,t-1 , and the representation sequence h i,t-n , h i,t-n+1 ,..., h i,t-1 is input into the long short-term memory network to extract the time series relationship therein; the calculation formula of the long short-term memory network is as follows:

[0027] X i,t = LSTM(h i,t-n , h i,t-n+1 ,..., h i,t-1 ) where X i,t represents the predicted infectious disease case data of the i-th region in the t-th time period, and h i,t-1Denote the infectious disease case data of the $i$-th region in the $(t - 1)$-th time period.

[0028] The technical solution adopted in the embodiment of this application further includes: The input of the infectious disease case data in the city into the infectious disease spatio-temporal prediction model, and the obtaining of the infectious disease prediction result in the city through the infectious disease spatio-temporal prediction model is specifically as follows:

[0029] Fuse the adjacency relationship, population movement relationship, and location attention relationship through the infectious disease spatio-temporal prediction model to obtain the urban infectious disease prediction result; the specific fusion method is:

[0030] Where $W$ adj , $W$ od and $W$ at are parameter matrices to be trained. and are the infectious disease prediction results at time $t$ obtained based on the adjacency matrix, population movement flow matrix, and location attention matrix respectively. Tanh is the activation function. is the final prediction result of the entire spatio-temporal prediction model at time $t$.

[0031] Another technical solution adopted in the embodiment of this application is: An urban epidemic spatio-temporal prediction system, including:

[0032] Data collection module: Used to collect the individual movement trajectory data and infectious disease case data in the city;

[0033] Flow calculation module: Used to process the individual movement trajectory data through a data-driven method, extract the population movement flow between regions, divide the population movement flow according to time attributes, and obtain the population movement relationship under each time attribute based on the regional adjacency relationship;

[0034] Similarity calculation module: Used to calculate the similarity of the infectious disease case data between regions using a histogram-based similarity algorithm, and obtain the location attention relationship between regions according to the similarity of the infectious disease case data;

[0035] Model construction module: Used to construct an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, adjacency relationship, and location attention relationship;

[0036] Infectious disease prediction module: Used to input the infectious disease case data in the city into the infectious disease spatio-temporal prediction model, and obtain the infectious disease prediction result in the city through the infectious disease spatio-temporal prediction model.

[0037] Another technical solution adopted in the embodiments of this application is as follows: A terminal, the terminal includes a processor and a memory coupled to the processor, wherein,

[0038] the memory stores program instructions for implementing the urban epidemic spatio-temporal prediction method;

[0039] the processor is used to execute the program instructions stored in the memory to control the urban epidemic spatio-temporal prediction.

[0040] Another technical solution adopted in the embodiments of this application is as follows: A storage medium stores program instructions that can be run by a processor, and the program instructions are used to execute the urban epidemic spatio-temporal prediction method.

[0041] Compared with the prior art, the beneficial effects produced by the embodiments of this application are as follows: The urban epidemic spatio-temporal prediction method, system, terminal and storage medium of the embodiments of this application extract the urban population movement flow through the individual movement trajectory data in the city, divide the population movement flow within a certain time period according to the time attribute, obtain the proximity relationship, population movement relationship and position attention relationship between regions according to the population movement flow with different time attributes, and construct an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the proximity relationship, population movement relationship and position attention relationship, and make a refined prediction of the infectious disease trend through the infectious disease spatio-temporal prediction model. By dividing the population movement flow according to the time attribute, the embodiments of this application can accurately capture the impact of population interaction and movement under different time attributes on the spread and transmission of infectious diseases within the city, make full use of the local spatial context information of each region, and consider various spatial relationships between regions, so as to better model the time dependence and spatial dependence in the infectious disease case data, greatly improve the spatial perception ability of the infectious disease spatio-temporal prediction model and the model prediction performance, realize the prediction of the development trend of infectious diseases within the city with a higher spatial resolution, complete the refined analysis of the infectious disease epidemic situation, help the government and public health departments timely and accurately understand the development trend of infectious diseases within the city, and carry out targeted epidemic prevention and control interventions, which can maximize the protection of people's life and health safety. Description of the Drawings

[0042] Figure 1 is a flowchart of the urban epidemic spatio-temporal prediction method according to the embodiments of this application;

[0043] Figure 2 is a schematic structural diagram of the urban epidemic spatio-temporal prediction system according to the embodiments of this application;

[0044] Figure 3 is a schematic structural diagram of the terminal according to the embodiments of this application;

[0045] Figure 4Schematic structural diagram of the storage medium according to the embodiment of the present application. Detailed implementation manners

