Data mining method and system applied to animal epidemic disease remote diagnosis center platform

Abstract

The invention provides a data mining method and system applied to an animal epidemic disease remote diagnosis center platform, and the method comprises the steps: obtaining a cross-domain attention knowledge vector of animal epidemic disease case data, and disassembling the cross-domain attention knowledge vector into a sample attention knowledge vector of a risk diffusion trend or a non-risk diffusion trend; the first risk diagnosis network is utilized to perform preliminary risk diagnosis on the sample attention knowledge vector, and the network is optimized according to the diagnosis result and the attention label attribute, so that the diagnosis accuracy is improved. Furthermore, a sample attention knowledge vector of a risk diffusion trend is extracted and loaded to the second risk diagnosis network, deeper risk diagnosis and optimization are carried out, and the accuracy of risk diagnosis is ensured. Finally, the two optimized risk diagnosis networks are connected to generate a target risk diagnosis model, whether a risk diffusion trend exists in the animal epidemic disease data mining process can be effectively diagnosed, data support is provided for remote diagnosis of animal epidemic diseases, and the efficiency and accuracy of animal epidemic disease prevention and control are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology. Specifically, it relates to a data mining method and system applied to the platform of an animal disease remote diagnosis center. Background Art

[0002] In the field of animal disease prevention and control, timely and accurately identifying the spread trend and potential risks of diseases is the key to ensuring animal health and maintaining public health security. Traditional animal disease diagnosis methods mainly rely on on-site investigations and laboratory tests. Although these methods are effective, they have problems such as long diagnosis cycles, high costs, and limited coverage, and it is difficult to meet the requirements of large-scale and rapid-response disease prevention and control.

[0003] With the development of technologies such as big data and artificial intelligence, data mining technology has gradually been introduced into the field of animal disease diagnosis. By constructing an animal disease data mining model, valuable information and patterns can be extracted from a large amount of animal disease case data, providing a scientific basis for disease prediction, early warning, and prevention and control. However, most of the existing animal disease data mining methods focus on the extraction and classification of disease characteristics, lacking in-depth analysis and diagnosis of the disease risk diffusion trend. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of this application, embodiments of this application provide a data mining method applied to the platform of an animal disease remote diagnosis center. The method includes: Obtain the cross-domain attention knowledge vector of the disease knowledge feature chain of the animal disease case data when the animal disease data mining model performs data mining on the animal disease case data. Based on the mining result of the animal disease data mining model for the animal disease case data, disassemble the cross-domain attention knowledge vector into a sample attention knowledge vector of a risk diffusion trend or a sample attention knowledge vector of a non-risk diffusion trend; Load the sample attention knowledge vector into the first risk diagnosis network, generate the first risk diagnosis result of the first risk diagnosis network for the sample attention knowledge vector, and optimize the first risk diagnosis network based on the first risk diagnosis result and the attention label attribute of the sample attention knowledge vector; Extract the sample attention knowledge vector with the attention label attribute of the risk diffusion trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, continue to load it into the second risk diagnosis network, generate the second risk diagnosis result of the second risk diagnosis network for the sample attention knowledge vector, and optimize the second risk diagnosis network based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector; The optimized first risk diagnosis network is connected to the second risk diagnosis network to generate a target risk diagnosis model, which is used to diagnose whether there is a risk diffusion trend in the animal disease data mining model when performing data mining.

[0005] On the other hand, an embodiment of the present application also provides an animal disease remote diagnosis system, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, the embodiments of the present application realize efficient and accurate mining and analysis of animal disease case data, and significantly improve the accuracy and timeliness of animal disease risk diagnosis. Specifically, by first obtaining the cross-domain attention knowledge vector of the animal disease data mining model during the mining process, and disassembling it into sample attention knowledge vectors of risk diffusion trends and non-risk diffusion trends, it is possible to accurately capture the key information directly related to the spread of disease risks, effectively avoid information redundancy, and improve the pertinence of risk identification. By introducing the first risk diagnosis network and the second risk diagnosis network, the present invention constructs a dual risk diagnosis mechanism. The first risk diagnosis network performs preliminary risk judgment, while the second risk diagnosis network performs a more in-depth analysis of the risk diffusion trend. This mechanism not only improves the accuracy of diagnosis, but also ensures the comprehensiveness of the diagnosis results. The risk diagnosis network can automatically adjust the network parameters and achieve self-optimization based on the comparison of the diagnosis results with the sample data. This capability enables the model to continuously adapt to new epidemic situations and data characteristics, and maintain the stability and advancement of its diagnostic performance. By connecting the optimized first risk diagnosis network with the second risk diagnosis network, the generated target risk diagnosis model can quickly and accurately determine whether there is a risk diffusion trend in the animal disease data mining results, which not only greatly improves the diagnosis efficiency, but also ensures the accuracy and reliability of the diagnosis results, and provides strong support for the timely prevention and control of animal diseases. Therefore, the technical solution of this embodiment is applicable to the animal disease remote diagnosis center platform, which can mine and analyze animal disease case data in real time and remotely, and provide timely and accurate epidemic information for decision makers. This helps to strengthen cross-regional prevention and control cooperation of animal diseases and improve the overall prevention and control effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the execution flow of the data mining method applied to the animal disease remote diagnosis center platform provided in the embodiment of the present application.

