Infectious disease prediction method and system based on global shipping and aviation network

Abstract

The invention relates to an infectious disease prediction method and system based on a global shipping and aviation network, and the method comprises the following steps: obtaining shipping information and aviation information, and constructing a shipping network model and an aviation network model; vectorizing the shipping network model and the aviation network model through a heterogeneous graph embedding algorithm to obtain shipping prediction data and aviation prediction data; obtaining entry data of the asymptomatic infected person and the patient and ballast water discharge data of the shipping ship according to the prediction data; integrating objective data of an entry country, entry data of an asymptomatic infected person and a patient and ballast water discharge data of a shipping ship to obtain a spatial high-dimensional feature vector of a multi-dimensional feature; and inputting the spatial high-dimensional feature vector into an infectious disease prediction model to obtain an infectious disease prediction result. Compared with the prior art, the infectious disease transmission condition is comprehensively considered from the two aspects of shipping and aviation, the external environment of virus transmission is comprehensively evaluated, and the accuracy of infectious disease prediction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of shipping and aviation network, and in particular to an infectious disease prediction method and system based on the global shipping and aviation network. Background Art

[0002] The global shipping and aviation network plays a crucial role in promoting the world economic development, and its influence spreads across multiple key fields such as economy, culture, politics and environment. However, the spread of infectious diseases globally will have a significant impact on global trade and shipping logistics, severely weakening the operating efficiency of the global shipping network. Whether passengers carry viruses on board ships or airplanes during voyages, or the discharge of ship ballast water, it may trigger the spread of infectious diseases, posing potential catastrophic risks to ports and even entire countries. Given the microscopic characteristics and strong transmissibility of viruses, it is difficult for current technical means to achieve a complete screening of all virus carriers, which poses a severe threat to port security and national security.

[0003] Currently, in the face of the complex real situation, the existing models have limitations in terms of considered factors. The parameter settings of early models often lack sufficient basis, which restricts the applicability of the models and the accuracy of predictions, resulting in a significant difference between the prediction results and the actual epidemic situation. In addition, the differences in prevention and control mechanisms among countries also increase the complexity of model predictions.

[0004] Traditional prediction models often fail to comprehensively consider the two major virus transmission routes of shipping and aviation. For example, the patent application CN116344065A discloses a method and system for analyzing the risk of disease transmission with ships, including the following steps: obtaining the spatio-temporal data of N ships in the AIS system within a specific water area; determining the berthing time of the N ships according to the spatio-temporal data of the N ships; determining the ship infection risk value during the period when the N ships berth at the output port according to the berthing time; constructing a maritime route topological network according to the data information in the AIS system; determining the epidemic risk value of the input port according to the ship infection risk value and the maritime route topological network; determining the epidemic transmission ductility calculation value according to the epidemic risk value of the input port; this method only considers the epidemic risk situation from the perspective of water transportation, but ignores the key role of the aviation network in virus transmission. At the same time, some models fail to comprehensively evaluate the external environment of virus transmission, such as the influence of the temperature and humidity of a country on the characteristics of the virus, and whether the country's sanitation conditions and economic strength are sufficient to cope with sudden epidemics. The current systems have deficiencies in the acquisition of monitoring data, early warning methods, timeliness, and data analysis of newly emerging infectious diseases, and fail to make full use of big data and advanced modeling and analysis technologies, which restricts the effectiveness and response ability of the early warning system. Summary of the Invention

[0005] The objective of the present invention is to overcome the deficiencies of the above-mentioned existing technologies and provide an infectious disease prediction method and system based on the global shipping and aviation networks, which comprehensively consider the spread of infectious diseases from both the shipping and aviation aspects, and comprehensively evaluate the external environment of virus transmission, thereby improving the accuracy of infectious disease prediction.