[0046] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] Aiming at the deficiencies of the prior art, the urban epidemic spatio-temporal prediction method of the embodiment of the present application constructs a spatio-temporal prediction model based on a graph neural network and a long short-term memory network based on the proximity relationship, population movement relationship and position attention relationship between regions in the city, so as to better model the time dependence and spatial dependence in spatio-temporal data. When modeling the spatial dependence, various spatial relationships between regions are considered and the corresponding relationship graph structure is constructed; when modeling the time dependence, the local spatial context information of each region is fully utilized to improve the prediction performance of the model; and according to the population movement patterns of the susceptible population of infectious diseases, the population movement flow is divided into different time attributes, and the directed graph structures of different time attributes are respectively constructed to improve the spatio-temporal perception ability of the spatio-temporal prediction model and further improve the prediction performance of the model.

[0048] Specifically, please refer to Figure 1 , which is a flowchart of the urban epidemic spatio-temporal prediction method according to the embodiment of the present application. The urban epidemic spatio-temporal prediction method according to the embodiment of the present application includes the following steps:

[0049] S1: Obtain the infectious disease case data in the city and collect the individual movement trajectory data in the city by using mobile devices;

[0050] In this step, the infectious disease case data includes but is not limited to data such as the number of infected cases, the number of death cases, clinical symptoms, and hospitalization personnel information. The individual movement trajectory data includes but is not limited to data such as the mobile phone number, signaling timestamp, and mobile phone location data of each individual. In order to better predict the incidence trend of infectious diseases within the city for susceptible populations such as minors, a large amount of individual movement trajectory data and infectious disease cases are obtained in the embodiment of the present application. The mobile devices include but are not limited to devices such as mobile phones or smart watches.

[0051] S2: Process the individual movement trajectory data by a data-driven method, extract the population movement flow between different regions, and obtain the urban population movement relationship;

[0052] In this step, processing the individual movement trajectory data by a data-driven method is specifically: extracting the population movement flow between different regions from the large-scale mobile phone location data and capturing the urban population movement relationship.

[0053] S3: Divide the population movement flow according to time attributes, construct a directed graph of the population movement flow with different time attributes, and determine whether there is an edge between two nodes in the directed graph based on the regional adjacency relationship, and generate an adjacency matrix of the population movement relationship under different time attributes;

[0054] In this step, the time attributes include but are not limited to weekdays, weekends or holidays. Suppose a city is given, a graph structure with time intervals of days, weeks or months is created, and the population movement flow is divided into population movement flows with two time attributes according to weekdays and weekends. The population movement flow of each time attribute is respectively converted into a weighted directed graph G=(V, E), where V represents the set of nodes and E represents the set of edges. The vertices represent the various regions within the city, and the edges are used to capture the movement patterns. For example, the weight from vertex v to vertex u represents the total number of people moving from region v to region u on weekdays in the t-th week, so as to obtain the adjacency matrices of the population movement relationships under the two time attributes of weekdays and weekends in the t-th week of the city respectively. and

[0055] Furthermore, according to the first law of geography, all things are related to other things, but things that are closer are more related than things that are farther away. Therefore, the embodiments of the present application consider the spatial relationship, that is, the adjacency relationship, between things that are close. Based on the regional adjacency relationship, it is determined whether there is an edge between two regions (nodes) according to whether the two regions are in contact with each other, and the adjacency matrix A of the population movement relationship corresponding to the adjacency relationship is obtained. adj . Specifically, assume that regions v and u are adjacent regions, and the weight from vertex v to vertex u represents the total number of individuals moving from region v as the departure place to region u on weekdays. The directed graph G can include the population flow behaviors of each adjacent region within the city. The mobility between regions v and u on weekdays forms an edge, and then multiplied by the number of cases in region v during this time period, a relative score is obtained, and this score represents how many infected people may flow from region v to region u. It can be understood that the case change patterns between two adjacent regions may not have mobility either. For example, regions i and j are adjacent regions, the flow of region i is fixed and always zero, and the flow of region j is dynamic and non-zero, then the case change patterns of these two adjacent regions have no correlation.