[0008] Figure 2It is a schematic diagram of the hardware architecture of the animal disease remote diagnosis system provided by an embodiment of the present application. Detailed implementation manners

[0009] The present application will be specifically described below with reference to the accompanying drawings of the specification. Figure 1 It is a schematic flowchart of a data mining method applied to an animal disease remote diagnosis center platform provided by an embodiment of the present application. The data mining method applied to the animal disease remote diagnosis center platform will be introduced in detail below.

[0010] Step S110: Obtain the cross-domain attention knowledge vector of the disease knowledge feature chain of the animal disease case data when the animal disease data mining model performs data mining on the animal disease case data. Based on the mining result of the animal disease data mining model on the animal disease case data, disassemble the cross-domain attention knowledge vector into a sample attention knowledge vector of a risk diffusion trend or a sample attention knowledge vector of a non-risk diffusion trend.

[0011] Specifically, the animal disease data mining model is a model for deeply mining and analyzing animal disease case data, which can extract useful information from a large amount of animal disease data to help users better understand the transmission law, risk factors, etc. of the disease. The animal disease case data refers to the specific case data of animal diseases, including detailed information such as the type of disease, occurrence time, location, types of affected animals, quantity, symptoms, treatment conditions, etc. The disease knowledge feature chain is a knowledge representation method used to describe various features related to the disease and their association relationships. Like a chain, it connects various features of the disease (such as symptoms, transmission routes, susceptible animals, etc.) to form a complete disease knowledge system.

[0012] The cross-domain attention knowledge vector is a multi-dimensional vector used to represent the attention degree to different domains (or called "domains") in the disease knowledge feature chain during the animal disease data mining process. Cross-domain means covering multiple different knowledge domains or aspects.

[0013] The sample attention knowledge vector of the risk diffusion trend refers to those sample attention knowledge vectors related to the risk diffusion trend of the disease, which usually contain some features or information indicating the possible diffusion of the disease, such as the transmission speed of the disease, the change of the infection range, etc. The sample attention knowledge vector of the non-risk diffusion trend is opposite to the sample attention knowledge vector of the risk diffusion trend, representing those features or information that are not directly related to the risk diffusion trend of the disease, reflecting the stability or controlled state of the disease in some aspects.

[0014] In this embodiment, when processing data related to animal diseases, relevant information during the data mining of animal disease case data by the animal disease data mining model needs to be obtained first. Specifically, the server stores a large amount of animal disease case data, which covers the disease situations of different regions and different species of animals. For example, in a large animal breeding base, there are multiple types of animals, such as pigs, cows, chickens, etc. For the disease case data of pigs, the disease case description data may include detailed information such as the age, breed, recent diet, and whether they have come into contact with foreign animals of the pigs, while the epidemic disease condition description data will record whether there are situations such as an accelerated virus transmission speed (this is the trend of disease risk factor condition change) in the disease case description data of diseases such as swine fever.

[0015] The server uses the animal disease data mining model to mine these animal disease case data. This data mining model includes a disease case feature extraction unit, a cross-domain association unit, and a knowledge reasoning unit (where the knowledge reasoning unit further includes an epidemic disease condition feature extraction subunit and an epidemic disease condition feature reduction subunit). Taking the pig disease data as an example, the disease case feature extraction unit will extract features such as the age and breed of the pigs in the disease case description data to generate a disease case feature vector. For example, for a 3-month-old white pig, its breed feature, diet feature, etc. are quantified as elements in the feature vector.

[0016] Then, according to the pre-set classification criteria for the knowledge fields related to animal diseases (this criterion may be formulated based on the research results of animal diseases over the years and industry norms. For example, it is divided according to the types of animal diseases such as infectious diseases and parasitic diseases; the categories of symptom manifestations such as fever and cough; and the sources of animal disease infections such as animal-to-animal transmission and environmental transmission), each element in the animal disease case feature vector is classified into different knowledge subsets. For example, the age of pigs may be related to the source of animal disease infection and is classified into the corresponding knowledge subset. Then, a cross-domain knowledge association rule set is set (for example, younger pigs may be more likely to be infected with certain animal diseases, which is a rule set based on the structure of the animal disease knowledge system and the prior requirements of animal disease diagnosis). For each knowledge subset, traverse the elements therein, judge the association relationship with the elements of other knowledge subsets and mark them. If it is found that the white pig breed is associated with certain specific infection sources, it is marked. After that, a framework for the animal disease knowledge feature chain is constructed. Starting from the core knowledge subset (such as the subset related to the source of animal disease infection), gradually connect the relevant knowledge subsets along the element association marks, and then arrange the elements in the knowledge subsets according to the advanced and refined association relationships for integrity inspection and optimization. For example, check whether all key knowledge fields of classical swine fever are covered and whether effective classical swine fever diagnosis can be carried out based on this feature chain. If there are problems, supplement the missing knowledge elements (such as supplementing the elements related to the newly discovered transmission route of classical swine fever), adjust the unreasonable association relationships (such as correcting the wrong association between breed and susceptibility), etc., and finally generate the animal disease knowledge feature chain.