[0006] The objective of the present invention can be achieved through the following technical solutions:

[0007] An infectious disease prediction method based on the global shipping and aviation networks, comprising the following steps:

[0008] Obtain shipping information and aviation information, and respectively construct a shipping network model and an aviation network model according to the shipping information and aviation information;

[0009] Respectively extract the network features of the shipping network model and the aviation network model through a heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model to obtain shipping prediction data and aviation prediction data;

[0010] Obtain the entry data of asymptomatic infected persons and patients according to the shipping prediction data and aviation prediction data, and obtain the ballast water discharge data of shipping vessels according to the shipping prediction data;

[0011] Obtain the objective data of the entry country and preprocess the objective data;

[0012] Integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features;

[0013] Input the spatial high-dimensional feature vector into an infectious disease prediction model to obtain an infectious disease prediction result.

[0014] Further, the shipping network model and the aviation network model are heterogeneous graphs, the nodes of the shipping network model are ports, and the edges are shipping routes and ship type classifications, and the nodes of the aviation network model are airports, and the edges are flight routes and aircraft type classifications.

[0015] Further, the specific steps of extracting the network features of the shipping network model and the aviation network model through a heterogeneous graph embedding algorithm and vectorizing the shipping network model and the aviation network model include:

[0016] Initialize the Metapath2Vec model and set basic parameters;

[0017] Define the connection information between the nodes of the shipping network model and the aviation network model respectively through an edge index dictionary, and construct the topological structures of the shipping network model and the aviation network model;

[0018] Define meta-paths according to the topological structure, guide the random walk process of the Metapath2Vec model, set the maximum number of steps of the random walk and the context window size of each node, capture the semantic information between the nodes, and obtain a node sequence;

[0019] Use the node sequence as input and obtain the embedding vectors of the nodes through the Skip-Gram model.

[0020] Further, the basic parameters include the dimension of the embedding vector, the learning rate, and the optimizer. The meta-paths of the shipping network model include port-ship type-port and port-route-port. The meta-paths of the aviation network model include airport-aircraft type-airport and airport-route-airport.

[0021] Further, the Skip-Gram model updates the embedding vectors through the stochastic gradient descent algorithm to minimize the loss function, thereby learning the optimal embedding representation of the nodes.

[0022] Further, the objective data includes the average of the highest temperature throughout the year, the average annual precipitation, the annual GDP, and the medical and health level of the entry country.

[0023] Further, the steps for preprocessing the objective data include data cleaning, data screening, data filling, data frequency conversion, and data normalization.

[0024] Further, the specific steps for integrating the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels include:

[0025] Simulate the infectious disease transmission path according to the route and the entry data of asymptomatic infected persons and patients;

[0026] Incorporate the ballast water discharge into the infectious disease transmission path according to the ballast water discharge data of shipping vessels, and the ballast water discharge data of shipping vessels is obtained according to the ship type and the number of shipping vessels of different ship types and the situation of ship ballast water discharge.

[0027] Further, the infectious disease prediction model is an LSTM model.

[0028] According to another aspect of the present invention, there is provided an infectious disease prediction system based on the global shipping and aviation networks, characterized by comprising:

[0029] A network model construction module for obtaining shipping information and aviation information, and respectively constructing a shipping network model and an aviation network model according to the shipping information and the aviation information;

[0030] A prediction data acquisition module, which is used to extract the network features of the shipping network model and the aviation network model respectively through a heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model, and obtain shipping prediction data and aviation prediction data;

[0031] A shipping vessel ballast water discharge data acquisition module, which is used to obtain the entry data of asymptomatic infected persons and patients according to the shipping prediction data and the aviation prediction data, and obtain the ballast water discharge data of shipping vessels according to the shipping prediction data;

[0032] An objective data and processing module, which is used to obtain the objective data of the entry country and preprocess the objective data;

[0033] A data integration module, which is used to integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features;

[0034] An infectious disease prediction module, which is used to input the spatial high-dimensional feature vector into an infectious disease prediction model to obtain an infectious disease prediction result.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The present invention respectively vectorizes the shipping network model and the aviation network model through a heterogeneous graph embedding algorithm to obtain shipping prediction data and aviation prediction data, accurately extracts feature vectors in the shipping network and the aviation network, obtains the entry data of asymptomatic infected persons and patients and the ballast water discharge data of shipping vessels according to the prediction data, and effectively predicts the potential spread of infectious diseases through LSTM based on the spatial high-dimensional feature vector with multi-dimensional features. Its practical application realizes the real-time monitoring and analysis of shipping data and aviation data, predicts the diffusion trend and potential risks of infectious diseases, and provides timely warning information for public health decision-makers.

