A social graph network-based animal disease early warning method and system

Abstract

The application discloses an early warning method and system for animal diseases based on a social graph network, wherein the early warning method for animal diseases based on the social graph network comprises the following steps: collecting health information and position information of a plurality of animals in real time; modeling the health information of the plurality of animals by using a long short-term memory (LSTM) model to obtain current health state information of the plurality of animals; generating a social graph network of the plurality of animals according to the position information of the plurality of animals; modeling the social graph network by using a graph neural network algorithm to obtain a social graph network model, and initializing the social graph network model by using the current health state information; and obtaining node features of the social graph network model, and early warning diseases of each animal according to the node features. The technical scheme of the application can solve the problems in the prior art that it is difficult to timely early warn the health conditions of animals and it is impossible to early warn the infection of animal diseases.

Description

Technical Field

[0001] This invention relates to the field of neural network technology, and in particular to an early warning method and system for animal diseases based on social graph networks. Background Technology

[0002] With societal development, more and more people are keeping pets. During pet ownership, due to contact between pets and the tendency for pathogens to spread, the spread of infectious diseases among pets is unavoidable. Because pet illnesses are often difficult to detect or are detected late, they can negatively impact a pet's physical and mental health. Therefore, a disease early warning system capable of detecting signs of animal illness promptly or even in advance is of great importance for early warning of animal diseases.

[0003] Current technologies for early warning of animal diseases typically employ monitoring various indicators in pets (such as body temperature, blood oxygen, or blood pressure). Specific disease monitoring methods include: 1. Rule-based methods: The disease warning system directly alarms when physiological indicators become abnormal, i.e., when they exceed the normal range; 2. Sequence-based methods: By monitoring the sequences of physiological indicators, methods such as Markov chains are used to detect the occurrence of sequence anomalies; 3. Model-based methods: By collecting sequences of physiological indicators, sequence models such as RNNs are trained, and the models are used to identify anomalies and even make early predictions based on the indicator sequences.

[0004] For animals, daily activities often involve contact with other animals, leading to interactions such as bacterial infections and viral transmission. However, the methods mentioned above typically only consider a single physiological indicator of the animal's condition and can only issue an alarm when the pet's health condition becomes abnormal, making it difficult to provide timely warnings and thus failing to provide true early warnings of the spread of animal diseases. Summary of the Invention

[0005] This invention provides an animal disease early warning method and system based on social graph networks, aiming to solve the problem that existing technologies only consider a single physiological indicator of the animal's own condition, and can only issue alarms when the pet's own health condition is abnormal, making it difficult to provide timely early warnings and thus failing to provide true early warning of the infectious behavior of animal diseases.

[0006] To achieve the above objectives, according to a first aspect of the present invention, an early warning method for animal diseases based on social graph networks is proposed, comprising:

[0007] Real-time collection of health and location information for multiple animals;

[0008] The health information of multiple animals was modeled using a Long Short-Term Memory (LSTM) model to obtain the current health status information of multiple animals.

[0009] Generate a social graph network of multiple animals based on their location information;

[0010] The social graph network is modeled using a graph neural network algorithm to obtain a social graph network model, and the social graph network model is initialized using the current health status information.

[0011] Obtain node features from a social graph network model and use these features to provide disease warnings for each animal.

[0012] Preferably, in the above-mentioned early warning method for animal diseases, the step of collecting real-time health and location information of multiple animals includes:

[0013] Set a unique identifier for each of the multiple animals;

[0014] The body temperature sensor is used to collect the body temperature index sequence of each animal in real time as health information for each animal;

[0015] Use position sensors to collect the location information of each animal in real time;

[0016] Establish a unique identifier that corresponds to both health information and location information.

[0017] Preferably, in the above-mentioned animal disease early warning method, the step of generating a social graph network of multiple animals based on the location information of multiple animals includes:

[0018] Set up multiple nodes, where each node represents a different animal;

[0019] Based on the location information of multiple animals, identify the contact between multiple animals;

[0020] Based on the contact between multiple animals, generate edges between multiple nodes and set the weights of the edges;

[0021] By updating the nodes, edges, and edge weights in real time, a social graph network of multiple animals is obtained.