[0017] Meanwhile, the epidemic situation and disease condition feature extraction sub-unit performs graph feature representation on the epidemic situation and disease condition description data to generate a knowledge graph of the disease condition trend. For example, for the classical swine fever epidemic situation, the virus transmission speed, the change of the infected range, etc. are constructed into a graph form. The epidemic situation and disease condition feature restoration sub-unit restores the features according to the cross-domain feature path data generated by the fusion of the animal disease knowledge feature chain and the knowledge graph of the disease condition trend, and generates the mining result of the animal disease data mining model for the animal disease case data. In this process, the server obtains the cross-domain attention knowledge vector of the animal disease knowledge feature chain of the animal disease data mining model for the animal disease case data.

[0018] Based on the mining result of the animal disease data mining model for the animal disease case data, the server disassembles the cross-domain attention knowledge vector into an example attention knowledge vector of the risk diffusion trend or an example attention knowledge vector of the non-risk diffusion trend. For example, if the mining result shows that the transmission speed of classical swine fever is accelerating in a certain breeding area and the number of infected pigs is on the rise, then the cross-domain attention knowledge vector related to this situation will be disassembled into an example attention knowledge vector of the risk diffusion trend; if the mining result shows that the classical swine fever epidemic situation is effectively controlled in a certain area, the transmission speed is slowing down, and the number of infected pigs is stable or decreasing, then the corresponding cross-domain attention knowledge vector will be disassembled into an example attention knowledge vector of the non-risk diffusion trend.

[0019] Step S120: Load the sample attention knowledge vector into the first risk diagnosis network to generate a first risk diagnosis result of the first risk diagnosis network for the sample attention knowledge vector, and optimize the first risk diagnosis network based on the first risk diagnosis result and the attention label attribute of the sample attention knowledge vector.

[0020] Specifically, the first risk diagnosis network is a network model for preliminarily diagnosing epidemic disease risks. It can receive the sample attention knowledge vector as input and output a preliminary diagnosis result on the risk diffusion trend of the epidemic disease. The first risk diagnosis result is the result output after the first risk diagnosis network processes the input sample attention knowledge vector, usually a probability value or a score, indicating the likelihood of the existence of a risk diffusion trend of the epidemic disease.

[0021] The attention label attribute refers to a type of label information carried in the sample attention knowledge vector, which is used to identify which epidemic disease risk state (such as risk diffusion trend, non-risk diffusion trend) the features or information represented by the vector are related to.

[0022] For example, in this embodiment, the server loads the sample attention knowledge vector obtained in the previous step into the first risk diagnosis network. Assume that the first risk diagnosis network is a neural network constructed based on a deep learning algorithm, and this neural network has been pre-set with some initial parameters. When the sample attention knowledge vector is loaded in, the first risk diagnosis network starts to perform operations. For example, for the sample attention knowledge vector of the risk diffusion trend obtained from previous swine epidemic cases (the knowledge vector converted from relevant information such as the spread speed of classical swine fever, the age distribution of infected pigs, and the breeding environment), the first risk diagnosis network will process this information according to its neuron structure and weight settings.

[0023] After the first risk diagnosis network analyzes the sample attention knowledge vector, it generates a first risk diagnosis result. For example, the diagnosis result may be a probability value, indicating a high risk diffusion trend (such as a probability value of 0.8) or a low risk diffusion trend (such as a probability value of 0.2) of classical swine fever under the current situation. At the same time, the sample attention knowledge vector has an attention label attribute, which was previously marked according to the actual situation of the epidemic disease case data. For example, for cases with an obvious risk diffusion trend of classical swine fever, it is marked with the "high risk diffusion" label.

[0024] Based on the attention label attribute of the first risk diagnosis result and the sample attention knowledge vector, the server starts to optimize the first risk diagnosis network. If the first risk diagnosis result indicates a high-risk spread trend of swine fever (probability value is 0.8), and the attention label attribute of the sample attention knowledge vector is also "high-risk spread", this shows that the diagnosis of the first risk diagnosis network is correct, but it may still need to be further optimized to improve accuracy. The server will adjust parameters such as the neuron weights in the first risk diagnosis network according to a certain optimization algorithm (such as the backpropagation algorithm). Suppose in the first risk diagnosis network, the neuron weight corresponding to the knowledge vector related to the spread speed of swine fever was 0.3 before optimization. After the feedback of this correct diagnosis, it may be adjusted to 0.32 according to the algorithm so that it can diagnose more accurately when encountering a similar sample attention knowledge vector next time. If the first risk diagnosis result does not match the attention label attribute, for example, the diagnosis result is low-risk spread (probability value is 0.2) while the label attribute is "high-risk spread", then the server will increase the amplitude of weight adjustment to correct the diagnostic deviation of the network.

[0025] Step S130, extract the sample attention knowledge vectors with the attention label attribute of risk spread trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, and continue to load them into the second risk diagnosis network to generate the second risk diagnosis result of the second risk diagnosis network for the sample attention knowledge vectors. Based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vectors, optimize the second risk diagnosis network.

[0026] Specifically, the sample training annotation data refers to the annotation information provided for the sample data when training the model. For the second risk diagnosis network, the sample training annotation data contains detailed information and evaluation results on the risk spread trend of the disease, which is used to guide the learning and optimization of the model. The second risk diagnosis network is a more refined and specialized network model than the first risk diagnosis network, which is used to further analyze and diagnose the sample attention knowledge vectors that have been preliminarily diagnosed to more accurately evaluate the risk spread trend of the disease. The second risk diagnosis result is the result output after the second risk diagnosis network processes the input sample attention knowledge vectors, and is usually more detailed and accurate than the first risk diagnosis result, and can provide more in-depth risk assessment information.