[0037] 2. The present invention obtains shipping information and aviation information, constructs a shipping network model and an aviation network model respectively according to the shipping information and the aviation information; and comprehensively considers various objective influencing factors to integrate the objective data of the entry country, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features, comprehensively considers the spread of infectious diseases from both aspects of shipping and aviation, and comprehensively evaluates the external environment of virus transmission, thereby improving the accuracy of infectious disease prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic flowchart of a method for predicting infectious diseases based on a global shipping and aviation network proposed by the present invention;

[0039] Figure 2 The figure is a schematic diagram of the process of preprocessing objective data in the present invention. DETAILED DESCRIPTION

[0040] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0041] Abbreviations involved:

[0042] Automatic Identification System, AIS

[0043] Long Short-Term Memory Network: Long Short-Term Memory, LSTM

[0044] Stochastic Gradient Descent: Stochastic Gradient Descent, SGD

[0045] Example 1

[0046] This embodiment provides a method for predicting infectious diseases based on global shipping and aviation networks. Figure 1 As shown, the following steps are included:

[0047] S1. Obtain shipping information and aviation information, and construct a shipping network model and an aviation network model according to the shipping information and aviation information respectively.

[0048] Through ship call data, with ports as nodes and routes and ship type classifications as edges, a shipping chain between ports and countries is constructed to obtain a shipping network model.

[0049] Through aircraft approach data, with airports as nodes and routes and aircraft types as edges, an aviation chain between airports and countries is constructed to obtain an aviation network model.

[0050] The shipping network model and the aviation network model are heterogeneous graphs. The steps for building the shipping network model include: importing the NetworkX library to build a heterogeneous graph and create a heterogeneous graph G; adding the port list as nodes to the graph G; adding the route list as edges to the graph G; and adding the ship type classification as edges to the graph G. The steps for building the aviation network model include: importing the NetworkX library to build a heterogeneous graph and create a heterogeneous graph G; adding the airport list as nodes to the graph G; adding the route list as edges to the graph G; and adding the aircraft type classification as edges to the graph G.

[0051] S2. Respectively extract the network features of the shipping network model and the aviation network model through the heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model, and obtain shipping prediction data and aviation prediction data.

[0052] The specific steps of extracting the network features of the shipping network model and the aviation network model through the heterogeneous graph embedding algorithm and vectorizing the shipping network model and the aviation network model are as follows:

[0053] Initialize the Metapath2Vec model and set the basic parameters, including the embedding vector dimension, learning rate, and optimizer.

[0054] Define the connection information between the nodes of the shipping network model and the aviation network model through the edge index dictionary respectively, and construct the topological structures of the shipping network model and the aviation network model.

[0055] Set the embedding vector dimension to 128 to ensure that the representation of each node or edge is a 128-dimensional vector.

[0056] Define metapaths according to the topological structure to guide the random walk process of the Metapath2Vec model. The metapaths of the shipping network model include port - ship type - port and port - route - port, and the metapaths of the aviation network model include airport - aircraft type - airport and airport - route - airport. Set the maximum number of steps of the random walk to 10 to control the length of the random walk. Set the context window size of each node to 3, which means considering the nearest 3 nodes around each node to capture the semantic information between nodes and obtain the node sequence. Use the Skip-Gram model to train the node sequence with the node sequence as the input, so that the embedding vector of each node can maximize the probability of the occurrence of its context nodes. The specific steps of training are as follows: conduct 100 epochs of training, clear the gradients with zero_grad each time an epoch is performed, calculate the loss, and backpropagate the gradients.

[0057] The Skip-Gram model updates the embedding vector through the Stochastic Gradient Descent (SGD) algorithm to minimize the loss function, thereby learning the optimal embedding representation of the nodes. The optimizer is the SparseAdam optimizer, and the learning rate is 0.01.

[0058] Use the obtained embedding vectors of the nodes as the prediction data.

[0059] S3. Obtain the entry data of asymptomatic infected persons and patients based on the shipping prediction data and the aviation prediction data, and obtain the ballast water discharge data of shipping vessels based on the shipping prediction data.