[0022] Preferably, in the above-mentioned early warning method for animal diseases, the steps of using a graph neural network algorithm to model a social graph network and initializing the social graph network model using current health status information include:

[0023] The information from the social graph network is input into the message passing algorithm for modeling, resulting in the social graph network model.

[0024] Using current health status information from multiple animals, initialize the node features of each node in the social graph network model.

[0025] Preferably, in the above-mentioned animal disease early warning method, the step of obtaining node features of the social graph network model and performing disease early warning for each animal based on the node features includes:

[0026] Based on the node feature information of each initialized node, perform multi-layer convolution operations on the social graph network model;

[0027] Obtain the node feature information of each node in the last convolutional layer of the social graph network model;

[0028] The node feature information of each node in the last convolutional layer is input into the fully connected layer of the message passing algorithm to obtain the disease warning information for each animal.

[0029] Preferably, in the above-mentioned early warning method for animal diseases, the step of performing multi-layer convolutional operations on the social graph network model based on the node feature information of each initialized node includes:

[0030] Using the node feature information of each initialized node, according to the formula The node feature information of each node in each convolutional layer of the social graph network model is obtained sequentially; where,

[0031] For the node feature information of the i-th node in the m-th convolutional layer of a social graph network model at time t, w m Let m be the parameters of the m-th convolutional layer. The node feature information of all neighbor nodes j of the i-th node in the (m-1)-th convolutional layer of the social graph network model at time t.

[0032] According to a second aspect of the present invention, the present invention provides an early warning system for animal diseases based on social graph networks, comprising:

[0033] The information acquisition module is used to collect health and location information of multiple animals in real time;

[0034] The first modeling module is used to model the health information of multiple animals using a Long Short-Term Memory (LSTM) model, and obtain the current health status information of multiple animals.

[0035] The social graph network generation module is used to generate a social graph network of multiple animals based on their location information.

[0036] The second modeling module is used to model the social graph network using graph neural network algorithms, obtain the social graph network model, and initialize the social graph network model using the current health status information;

[0037] The disease early warning module is used to obtain the node features of the social graph network model and to provide disease warnings for each animal based on the node features.

[0038] Preferably, in the above-mentioned early warning system for animal diseases, the information collection module includes:

[0039] The identifier setting submodule is used to set a unique identifier for each of the multiple animals;

[0040] The body temperature acquisition submodule is used to collect the body temperature index sequence of each animal in real time using a body temperature sensor, as the health information of each animal;

[0041] The location acquisition submodule is used to collect the location information of each animal in real time using location sensors;

[0042] The relationship establishment submodule is used to establish the correspondence between the unique identifier and the health information and the location information, respectively.

[0043] Preferably, in the aforementioned animal disease early warning system, the social graph network generation module includes:

[0044] The node settings submodule is used to set up multiple nodes, which represent multiple animals;

[0045] The contact recognition submodule is used to identify the contact between multiple animals based on their location information.

[0046] The edge generation submodule is used to generate edges between multiple nodes and set the weights of the edges based on the contact situation of multiple animals.

[0047] The network acquisition submodule is used to update nodes, edges, and edge weights in real time to obtain the social graph network of multiple animals.

[0048] Preferably, in the above-mentioned early warning system for animal diseases, the second modeling module includes:

[0049] The algorithm input submodule is used to input the information of the social graph network into the message passing algorithm to learn and update the graph network model, and obtain the social graph network model.

[0050] The model initialization submodule is used to initialize the node features of each node in the social graph network model using the current health status information of multiple animals.

[0051] Preferably, in the above-mentioned animal disease early warning system, the disease early warning module includes:

[0052] The multi-layer convolution submodule is used to perform multi-layer convolution operations on the social graph network model based on the node feature information of each initialized node.

[0053] The feature acquisition submodule is used to acquire the node feature information of each node in the last convolutional layer of the social graph network model;

[0054] The feature input submodule is used to input the node feature information of each node in the last convolutional layer into the fully connected layer of the message passing algorithm to obtain the disease warning information for each animal.