[0056] S4: Construct a histogram of the infectious disease case data, calculate the similarity of the infectious disease case data between regions using a histogram-based similarity algorithm, and obtain an adjacency matrix representing the position attention relationship between all regions according to the similarity of the infectious disease case data;

[0057] In this step, the correlation between two regions may be affected by geographical distance. That is, adjacent regions may have similar topographical or climatic characteristics, making them have similar trends of infectious disease outbreaks. However, due to population mobility and similar geographical features, non-adjacent regions may also have potential dependencies, but it is difficult to simulate all relevant factors of infectious disease outbreaks. Therefore, the present invention uses a histogram-based similarity algorithm to calculate the correlation of infectious disease case data between regions. If the similarity of infectious disease case data between two regions is high, correspondingly, their case change patterns will also be relatively similar. In the embodiments of the present application, by calculating the similarity of infectious disease case data of these non-adjacent but highly correlated regions, the case change trend can be captured from a more global spatial perspective.

[0058] The histogram-based similarity algorithm is specifically as follows: First, construct corresponding histograms for the time series of infectious disease case data samples in different regions respectively; secondly, calculate the similarity of the infectious disease case data between two regions. If the similarity of the infectious disease case data between two regions is higher than a set threshold, it is considered that the infectious disease outbreak trends of these two regions are relatively similar and there is a strong correlation between them. A connecting edge on the graph is constructed between these two regions in the directed graph, thereby generating an adjacency matrix representing the positional attention relationship between all regions. The calculation formula is as follows:

[0059]

[0060] where θ is the threshold for establishing a connecting edge by the histogram-based similarity algorithm. When the similarity w between region i and region j i,j is higher than the threshold θ, a connecting edge is created between region i and region j, and finally an adjacency matrix representing the positional attention relationship corresponding to each region in the city is obtained.

[0061] Specifically, the histogram construction algorithm is as follows:

[0062]

[0063] The histogram-based similarity algorithm is as follows:

[0064] Similarity algorithm

[0065]

[0066] S5: Train an infectious disease spatio-temporal prediction model based on a graph neural network (Graph Neural Networks, GNN) and a long short-term memory network (Long short-term memory, LSTM) using the proximity relationship, population movement relationship, and positional attention relationship.

[0067] In this step, the graph neural network is a neighbor aggregation strategy, and the representation vector of a node is calculated by its neighbor nodes through cyclic aggregation and transfer of the representation vector. The framework of the graph neural network is the Message Passing Neural Network (MPNN). The Message Passing Neural Network is a formal framework for spatial graph convolution. In the embodiments of the present application, the graph neural network uses the following neighborhood aggregation method to normalize the input directed graph structure matrix A:

[0068]

[0069] where H i is a matrix that contains the node representations of the previous layer. Initially, H 0 is set to the historical data representing the changes in infectious disease cases in each region. W i represents the trainable parameter matrix of the i-th layer, and f is a non-linear activation function, such as the ReLU function. Normalize the input directed graph structure matrix A so that the weighted in-edges of each node are equal to 1, and obtain the normalized directed graph structure matrix

[0070] The long short-term memory network is a special type of Recurrent Neural Network (RNN) that can be used to process sequential data. The long short-term memory network is mainly designed to solve the problems of vanishing gradients and exploding gradients during the training of long sequences. Use an MPNN at each time step to obtain a representation sequence h i,t-n , h i,t-n+1 ,..., h i,t-1 . Input these representation sequences into the long short-term memory network to extract the temporal features. The LSTM calculation formula is expressed as:

[0071] X i,t = LSTM(h i,t-n , h i,t-n+1 ,..., h i,t-1 ) (3)

[0072] where X i,t represents the predicted infectious disease case data of the i-th region at the t-th time period, and h i,t-1 represents the representation of input features such as the infectious disease case data of the i-th region at the (t-1)-th time period.

[0073] S6: Fuse the adjacency relationship, population movement relationship, and location attention relationship through the infectious disease spatio-temporal prediction model to obtain the urban infectious disease prediction result;

[0074] Among them, in order to simultaneously consider the influence of the proximity relationship, population movement relationship, and location attention relationship on the prediction of urban infectious diseases, the present invention adopts a fusion method based on a parameter matrix, specifically as follows:

[0075]

[0076] Among them, W adj 、W od and W at are parameter matrices to be trained, and are the prediction results of infectious diseases at time t obtained based on the adjacency matrix, population movement flow matrix, and location attention matrix respectively. Tanh is the activation function, is the final prediction result of the entire infectious disease spatio-temporal prediction model at time t.

[0077] Since the embodiments of the present application use artificial intelligence deep learning to predict the trend of infectious diseases, the model parameters can be updated by learning new data after multiple predictions, making the model more intelligent and efficient.