[0060] Obtain the entry data of asymptomatic infected persons and patients based on the shipping prediction data and the aviation prediction data.

[0061] Obtain the number of shipping vessels of different types in the shipping network model based on shipping forecast data. Considering different ship types and the ballast water discharge of vessels with different quantities, incorporate the ballast water discharge into a transmission path of infectious disease viruses. The ballast water discharge data of shipping vessels is obtained based on the ship type, the number of shipping vessels of different types, and the situation of ballast water discharge of vessels.

[0062] S4. Obtain the objective data of the entry country and preprocess the objective data.

[0063] The objective data includes the average value of the annual highest temperature, the average value of the annual precipitation, the annual GDP, and the medical and health level of the entry country.

[0064] Considering the impact of temperature on infectious disease viruses, take the average value of the annual highest temperature of the entry country;

[0065] Considering the impact of humidity on infectious disease viruses, take the average value of the annual precipitation of the entry country;

[0066] Considering the impact of the country's economic conditions on whether it is sufficient to resist virus invasion, take the annual GDP of the required country;

[0067] Considering the impact of the country's health conditions on whether it is sufficient to resist virus invasion, obtain the medical and health level of the entry country through the WHO.

[0068] The steps for preprocessing the objective data are as Figure 2 shown, including data cleaning, data screening, data filling, data frequency conversion, and data normalization.

[0069] S5. Integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a high-dimensional feature vector of multi-dimensional features in space.

[0070] The specific steps for integrating the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels include:

[0071] Based on the shipping routes and the entry data of asymptomatic infected persons and patients, simulate the transmission path of infectious diseases;

[0072] Incorporate the ballast water discharge into the infectious disease transmission path according to the ballast water discharge data of shipping vessels. The ballast water discharge data of shipping vessels is obtained based on the ship type, the number of shipping vessels of different types, and the situation of ballast water discharge of vessels;

[0073] Integrate the ballast water discharge data of shipping vessels, the preprocessed objective data, and the entry data of asymptomatic infected persons and patients to obtain a high-dimensional feature vector of multi-dimensional features in space.

[0074] S6. Input the spatial high-dimensional feature vector into the infectious disease prediction model to obtain the infectious disease prediction result.

[0075] The infectious disease prediction model is an LSTM model. The hidden state calculated based on the internal cell state of the LSTM processes time series data, captures long-term dependencies, and predicts future trends based on this information, forming an infectious disease prediction model based on the global shipping and aviation networks. The predicted number of infectious disease cases is finally obtained through the infectious disease prediction model according to the spatial high-dimensional feature vector.

[0076] Embodiment 2

[0077] This embodiment provides an infectious disease prediction system based on the global shipping and aviation networks, including:

[0078] A network model construction module, configured to obtain shipping information and aviation information, and respectively construct a shipping network model and an aviation network model according to the shipping information and aviation information;

[0079] A prediction data acquisition module, configured to respectively extract the network features of the shipping network model and the aviation network model through the heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model to obtain shipping prediction data and aviation prediction data;

[0080] A shipping vessel ballast water discharge data acquisition module, configured to obtain the entry data of asymptomatic infected persons and patients according to the shipping prediction data and the aviation prediction data, and obtain the ballast water discharge data of shipping vessels according to the shipping prediction data;

[0081] An objective data and processing module, configured to obtain the objective data of the entry country and preprocess the objective data;

[0082] A data integration module, configured to integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features;

[0083] An infectious disease prediction module, configured to input the spatial high-dimensional feature vector into the infectious disease prediction model to obtain the infectious disease prediction result.

[0084] The rest is the same as Embodiment 1.

[0085] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations according to the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of the present invention through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should be within the protection scope determined by the claims.

Claims

1. An infectious disease prediction method based on the global shipping and aviation networks, characterized in that, It includes the following steps: Obtain shipping information and aviation information, and respectively construct a shipping network model and an aviation network model according to the shipping information and aviation information; Extract the network features of the shipping network model and the aviation network model respectively through the heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model, and obtain shipping prediction data and aviation prediction data; Obtain the entry data of asymptomatic infected persons and patients according to the shipping prediction data and aviation prediction data, and obtain the ballast water discharge data of shipping vessels according to the shipping prediction data; Obtain the objective data of the entry country and preprocess the objective data; Integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features; Input the spatial high-dimensional feature vector into an infectious disease prediction model to obtain an infectious disease prediction result.