[0055] In summary, the animal disease early warning scheme based on social graph networks provided in this application first collects the health and location information of multiple animals. This allows for the use of graph neural network algorithms to model the social graph network using the location information of multiple animals, resulting in a social graph network model that reflects the social relationships and intimacy among the animals. Furthermore, an LSTM model is used to model the health information of multiple animals, obtaining the current health status information of each animal to preliminarily determine whether the animal has a disease. The social graph network model is initialized using this current health status information, thus obtaining the initial node features of the social graph network model. The social graph network model is a neural network model capable of machine learning the current health status of each animal to determine its disease status, thereby providing disease warnings for each animal in the graph based on the node features. This animal disease early warning scheme based on social graph networks integrates animal physiological indicators and daily activities to provide early warnings of animal disease transmission, improving the accuracy of animal disease early warnings and reducing false positives and false negatives. This solves the problem that existing technologies struggle to provide timely warnings of animal health status and early warnings of animal disease transmission. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0057] Figure 1 This is a schematic diagram of the structure of a social graph network model provided in an embodiment of the present invention;

[0058] Figure 2 This is a schematic flowchart of an animal disease early warning method based on social graph networks provided in an embodiment of the present invention;

[0059] Figure 3 yes Figure 2 The illustrated embodiment provides a flowchart of an animal information collection method;

[0060] Figure 4 yes Figure 2 The illustrated embodiment provides a flowchart of a method for generating a social graph network;

[0061] Figure 5 yes Figure 2 The illustrated embodiment provides a flowchart of a social graph network modeling method.

[0062] Figure 6 yes Figure 2 The illustrated embodiment provides a flowchart of a disease early warning method;

[0063] Figure 7 This is a schematic flowchart of an animal disease early warning method based on social graph networks provided in an embodiment of the present invention;

[0064] Figure 8 yes Figure 7 The illustrated embodiment provides a structural diagram of an information acquisition module;

[0065] Figure 9 yes Figure 7 The illustrated embodiment provides a structural diagram of a social graph network generation module;

[0066] Figure 10 yes Figure 7 The illustrated embodiment provides a structural diagram of a second modeling module;

[0067] Figure 11 yes Figure 7 The illustrated embodiment provides a structural schematic diagram of a disease early warning module.

[0068] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0069] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0070] The main technical problems solved by the embodiments of the present invention are as follows:

[0071] Animals are prone to coming into contact with other animals during their daily activities, which can lead to bacterial infections and viral transmission. However, existing methods for collecting animal physiological indicators usually only consider a single physiological indicator such as the animal's own condition. They can only issue an alarm when the pet's health condition is abnormal, making it difficult to provide timely warnings and thus failing to provide a true early warning of the spread of animal diseases.

[0072] To solve the above problem, see Figure 1This invention provides an animal disease early warning scheme based on social graph networks. By sampling animal health information (such as body temperature characteristics), a body temperature sampling sequence is obtained. This body temperature sampling sequence is then processed by a sequence model network for machine learning. The current health status information of each animal can be obtained through an LSTM network. Then, by collecting the location information of multiple animals, a social graph network of the above-mentioned animals is generated. Each node in the social graph network represents an animal. The node features of each animal are initialized using the current health status information. Multiple convolutions are performed on the social graph network model to obtain multiple convolutional layers. The node features of the last convolutional layer (layer3 in the figure) are used for binary classification to predict whether an animal is at risk of disease infection in advance, and to provide early disease warning for each animal.

[0073] To achieve the above objectives, please refer to Figure 2 , Figure 2 This is a flowchart illustrating an animal disease early warning method based on social graph networks provided in an embodiment of the present invention. Figure 2 As shown, this animal disease early warning method based on social graph networks includes:

[0074] S110: Real-time collection of health and location information of multiple animals. Health information includes information such as animal body temperature, blood oxygen, or blood pressure. In the following embodiments of this application, body temperature is used as the animal's health information, and location information is the real-time collection of the animal's coordinates. This allows the animal's movement trajectory and social status to be determined based on its coordinates.

[0075] Specifically, as a preferred embodiment, such as Figure 3 As shown, the steps for real-time collection of health and location information of multiple animals include:

[0076] S111: Set a unique identifier for each of the multiple animals; in this embodiment of the application, an animal ID is constructed as the unique identifier of the animal. Specifically, a chip can be implanted in the animal body, and the chip integrates the ID.