[0027] In this embodiment, the server extracts the sample attention knowledge vectors with the attention label attribute of risk spread trend from the sample attention knowledge vectors that have been loaded into the first risk diagnosis network. Continuing with the above example of swine diseases, from the previously processed sample attention knowledge vectors related to swine fever, find those knowledge vectors labeled with a risk spread trend (such as the rapid spread of swine fever in the breeding area).

[0028] Then, the extracted sample attention knowledge vectors are continuously loaded into the second risk diagnosis network. Assume that the second risk diagnosis network is a network specifically for further refining the diagnosis of the risk diffusion trend, and it may have different structures and parameters from the first risk diagnosis network. For example, it may focus more on analyzing the specific patterns of risk diffusion, such as details like whether the risk diffusion of swine fever is caused by contact transmission or airborne transmission.

[0029] After receiving these sample attention knowledge vectors, the second risk diagnosis network processes them and generates a second risk diagnosis result. For example, regarding the risk diffusion trend of swine fever, the second risk diagnosis network may diagnose that the risk diffusion of swine fever is mainly caused by direct contact transmission between pigs, and give a specific risk diffusion assessment result based on factors such as the transmission speed and contact frequency, such as the severity of the risk diffusion being "high" (which may be 0.7 in numerical representation).

[0030] Meanwhile, these sample attention knowledge vectors have sample training annotation data, which is annotated based on in-depth analysis and professional judgment of a large number of swine disease cases before. For example, for the risk diffusion situation of swine fever caused by contact transmission, detailed information such as the specific risk diffusion level and the credibility of the transmission route is annotated according to actual observation and analysis.

[0031] Based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vectors, the server optimizes the second risk diagnosis network. If the second risk diagnosis result is consistent with the sample training annotation data, for example, it is diagnosed that the risk diffusion of swine fever caused by contact transmission is high (the value is 0.7), and the sample training annotation data also indicates that the risk diffusion level is high in this case, then the server will fine-tune the parameters of the second risk diagnosis network according to the optimization algorithm. Assume that in the second risk diagnosis network, the neuron weight corresponding to the knowledge vector related to the contact transmission of swine fever is 0.4 before optimization and may be adjusted to 0.41. If the second risk diagnosis result is inconsistent with the sample training annotation data, for example, the diagnosis result is that the risk diffusion level is "medium" (the value is 0.5), while the annotation data is "high", then the server will significantly adjust parameters such as the weight to improve the accuracy of the network.

[0032] Step S140, connect the optimized first risk diagnosis network and the second risk diagnosis network to generate a target risk diagnosis model, and the target risk diagnosis model is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model conducts data mining.

[0033] In this embodiment, the server connects the optimized first risk diagnosis network with the second risk diagnosis network to generate a target risk diagnosis model. For example, the first risk diagnosis network is mainly responsible for preliminarily judging the general situation of the risk diffusion trend in the mining results of the animal disease data mining model, while the second risk diagnosis network conducts a more in-depth and detailed analysis of the risk diffusion trend. After connecting them, a more comprehensive and accurate target risk diagnosis model is formed.

[0034] This target risk diagnosis model is specifically used to diagnose whether there is a risk diffusion trend when the animal disease data mining model conducts data mining. For example, when the animal disease data mining model mines new pig disease case data, the target risk diagnosis model can accurately diagnose whether there is a risk diffusion trend of swine fever during the mining process. If the animal disease data mining model mines that the health data of pigs in a certain breeding area is abnormal, the target risk diagnosis model can comprehensively judge the overall risk trend by the first risk diagnosis network and the specific pattern and degree of risk diffusion by the second risk diagnosis network, and accurately give the diagnosis result of whether there is a risk diffusion trend of swine fever, so as to provide a strong decision-making basis for the prevention and control of animal diseases.

[0035] Based on the above steps, the embodiment of the present application realizes efficient and accurate mining and analysis of animal disease case data, and significantly improves the accuracy and timeliness of animal disease risk diagnosis. Specifically, first, by obtaining the cross-domain attention knowledge vector of the animal disease data mining model during the mining process, and disassembling it into sample attention knowledge vectors of risk diffusion trend and non-risk diffusion trend, it is possible to accurately capture the key information directly related to the spread of disease risk, effectively avoid information redundancy, and improve the pertinence of risk identification. By introducing the first risk diagnosis network and the second risk diagnosis network, the present invention constructs a dual risk diagnosis mechanism. The first risk diagnosis network performs preliminary risk judgment, while the second risk diagnosis network performs a more in-depth analysis of the risk diffusion trend. This mechanism not only improves the accuracy of diagnosis, but also ensures the comprehensiveness of the diagnosis results. The risk diagnosis network can automatically adjust the network parameters according to the comparison of the diagnosis results with the sample data to achieve self-optimization. This capability enables the model to continuously adapt to new epidemic situations and data characteristics, and maintain the stability and advancement of its diagnostic performance. By connecting the optimized first risk diagnosis network with the second risk diagnosis network, the generated target risk diagnosis model can quickly and accurately determine whether there is a risk diffusion trend in the animal disease data mining results, which not only greatly improves the diagnosis efficiency, but also ensures the accuracy and reliability of the diagnosis results, and provides strong support for the timely prevention and control of animal diseases. Therefore, the technical solution of this embodiment is applicable to the animal disease remote diagnosis center platform, which can mine and analyze animal disease case data in real time and remotely, and provide timely and accurate epidemic information for decision makers. This helps to strengthen cross-regional prevention and control cooperation of animal diseases and improve the overall prevention and control effect.