[0078] Based on the above, the urban epidemic spatio-temporal prediction method of the embodiments of the present application extracts the urban population movement flow through the individual movement trajectory data within the city, divides the population movement flow within a certain time period according to the time attribute, obtains the proximity relationship, population movement relationship, and location attention relationship between regions based on the population movement flow with different time attributes, and constructs an infectious disease spatio-temporal prediction model based on the graph neural network and long short-term memory network according to the proximity relationship, population movement relationship, and location attention relationship. The infectious disease spatio-temporal prediction model makes a refined prediction of the infectious disease trend. By dividing the population movement flow according to the time attribute, the embodiments of the present application can accurately capture the impact of population interaction and movement under different time attributes on the spread and transmission of infectious diseases within the city, fully utilize the local spatial context information of each region, and consider various spatial relationships between regions, so as to better model the time dependence and spatial dependence in the infectious disease case data, greatly improve the spatial perception ability of the infectious disease spatio-temporal prediction model and the model prediction performance, realize the prediction of the development trend of infectious diseases within the city with a higher spatial resolution, complete the refined analysis of the infectious disease epidemic, help the government and public health departments timely and accurately understand the development trend of infectious diseases within the city, and carry out targeted epidemic prevention and control interventions, which can maximize the protection of people's life and health safety.

[0079] Please refer to Figure 2 , which is a schematic structural diagram of the urban epidemic spatio-temporal prediction system of the embodiments of the present application. The urban epidemic spatio-temporal prediction system 40 of the embodiments of the present application includes:

[0080] Data collection module 41: used to collect individual movement trajectory data and infectious disease case data within the city;

[0081] Flow calculation module 42: used to process individual movement trajectory data through a data-driven method, extract the population movement flow between regions, divide the population movement flow according to time attributes, and obtain the population movement relationship under each time attribute based on the proximity relationship between regions;

[0082] Similarity calculation module 43: used to calculate the similarity of infectious disease case data between regions using a histogram-based similarity algorithm, and obtain the location attention relationship between regions according to the similarity of infectious disease case data;

[0083] Model construction module 44: used to construct an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship, and location attention relationship;

[0084] Infectious disease prediction module 45: used to input the infectious disease case data within the city into the infectious disease spatio-temporal prediction model, and obtain the infectious disease prediction results within the city through the infectious disease spatio-temporal prediction model.

[0085] Please refer to Figure 3 , which is a schematic diagram of the terminal structure according to an embodiment of the present application. The terminal 50 includes a processor 51 and a memory 52 coupled to the processor 51.

[0086] The memory 52 stores program instructions for implementing the above-mentioned urban epidemic spatio-temporal prediction method.

[0087] The processor 51 is used to execute the program instructions stored in the memory 52 to control the urban epidemic spatio-temporal prediction.

[0088] Among them, the processor 51 can also be called a CPU (Central Processing Unit, central processing unit). The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0089] Please refer to Figure 4, which is a schematic structural diagram of the storage medium according to the embodiments of the present application. The storage medium according to the embodiments of the present application stores a program file 61 that can implement all the above methods. Among them, the program file 61 can be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods according to various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, or terminal devices such as computers, servers, mobile phones, and tablets.

[0090] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined in the present application can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown in the present application, but will be accorded the widest scope consistent with the principles and novel features disclosed in the present application.

Claims

1. A method for spatio-temporal prediction of urban epidemic situations, characterized in that, Including: Collecting individual movement trajectory data and infectious disease case data within the city; Processing the individual movement trajectory data by a data-driven method, extracting the population movement flow between regions, dividing the population movement flow according to time attributes, and obtaining the population movement relationship under each time attribute based on the proximity relationship between regions; Calculating the similarity of infectious disease case data between regions by using a histogram-based similarity algorithm, and obtaining the position attention relationship between regions according to the similarity of the infectious disease case data; Constructing an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship and position attention relationship; Inputting the infectious disease case data within the city into the infectious disease spatio-temporal prediction model, and obtaining the infectious disease prediction result within the city through the infectious disease spatio-temporal prediction model; The constructing of the infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship and position attention relationship includes: Inputting the graph structure into the graph neural network, and the graph neural network normalizes the directed graph by using a neighborhood aggregation method so that the weighted incoming edges of each node are equal to 1; Among them, H i is a matrix that contains the node representations of the previous layer. The initial H 0 is set to the historical data representing the changes in the number of infectious disease cases in each region. W i represents the trainable parameter matrix of the i-th layer, and f is a non-linear activation function.