[0077] S112: Use a body temperature sensor to collect the body temperature index sequence of each animal in real time as the health information of each animal; the body temperature data collected in this embodiment is obtained by real-time sampling, so that a serialized body temperature index sequence can be obtained, thereby serving as the health information of the animal.

[0078] S113: Use position sensors to collect the position information of each animal in real time; the position information can be collected by multiple sensors, or by using the above-mentioned implanted chip to integrate positioning function, to obtain the animal's position coordinates in real time, thereby determining the animal's movement trajectory and finding the social relationship between multiple animals.

[0079] S114: Establish a unique identifier that corresponds to both health information and location information. Because an implanted chip is placed inside the animal, integrating the animal's ID, and capable of collecting body temperature sequences and location information in real time, a correspondence is established between the animal's relevant characteristics, facilitating identification.

[0080] After collecting health and location information of multiple animals in real time Figure 2 The illustrated early warning method for animal diseases also includes the following steps:

[0081] S120: Use a Long Short-Term Memory (LSTM) model to model the health information of multiple animals and obtain the current health status information of multiple animals.

[0082] In this embodiment, the sequence model uses an LSTM model, which is a recurrent model capable of reorganizing the health information of multiple animals, such as body temperature index sequences, according to the length of time memory. This allows for modeling the health information of each animal and obtaining current health status information. This current health status information includes the animal's current health condition, such as current body temperature characteristics. These current body temperature characteristics are obtained by the LSTM model processing the body temperature characteristics at each time point within a predetermined time period. Therefore, the obtained current health status information can more accurately predict the animal's health status. For example, the animal's body temperature index sequence is input into the LSTM model for sequence modeling, and the input at each time point is denoted as T. t The output is denoted as Y. t Y t To some extent, it represents the current health status of the animal node at the current moment.

[0083] S130: Generate a social graph network of multiple animals based on their location information;

[0084] As a preferred embodiment, such as Figure 4 As shown, the steps for generating a social graph network of multiple animals based on their location information include:

[0085] S131: Set multiple nodes, where multiple nodes represent multiple animals; in this embodiment of the application, the IDs of an animal and each animal it socially contacts can be recorded as nodes in the social graph network.

[0086] S132: Based on the location information of multiple animals, identify the contact situation between multiple animals; the contact situation between animals includes information such as the contact time and the number of contact times, and specifically, determine whether the animals are in contact based on whether the coordinates of the two animals in the above location information overlap.

[0087] S133: Based on the contact status of multiple animals, generate edges between multiple nodes and set the weights of the edges. Contact between animals is defined by setting an edge between them. If no contact occurs within a predetermined time, the edge is disconnected. The weights of the edges can be determined by the duration of contact and the number of contacts.

[0088] S134: Update nodes, edges, and edge weights in real time to obtain a social graph network of multiple animals.

[0089] The technical solution provided in this application embodiment involves implanting a chip in an animal. This implanted chip integrates a positioning function, collecting the animal's location information in real time, and then constructing an animal social graph network based on this location information. Specifically, multiple nodes are set, each represented by an animal's ID. Edges between nodes are established based on the contact between animals, and edge weights are assigned to each edge, thereby generating the animal's social graph network. The animal social graph network consists of nodes and edges, i.e., G = (V, E), where V represents a node, E represents an edge, and G represents the social graph network. Nodes (V): represent the pets, represented by the ID of the pet chip; Edges (E): represent the connections between animals. Whether animals have contact is determined by the overlap of their locations, and this is used to establish an edge. The contact time and frequency are used as the edge weights.

[0090] Because the animals move, this social graph is dynamic and changes over time. When animals are in contact, there are edges between the nodes, and when they are not in contact, the edges are broken.

[0091] After generating the social graph network of the aforementioned animals, Figure 2 The illustrated early warning method for animal diseases also includes the following steps:

[0092] S140: A graph neural network algorithm is used to model the social graph network, resulting in a social graph network model. The current health status information is then used to initialize the social graph network model. In this embodiment, the graph neural network algorithm employs the message-passing algorithm. This algorithm is used to model the previously obtained social graph network. Specifically, at time t, the social graph network information at that time is used to update the node information, reflecting the dynamic social state among animals. The aforementioned current health status information serves as the initial feature of the nodes in the social graph network model. Through continuous convolution of the social graph network model, the body temperature characteristics of each animal in the social graph network can be accurately obtained.