[0036] In a possible implementation, step S110 includes: Step S111, obtaining animal disease case data, wherein the animal disease case data includes a disease case description data and an epidemic condition description data, wherein the epidemic condition description data is used to record whether there is a disease condition change trend of an epidemic risk factor in the disease case description data.

[0037] Step S112, using the animal disease data mining model to obtain a disease knowledge feature chain based on the disease case description data, and to obtain a disease trend knowledge graph based on the epidemic condition description data, and based on the disease knowledge feature chain and the disease trend knowledge graph, generate a cross-domain attention knowledge vector of the disease knowledge feature chain.

[0038] In this embodiment, taking the case of swine diseases as an example, the disease case description data contains information in many aspects, such as the breed of the pig being Landrace, the age being 6 months, the weight being 100 kilograms, the breeding density of the farm where it is located being 1.5 pigs per square meter, and the recent feed replacement situation, etc. The epidemic disease condition description data records the trend of the disease condition changes of the epidemic risk factors. For example, in this pig group, whether the infection rate of the classical swine fever virus is increasing, whether the duration of abnormal body temperature of the infected pigs is getting longer, etc.

[0039] Next, the server uses the animal disease data mining model to obtain the disease knowledge feature chain based on the disease case description data. The disease case feature extraction unit in the animal disease data mining model will extract features from the disease case description data. For example, for features such as the breed, age, and weight of pigs, they are quantified into disease case feature vectors through specific algorithms. For the breed of Landrace pigs, there may be corresponding encodings in the vector according to the breed characteristics; an age of 6 months, a weight of 100 kg, etc. will also be converted into specific numerical representations. Then, the cross-domain association unit divides each element in the disease case feature vector into different knowledge subsets according to the pre-set classification criteria for disease-related knowledge domains. This classification criteria may be based on factors such as the type of disease, symptom manifestation categories, disease infection sources, etc. For example, the age of pigs may be related to the disease infection source and is divided into the corresponding knowledge subset; the breed of pigs may be related to the susceptibility to the disease and is also divided into the corresponding subset. Based on the structure of the disease knowledge system and the prior requirements of animal disease diagnosis, a cross-domain knowledge association rule set for the cross-domain association unit is set. For example, the breed of Landrace pigs may be more susceptible to classical swine fever in certain environments, which is an association rule. For each knowledge subset, traverse the elements in it, and judge whether there is an association relationship between this element and the elements in other knowledge subsets according to the cross-domain knowledge association rule set. If there is, mark the name or identifier of the other knowledge subset associated with this element for this element. For example, if it is found that the weight of pigs is associated with the breeding density, the weight element is marked with the association with the breeding density subset. Based on the knowledge subsets with association marks, a preliminary framework of the disease knowledge feature chain is constructed. Starting from the core knowledge subset, such as the subset related to the disease infection source, along the association marks of the elements, gradually connect the relevant knowledge subsets, arrange them in a logical order to form a chain-like structure framework, and the position of each knowledge subset is determined according to its association relationship with other knowledge subsets and the importance weight in disease diagnosis. For example, the subset related to the disease infection source is at the front end of the chain because it plays a key role in the occurrence of the disease. For each knowledge subset in the preliminary disease knowledge feature chain framework, again according to the association marks, arrange the elements in the knowledge subset according to the advanced and refined association relationships to generate a refined disease knowledge feature chain framework. Finally, perform an integrity check and optimization on the refined disease knowledge feature chain framework. From the perspective of the disease knowledge system, check whether all key knowledge areas of the disease are covered. From the perspective of disease diagnosis, check whether effective disease diagnosis can be carried out based on this framework. If there is a situation of missing key information or unreasonable association relationships, such as finding that information related to a newly emerging transmission route of classical swine fever is not covered, make supplements or adjustments, so as to generate the disease knowledge feature chain.

[0040] Meanwhile, the server uses the animal disease data mining model to obtain the disease trend knowledge graph based on the epidemic situation description data. The epidemic situation disease feature extraction subunit performs graph feature representation on the epidemic situation description data. For the swine fever epidemic, the upward trend of the swine fever virus infection rate, the change in the duration of abnormal body temperature of infected pigs, etc. are constructed into a graph form, where the nodes represent different disease factors and the edges represent the relationships between the factors. For example, there may be a positive correlation between the increase in the infection rate and the lengthening of the duration of abnormal body temperature.