2. The infectious disease prediction method based on the global shipping and aviation network according to claim 1, wherein The shipping network model and the aviation network model are heterogeneous graphs. The nodes of the shipping network model are ports, and the edges are shipping routes and ship type classifications. The nodes of the aviation network model are airports, and the edges are flight routes and aircraft type classifications.

3. The method for predicting infectious diseases based on the global shipping and aviation network according to claim 1, wherein The specific steps of extracting the network features of the shipping network model and the aviation network model through the heterogeneous graph embedding algorithm and vectorizing the shipping network model and the aviation network model include: Initialize the Metapath2Vec model and set basic parameters; Define the connection information between the nodes of the shipping network model and the aviation network model respectively through an edge index dictionary, and construct the topological structures of the shipping network model and the aviation network model; Define meta-paths according to the topological structures, guide the random walk process of the Metapath2Vec model, set the maximum number of steps of the random walk and the context window size of each node, capture the semantic information between the nodes, and obtain a node sequence; Use the node sequence as input and obtain the embedding vector of the node through the Skip-Gram model.

4. The infectious disease prediction method based on the global shipping and aviation network according to claim 3, wherein The basic parameters include the dimension of the embedding vector, the learning rate, and the optimizer. The meta-paths of the shipping network model include port-ship type-port and port-route-port. The meta-paths of the aviation network model include airport-aircraft type-airport and airport-route-airport.

5. The infectious disease prediction method based on the global shipping and aviation network according to claim 3, characterized in that, The Skip-Gram model updates the embedding vector through the stochastic gradient descent algorithm to minimize the loss function, thereby learning the optimal embedding representation of the node.

6. The infectious disease prediction method based on the global shipping and aviation network according to claim 1, wherein The objective data includes the average of the highest temperature throughout the year, the average annual precipitation, the annual GDP, and the medical and health level of the entry country.

7. The method for predicting infectious diseases based on the global shipping and aviation network according to claim 1, wherein The steps of preprocessing the objective data include data cleaning, data screening, data filling, data frequency conversion, and data normalization.

8. The method for predicting infectious diseases based on the global shipping and aviation network according to claim 1, wherein The specific steps of integrating the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of shipping vessels include: Simulate the infectious disease transmission path according to the flight route and the entry data of asymptomatic infected persons and patients; The ballast water discharge of the shipping vessels is included in the infectious disease transmission path according to the ballast water discharge data of the shipping vessels, and the ballast water discharge data of the shipping vessels is obtained according to the ship type, the number of shipping vessels of different ship types, and the ballast water discharge situation of the vessels.

9. The infectious disease prediction method based on the global shipping and aviation network according to claim 1, wherein The infectious disease prediction model is an LSTM model.

10. An infectious disease prediction system based on the global shipping and aviation networks, characterized in that, It includes: A network model construction module, configured to obtain shipping information and aviation information, and respectively construct a shipping network model and an aviation network model according to the shipping information and the aviation information; A prediction data acquisition module, configured to respectively extract the network features of the shipping network model and the aviation network model through a heterogeneous graph embedding algorithm, vectorize the shipping network model and the aviation network model, and obtain shipping prediction data and aviation prediction data; A ballast water discharge data acquisition module of shipping vessels, configured to obtain the entry data of asymptomatic infected persons and patients according to the shipping prediction data and the aviation prediction data, and obtain the ballast water discharge data of the shipping vessels according to the shipping prediction data; An objective data and processing module, configured to obtain the objective data of the entry country and preprocess the objective data; A data integration module, configured to integrate the preprocessed objective data, the entry data of asymptomatic infected persons and patients, and the ballast water discharge data of the shipping vessels to obtain a spatial high-dimensional feature vector with multi-dimensional features; An infectious disease prediction module, configured to input the spatial high-dimensional feature vector into an infectious disease prediction model to obtain an infectious disease prediction result.

Images