[0093] Specifically, such as Figure 5 As shown, the method for modeling social graph networks using graph neural network algorithms includes the following steps:

[0094] S141: Input the information of the social graph network into the message passing algorithm for learning and updating to obtain the social graph network model; the message passing algorithm is a mainstream graph neural network algorithm that can reflect the state of the social graph network in real time, that is, the social situation of each animal. Input the information of the social graph network, such as nodes, edges and edge weights, into the message passing algorithm for modeling to obtain the real-time social situation of animals.

[0095] S142: Using the current health status information of multiple animals, initialize the node features of each node in the social graph network model. By using the current health status information as the initial node features of each node in the social graph network model, the health status of each animal in the social graph can be obtained in real time as the social graph network model performs convolutional calculations, thereby providing timely and accurate early warnings of animal disease transmission.

[0096] After obtaining the social graph network model Figure 2 The animal disease early warning methods shown also include:

[0097] S150: Obtain the node features of the social graph network model, and provide disease warnings for each animal based on the node features. In this embodiment, the node features reflect the health status information of the animals, which is updated as animals come into contact. Based on the contact situation and health status of each animal, a binary classification algorithm using the social graph network model is used to obtain disease warning information for each animal.

[0098] Specifically, as a preferred embodiment, such as Figure 6 As shown, the steps for obtaining node features of the social graph network model and providing disease warnings for each animal based on these node features include:

[0099] S151: Perform multi-layer convolution operations on the social graph network model based on the node feature information of each initialized node.

[0100] In one preferred embodiment, the steps of performing multi-layer convolution operations on the social graph network model based on the node feature information of each initialized node are as follows:

[0101] Using the node feature information of each initialized node, according to the formula The node feature information of each node in each convolutional layer of the social graph network model is obtained sequentially; where,

[0102] For the node feature information of the i-th node in the m-th convolutional layer of a social graph network model at time t, w mLet m be the parameters of the m-th convolutional layer. Let j represent the node feature information of all neighboring nodes of the i-th node in the (m-1)-th convolutional layer of the social graph network model at time t, where j represents the neighboring nodes of node i.

[0103] See details Figure 1 The feature representation of each node in Layer 0 is as follows: That is, the output of the LSTM model at node i at time t, here we choose Y. 0 Let i represent the Layer0 convolutional layer, and let i represent the i-th node, which corresponds to circles 1-5 in layer0.

[0104] According to the above formula, the feature representation of each node in Layer 1 is as follows: Where j represents all neighboring nodes of node i at time t, and the total number of its neighbors is n; similarly, the feature representation of each node in Layer 2 is as follows: Where j represents all neighboring nodes of node i at time t, and the total number of neighbors is n.

[0105] Layer 3 and so on, thus obtaining all convolutional layers.

[0106] S152: Obtain the node feature information of each node in the last convolutional layer of the social graph network model.

[0107] S153: Input the node feature information of each node in the last convolutional layer into the fully connected layer of the message passing algorithm to obtain the disease warning information for each animal.

[0108] Assuming the last convolutional layer is Layer 3, the node feature information of each node in this convolutional layer, i.e. the health status, is input into the fully connected layer of the graph neural network, thereby obtaining disease warning information for each animal, so as to provide early warning of the spread of animal diseases.

[0109] In summary, the animal disease early warning method based on social graph networks provided in this application first collects the health and location information of multiple animals. This allows for the use of graph neural network algorithms to model the social graph network using the location information of multiple animals, resulting in a social graph network model that reflects the social relationships and intimacy among the animals. Furthermore, an LSTM model is used to model the health information of multiple animals, obtaining the current health status information of each animal to preliminarily determine whether the animal has a disease. The social graph network model is initialized using this current health status information, thus obtaining the initial node features of the social graph network model. The social graph network model is a neural network model capable of machine learning the current health status of each animal to determine its disease status, thereby providing disease early warning for each animal in the graph based on the node features. This animal disease early warning scheme based on social graph networks integrates animal physiological indicators and daily activities to provide early warning of animal disease transmission, improving the accuracy of animal disease early warning and reducing false positives and false negatives. This solves the problem that existing technologies struggle to provide timely early warnings of animal health status and early warnings of animal disease transmission.