[0041] Finally, based on the disease knowledge feature chain and the disease trend knowledge graph, the server generates a cross-domain attention knowledge vector for the disease knowledge feature chain. This vector synthesizes various association information in the disease knowledge feature chain and the disease change trend information in the disease trend knowledge graph. For example, the association relationships between factors such as pig breed, age, weight, and breeding density in the disease knowledge feature chain, combined with the information such as the upward trend of the swine fever virus infection rate and the lengthening of the duration of abnormal body temperature of infected pigs in the disease trend knowledge graph, generate a cross-domain attention knowledge vector through a specific algorithm. This vector can comprehensively reflect the comprehensive situation of animal disease case data in terms of disease knowledge features and disease trends, providing an important basis for subsequent analysis and diagnosis.

[0042] In a possible implementation manner, the animal disease data mining model includes a disease case feature extraction unit, a cross-domain association unit, and a knowledge reasoning unit, and the knowledge reasoning unit includes an epidemic situation disease feature extraction subunit and an epidemic situation disease feature restoration subunit.

[0043] Step S112 includes: Step S1121, using the disease case feature extraction unit in the animal disease data mining model to extract features from the disease case description data to generate a disease case feature vector, and using the cross-domain association unit of the animal disease data mining model to perform cross-domain knowledge association on the disease case feature vector to generate the disease knowledge feature chain of the animal disease case data.

[0044] Step S1122, using the epidemic situation disease feature extraction subunit of the animal disease data mining model to perform graph feature representation on the epidemic situation description data to generate the disease trend knowledge graph of the animal disease case data.

[0045] Step S1123: Using the epidemic disease condition feature reduction sub-unit of the animal epidemic disease data mining model, perform feature reduction based on the cross-domain feature path data generated by fusing the disease knowledge feature chain and the disease condition trend knowledge graph, generate the mining result of the animal epidemic disease data mining model for the animal epidemic disease case data, and obtain the cross-domain attention knowledge vector of the disease knowledge feature chain of the animal epidemic disease case data by the animal epidemic disease data mining model when generating the mining result.

[0046] In this embodiment, the server uses the disease case feature extraction unit in the animal epidemic disease data mining model to extract features from the disease case description data to generate a disease case feature vector. Taking the epidemic disease case of pigs as an example, the disease case description data includes many pieces of information such as the age, breed, breeding environment, and recent contact history of the pigs. The disease case feature extraction unit will perform quantization processing on this information. If the age of a pig is 3 months, it will be converted into a specific numerical code; if the breed is Duroc pigs, it will also be coded according to a pre-set coding system; parameters such as temperature and humidity in the breeding environment will also be converted into corresponding numerical representations. For example, a temperature of 20 degrees Celsius and a humidity of 60% will be mapped to a specific numerical range. Combining the quantized data of the age, breed, breeding environment, etc. of the pigs generates a disease case feature vector.

[0047] Next, the server uses the cross - domain association unit of the animal disease data mining model to perform cross - domain knowledge association on the disease case feature vectors to generate the disease knowledge feature chain of animal disease case data. The cross - domain association unit first divides the disease case feature vectors into different knowledge subsets according to the pre - set classification criteria for disease - related knowledge domains, which are based on factors such as the type of disease, symptom manifestation categories, and disease infection sources. For example, for pig diseases, the disease types may be divided into viral, bacterial, etc.; symptom manifestation categories include fever, cough, etc.; disease infection sources include contact between pigs, feed contamination, etc. Elements in the disease case feature vectors are classified into different knowledge subsets according to this classification criteria. For instance, the age of pigs may be related to the disease infection source and is thus classified into the disease infection source knowledge subset; the breed is related to the susceptibility of the disease and is classified into the corresponding knowledge subset. Then, a cross - domain knowledge association rule set is set based on the structure of the disease knowledge system and the prior requirements of animal disease diagnosis. For example, young pigs may be more susceptible to certain viral diseases, which is an association rule. For each knowledge subset, traverse the elements in it and judge the association relationship between the elements and those in other knowledge subsets according to the cross - domain knowledge association rule set. If it is found that the Duroc pig breed is related to a specific disease infection source, mark the name or identifier of the other knowledge subsets associated with this element. Based on the knowledge subsets with association marks, construct a preliminary framework of the disease knowledge feature chain. Starting from the core knowledge subset, such as the disease infection source - related subset, along the association marks of the elements, gradually arrange and connect the relevant knowledge subsets in logical order to form a chain - like structure framework, and the position of each knowledge subset is determined according to its association relationship with other knowledge subsets and its importance weight in disease diagnosis. For example, the disease infection source subset is in a key position in the disease knowledge feature chain framework because it plays an important role in the occurrence and spread of diseases. For each knowledge subset in the preliminary disease knowledge feature chain framework, again according to the association marks, arrange the elements in the knowledge subset according to the advanced and refined association relationships to generate a refined disease knowledge feature chain framework. Finally, conduct an integrity check and optimization on the refined disease knowledge feature chain framework. Check from the perspective of the disease knowledge system whether all key knowledge domains of the disease are covered, and check from the perspective of disease diagnosis whether an effective diagnosis can be made based on this framework. If there is a lack of key information, such as the lack of information related to the transmission route of a new pig disease, or if the association relationship is unreasonable, make supplements or adjustments to generate the disease knowledge feature chain.