2. The urban epidemic spatio-temporal prediction method according to claim 1, wherein The dividing of the population movement flow according to time attributes and obtaining the population movement relationship under each time attribute based on the proximity relationship between regions specifically are: The time attributes include weekdays, weekends or holidays; the population movement flow of each time attribute is respectively converted into a weighted directed graph G=(V,E), where V represents the set of nodes, E represents the set of edges, the vertices represent the regions within the city, and the edges are used to capture the movement patterns; and based on the proximity relationship between regions, it is determined whether there is an edge between two nodes in the directed graph according to whether the two regions are in contact with each other, and the adjacency matrix of the population movement relationship corresponding to the proximity relationship is obtained.

3. The urban epidemic spatio-temporal prediction method according to claim 2, characterized in that The calculating of the similarity of infectious disease case data between regions by using a histogram-based similarity algorithm and obtaining the position attention relationship between regions according to the similarity of the infectious disease case data includes: Constructing corresponding histograms for the infectious disease case data of each region respectively; Calculating the similarity of infectious disease case data between two regions. If the similarity of infectious disease case data between two regions is higher than the set threshold, it is considered that the infectious disease outbreak trends of the two regions are similar and correlated, and an edge on the graph is constructed between the two regions in the directed graph, generating an adjacency matrix representing the position attention relationship between all regions.

4. The urban epidemic spatio-temporal prediction method according to claim 3, wherein The calculation formula of the position attention relationship is: where θ is the threshold for establishing connections using the histogram-based similarity algorithm; when the similarity w i,j between region i and region j is higher than the threshold θ, an edge is created between region i and region j, and finally an adjacency matrix of the attention relationships at the corresponding positions of each region within the city is obtained.

5. The urban epidemic spatio-temporal prediction method according to claim 4, characterized in that, The constructing of the infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship and position attention relationship further includes: At each time step, a message propagation neural network is used to obtain a representation sequence h i,t-n , h i,t-n+1 ,..., h i,t-1 . The representation sequence h i,t-n , h i,t-n+1 ,..., h i,t-1 is input into a long short-term memory network to extract the time series relationship therein; the calculation formula of the long short-term memory network is: X i,t = LSTM(h i,t-n , h i,t-n+1 ,..., h i,t-1 ) Among them, X i,t represents the predicted infectious disease case data of the i-th region in the t-th time period, h i,t-1 represents the infectious disease case data of the i-th region in the (t - 1)-th time period.

6. The urban epidemic spatio-temporal prediction method according to claim 5, wherein, The inputting of the infectious disease case data within the city into the infectious disease spatio-temporal prediction model and obtaining the infectious disease prediction result within the city through the infectious disease spatio-temporal prediction model specifically are: Fusing the proximity relationship, population movement relationship and position attention relationship through the infectious disease spatio-temporal prediction model to obtain the urban infectious disease prediction result; the specific fusion method is: Among them, W adj , W od and W at are parameter matrices to be trained. and are the infectious disease prediction results at time t obtained based on the adjacency matrix, population movement flow matrix, and position attention matrix respectively. Tanh is the activation function. is the final prediction result of the entire infectious disease spatio-temporal prediction model at time t.

7. A city epidemic spatio-temporal prediction system using the city epidemic spatio-temporal prediction method described in claim 1, characterized in that, Including: Data collection module: used to collect individual movement trajectory data and infectious disease case data within the city; Flow calculation module: used to process the individual movement trajectory data through a data-driven method, extract the population movement flow between regions, divide the population movement flow according to time attributes, and obtain the population movement relationship under each time attribute based on the proximity relationship of regions; Similarity calculation module: used to calculate the similarity of infectious disease case data between regions using a histogram-based similarity algorithm, and obtain the location attention relationship between regions according to the similarity of the infectious disease case data; Model construction module: used to construct an infectious disease spatio-temporal prediction model based on a graph neural network and a long short-term memory network according to the population movement relationship, proximity relationship, and location attention relationship; Infectious disease prediction module: used to input the infectious disease case data within the city into the infectious disease spatio-temporal prediction model, and obtain the infectious disease prediction result within the city through the infectious disease spatio-temporal prediction model.

8. A terminal, characterized in that, The terminal includes a processor and a memory coupled to the processor, where The memory stores program instructions for implementing the urban epidemic spatio-temporal prediction method according to any one of claims 1-6; The processor is used to execute the program instructions stored in the memory to control the urban epidemic spatio-temporal prediction.

9. A storage medium, characterized in that, Stores program instructions that can be run by a processor, and the program instructions are used to execute the urban epidemic spatio-temporal prediction method according to any one of claims 1 to 6.

Images