[0110] Based on the same concept as the above method embodiments, this invention also proposes an animal disease early warning system based on social graph networks to implement the above method of this invention. Since the principle and method of solving the problem in this system embodiment are similar, it has at least all the beneficial effects brought about by the technical solutions of the above embodiments, and will not be described in detail here.

[0111] See Figure 7 , Figure 7 This is a schematic diagram illustrating the structure of an animal disease early warning system based on a social graph network, as provided in an embodiment of the present invention. Figure 7 As shown, the early warning system for this animal disease includes:

[0112] Information acquisition module 110 is used to collect health and location information of multiple animals in real time;

[0113] The first modeling module 120 is used to model the health information of multiple animals using a long short-term memory (LSTM) model to obtain the current health status information of multiple animals.

[0114] The social graph network generation module 130 is used to generate a social graph network of multiple animals based on the location information of multiple animals.

[0115] The second modeling module 140 is used to model the social graph network using a graph neural network algorithm to obtain a social graph network model, and to initialize the social graph network model using the current health status information.

[0116] The disease early warning module 150 is used to acquire the node features of the social graph network model and to provide disease early warning for each animal based on the node features.

[0117] In summary, the animal disease early warning system based on social graph networks provided in this application first collects the health and location information of multiple animals through the information collection module 110. Then, the social graph network generation module 130 uses the location information of multiple animals and a graph neural network algorithm to model the social graph network, obtaining a social graph network model that reflects the social relationships and intimacy of multiple animals. Furthermore, the first modeling module 120 uses an LSTM model to model the health information of multiple animals, obtaining the current health status information of each animal to preliminarily determine whether the animal has a disease. The social graph network model is initialized using the current health status information, thus obtaining the initial node features of the social graph network model. The social graph network model is a neural network model that can perform machine learning on the current health status of each animal to determine the disease status of each animal. Therefore, the disease early warning module 150 can provide disease warnings for each animal in the graph based on the node features. The aforementioned animal disease early warning scheme based on social graph networks can integrate animal physiological indicators and daily activities to provide early warnings of animal disease transmission, improve the accuracy of animal disease early warnings, and reduce false and missed reports of animal diseases. This solves the problem that existing technologies are unable to provide timely early warnings of animal health status and cannot provide early warnings of animal disease transmission.

[0118] As a preferred embodiment, such as Figure 8 As shown, in the above-mentioned animal disease early warning system, the information collection module 110 includes:

[0119] The identifier setting submodule 111 is used to set a unique identifier for each of the plurality of animals;

[0120] The body temperature acquisition submodule 112 is used to collect the body temperature index sequence of each animal in real time using a body temperature sensor, as the health information of each animal;

[0121] The location acquisition submodule 113 is used to acquire the location information of each animal in real time using a location sensor;

[0122] The relationship establishment submodule 114 is used to establish the correspondence between the unique identifier and the health information and the location information, respectively.

[0123] As a preferred embodiment, such as Figure 9 As shown, in the aforementioned animal disease early warning system, the social graph network generation module 130 includes:

[0124] The node setting submodule 131 is used to set multiple nodes, which represent multiple animals.

[0125] The contact recognition submodule 132 is used to identify the contact between multiple animals based on the location information of multiple animals;

[0126] The edge generation submodule 133 is used to generate edges between multiple nodes and set the weights of the edges based on the contact situation of multiple animals; the contact situation of the animals includes the contact time and number of contact of multiple animals.

[0127] The network acquisition submodule 134 is used to update the nodes, edges, and edge weights in real time to obtain the social graph network of multiple animals.

[0128] As a preferred embodiment, such as Figure 10 As shown, in the above-mentioned animal disease early warning system, the second modeling module 140 includes:

[0129] The algorithm input submodule 141 is used to input the information of the social graph network into the message passing algorithm for modeling, and obtain the social graph network model.

[0130] The model initialization submodule 142 is used to initialize the node features of each node in the social graph network model using the current health status information of multiple animals.