[0048] After that, the server uses the epidemic situation and disease condition feature extraction subunit of the animal disease data mining model to perform graph feature representation on the epidemic situation and disease condition description data to generate a disease condition trend knowledge graph of animal disease case data. Taking the classical swine fever epidemic as an example, the epidemic situation and disease condition description data includes information such as the change in the infection rate of classical swine fever virus in the pig population and the development trend of the symptoms of infected pigs. The epidemic situation and disease condition feature extraction subunit will construct these information into a graph structure. The infection rate of classical swine fever virus is used as a node, and the development trend of the symptoms of infected pigs is used as another node. If there is an association between the increase in the infection rate and the aggravation of the symptoms, these two nodes will be connected by an edge. Different disease condition factors are used as nodes, and the relationships between the factors are used as edges, thus constructing a disease condition trend knowledge graph. This graph can intuitively reflect the relationships between various factors in the epidemic situation and disease condition description data and the development trend of the disease condition.

[0049] Finally, the server uses the epidemic situation and disease condition feature restoration subunit of the animal disease data mining model to perform feature restoration based on the cross-domain feature path data generated by fusing the disease knowledge feature chain and the disease condition trend knowledge graph, generate the mining result of the animal disease data mining model for the animal disease case data, and obtain the cross-domain attention knowledge vector of the disease knowledge feature chain of the animal disease case data by the animal disease data mining model when generating the mining result. The disease knowledge feature chain includes various features related to the disease such as the breed and age of pigs and their associated relationships. The disease condition trend knowledge graph includes information such as the infection rate of classical swine fever virus and the development trend of the symptoms of infected pigs. The cross-domain feature path data after fusing the two contains rich comprehensive information. The epidemic situation and disease condition feature restoration subunit performs feature restoration on these cross-domain feature path data according to a specific algorithm. For example, according to the relationship between the pig breed and the disease susceptibility in the disease knowledge feature chain, combined with the change in the infection rate of classical swine fever in the disease condition trend knowledge graph, mining results such as the risk assessment result of the pig population infected with classical swine fever are restored. During the process of generating the mining result, the server obtains the cross-domain attention knowledge vector of the disease knowledge feature chain of the animal disease case data by the animal disease data mining model. This cross-domain attention knowledge vector synthesizes various associated relationships in the disease knowledge feature chain and the disease development trend in the disease condition trend knowledge graph and other multi-faceted information, providing an important basis for subsequent epidemic risk analysis and other operations.

[0050] In a possible implementation manner, step S1121 includes: Step S1121-1, using the cross-domain association unit, according to the pre-set classification criteria for the disease-related knowledge fields, divide each element in the disease case feature vector into different knowledge subsets, and the knowledge subsets are divided based on the type of the disease, the category of symptom manifestations, and the disease infection source.

[0051] Step S1121-2: Based on the structure of the disease knowledge system and the prior requirements of animal disease diagnosis, set the cross-domain knowledge association rule set of the cross-domain association unit. The setting of the cross-domain knowledge association rules is carried out around the potential connections between different knowledge subsets.

[0052] Step S1121-3: For each knowledge subset, traverse the elements in the knowledge subset, and judge whether there is an association relationship between the element and the elements in other knowledge subsets according to the cross-domain knowledge association rule set. If there is, mark the name or identifier of the other knowledge subsets associated with the element, so that the elements within each knowledge subset are clear about the association relationships with other knowledge subsets, and generate a knowledge subset with association marks.

[0053] Step S1121-4: Based on the knowledge subsets with association marks, construct a preliminary framework of the disease knowledge feature chain. The construction process of this preliminary framework of the disease knowledge feature chain starts from the core knowledge subset, and along the association marks of the elements, gradually connects the relevant knowledge subsets. The specific connection method is to arrange the knowledge subsets with association relationships in a logical order to form a chain structure framework. In this chain structure framework, the position of each knowledge subset is determined according to the association relationship between the knowledge subset and other knowledge subsets and the importance weight in disease diagnosis.

[0054] Step S1121-5: For each knowledge subset in the preliminary framework of the disease knowledge feature chain, again according to the association marks, arrange the elements in the knowledge subset according to the advanced refined association relationship, and generate a refined framework of the disease knowledge feature chain.

[0055] Step S1121-6: Conduct integrity inspection and optimization on the refined framework of the disease knowledge feature chain. Among them, from the perspective of the disease knowledge system, check whether the refined framework of the disease knowledge feature chain covers all key knowledge areas of the disease; from the perspective of disease diagnosis, check whether the refined framework of the disease knowledge feature chain can be used for effective disease diagnosis according to the refined framework of the disease knowledge feature chain, and whether there are cases of missing key information or unreasonable association relationships in the refined framework of the disease knowledge feature chain. When it is determined that there are integrity problems in the refined framework of the disease knowledge feature chain, optimize the refined framework of the disease knowledge feature chain. The optimization methods include supplementing missing knowledge elements, adjusting unreasonable association relationships, rearranging the order of knowledge subsets or elements, thereby generating the disease knowledge feature chain, which is used to reflect the disease knowledge features in the animal disease case data and the association relationships between the disease knowledge features.