[0131] As a preferred embodiment, such as Figure 11 As shown, in the above-mentioned animal disease early warning system, the disease early warning module 150 includes:

[0132] The multi-layer convolutional submodule 151 is used to perform multi-layer convolutional operations on the social graph network model based on the node feature information of each initialized node.

[0133] The feature acquisition submodule 152 is used to acquire the node feature information of each node in the last convolutional layer of the social graph network model;

[0134] The feature input submodule 153 is used to input the node feature information of each node in the last convolutional layer into the fully connected layer of the message passing algorithm to obtain the disease warning information for each animal.

[0135] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0137] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0139] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0140] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0141] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for early warning of animal diseases based on social graph networks, characterized in that, include: Real-time collection of health and location information for multiple animals; The health information of the multiple animals is modeled using a Long Short-Term Memory (LSTM) model to obtain the current health status information of the multiple animals. Based on the location information of the multiple animals, a social graph network of the multiple animals is generated; The social graph network is modeled using a graph neural network algorithm to obtain a social graph network model, and the social graph network model is initialized using the current health status information. This includes: inputting the information of the social graph network into a message passing algorithm to learn and update the graph network model to obtain the social graph network model; and initializing the node features of each node in the social graph network model using the current health status information of the multiple animals. Obtain the node features of the social graph network model, and provide disease warnings for each animal based on the node features; The step of obtaining node features of the social graph network model and providing disease warnings for each animal based on the node features includes: Based on the node feature information of each initialized node, perform multi-layer convolution operations on the social graph network model; Obtain the node feature information of each node in the last convolutional layer of the social graph network model; The node feature information of each node in the last convolutional layer is input into the fully connected layer of the message passing algorithm to obtain the disease warning information for each animal. The step of performing multi-layer convolution operations on the social graph network model based on the node feature information of each initialized node includes: Using the node feature information of each initialized node, according to the formula The node feature information of each node in each convolutional layer of the social graph network model is obtained sequentially; wherein, This refers to the node feature information of the i-th node in the m-th convolutional layer of the social graph network model at time t. Let m be the parameters of the m-th convolutional layer. This refers to the node feature information of all neighbor nodes j of the i-th node in the (m-1)-th convolutional layer of the social graph network model at time t; the total number of neighbors is n.

2. The method for early warning of animal diseases according to claim 1, characterized in that, The steps for real-time collection of health and location information of multiple animals include: Set a unique identifier for each of the multiple animals; The body temperature index sequence of each animal is collected in real time using a body temperature sensor, which serves as the health information of each animal; Use position sensors to collect the location information of each animal in real time; Establish a correspondence between the unique identifier and the health information and the location information, respectively.

3. The method for early warning of animal diseases according to claim 1, characterized in that, The step of generating a social graph network of the multiple animals based on their location information includes: Multiple nodes are set up, and the multiple nodes represent the multiple animals; Based on the location information of the multiple animals, the contact between the multiple animals can be identified; Based on the contact situation of the multiple animals, generate edges between the multiple nodes and set the weights of the edges; The nodes, edges, and edge weights are updated in real time to obtain the social graph network of the multiple animals.

4. An early warning system for animal diseases based on social graph networks, characterized in that, The method for early warning of animal diseases based on social graph networks as described in claim 1 includes: The information acquisition module is used to collect health and location information of multiple animals in real time; The first modeling module is used to model the health information of the multiple animals using a Long Short-Term Memory (LSTM) model, and obtain the current health status information of the multiple animals. A social graph network generation module is used to generate a social graph network of the multiple animals based on their location information. The second modeling module is used to model the social graph network using a graph neural network algorithm to obtain a social graph network model, and to initialize the social graph network model using the current health status information; The disease early warning module is used to acquire the node features of the social graph network model and to provide disease early warning for each animal based on the node features.

5. The animal disease early warning system according to claim 4, characterized in that, The social graph network generation module includes: The node setting submodule is used to set multiple nodes, which represent the multiple animals; The contact recognition submodule is used to identify the contact situation between the multiple animals based on their location information. The edge generation submodule is used to generate edges between the multiple nodes and set the weights of the edges based on the contact situation of the multiple animals; The network acquisition submodule is used to update the nodes, edges, and edge weights in real time to obtain the social graph network of the multiple animals.

Images