[0056] In this embodiment, first, the server uses the cross-domain association unit to divide each element in the disease case feature vector into different knowledge subsets according to the pre-set classification criteria for the disease-related knowledge fields. These knowledge subsets are divided based on the type of the disease, the category of symptom manifestations, and the disease infection source. Taking the case of pig diseases as an example, assume that the disease case feature vector contains elements such as the age of the pig (3 months), breed (Landrace), weight (50 kg), recent contact history (contact with pigs from another farm), and the hygiene status of the farm where it is located (average). According to the classification criteria for the disease-related knowledge fields, elements such as age and weight may be related to the disease infection source. For example, young and lighter-weight pigs may be more sensitive to certain disease infection sources, so these elements will be divided into the knowledge subset of the disease infection source. The breed of the pig may be related to the type of the disease. The Landrace breed may have a higher susceptibility to certain specific types of diseases. Therefore, the breed element is divided into the knowledge subset of the disease type. The recent contact history and the hygiene status of the farm are clearly related to the disease infection source and are also divided into the knowledge subset of the disease infection source.

[0057] Next, based on the structure of the disease knowledge system and the prior requirements of animal disease diagnosis, the server sets the cross-domain knowledge association rule set of the cross-domain association unit. The setting of the cross-domain knowledge association rules focuses on the potential connections between different knowledge subsets. For pig diseases, starting from the structure of the disease knowledge system, different types of diseases may be related to specific symptom manifestation categories and infection sources. For example, viral diseases may be more likely to spread in an environment with a large density of pigs and poor hygiene conditions, and may be accompanied by symptom manifestations such as fever and cough. From the prior requirements of animal disease diagnosis, if we want to quickly diagnose a viral disease such as classical swine fever, we need to pay attention to the associations between factors such as the breed, age, recent contact history of the pig, and the hygiene status of the farm. So the set association rules may include: If the breed of the pig is Landrace and the age is young, after contact with foreign pigs, when the hygiene status of the farm is poor, the risk of infecting a viral disease (such as classical swine fever) increases.

[0058] Then, for each knowledge subset, the server traverses the elements in the knowledge subset, and determines whether the element has an association relationship with the elements in other knowledge subsets based on the cross-domain knowledge association rule set. If so, the name or identifier of the other knowledge subset associated with it is marked for the element, so that the elements within each knowledge subset have a clear association relationship with other knowledge subsets, and a knowledge subset with an association mark is generated. Continuing to take swine epidemic as an example, in the epidemic infection source knowledge subset, for the element "recent contact history", according to the association rule, it is found that there is an association relationship with "viral epidemic disease" in the epidemic type knowledge subset, because recent contact with foreign pigs increases the risk of contracting viral epidemic diseases, so the "recent contact history" element is marked with an identifier associated with "viral epidemic disease". Similarly, in the epidemic type knowledge subset, for the element "swine fever", because swine fever is more easily transmitted in young pigs (age element), the "swine fever" element is marked with an association identifier with the "age" element in the epidemic infection source knowledge subset. In this way, a knowledge subset with an association mark is generated.

[0059] Afterwards, based on the knowledge subsets carrying association marks, the server constructs a preliminary disease knowledge feature chain framework. The construction process of the preliminary disease knowledge feature chain framework starts from the core knowledge subset, and gradually connects the related knowledge subsets along the association marks of the elements. The specific connection method is to arrange the knowledge subsets with association relationships in a logical order to form a chain structure framework. In the chain structure framework, the position of each knowledge subset is determined according to the association relationship between the knowledge subset and other knowledge subsets and the importance weight in disease diagnosis. In the case of swine epidemics, the disease infection source knowledge subset may be one of the core knowledge subsets because it is the source factor of disease transmission. Starting from this core knowledge subset, according to the association mark, if the disease infection source is found to be associated with the disease type, the disease type knowledge subset is connected to the disease infection source knowledge subset. For example, since the infection source factors such as the pig's contact history and hygiene conditions are associated with the disease type such as swine fever, the disease infection source knowledge subset is connected to the disease type knowledge subset in this logical order. When determining the position of each knowledge subset, its importance in disease diagnosis is taken into consideration. For example, the source of disease infection is critical in the early diagnosis of swine disease, so it is positioned relatively high in the chain structure framework; while the disease type is also important, but relatively speaking, it is more helpful for further diagnosis after the source of infection is determined, so it is positioned relatively low.

[0060] For each knowledge subset in the preliminary framework of the disease knowledge feature chain, the server again arranges the elements in the knowledge subset according to the refined association relationship based on the association tags, generating a refined framework of the disease knowledge feature chain. In the knowledge subset of the disease infection source, elements such as the recent contact history of pigs and the hygiene status of the farm were originally included. According to a more detailed association relationship, it may be found that different sources of contact with foreign pigs in the recent contact history have different impacts on the risk of disease infection. For example, pigs from disease-prone areas bring a higher risk of infection. Therefore, according to this refined association relationship, the elements in the recent contact history are arranged according to the risk level of the source of the pigs contacted. Similarly, in the knowledge subset of the disease type, for a disease such as swine fever, the relevant elements of swine fever may be refined and arranged according to the pathogenicity differences of different strains of the swine fever virus in pigs of different ages and breeds.

[0061] Finally, the server conducts integrity checks and optimizations on the refined framework of the disease knowledge feature chain. From the perspective of the disease knowledge system, it checks whether the refined framework of the disease knowledge feature chain covers all key knowledge areas of the disease. For example, for swine diseases, in addition to the factors such as breed, age, contact history, hygiene status, and disease type that have been considered, it is also necessary to check whether factors such as vaccination status and feed source that may affect the disease are covered. If it is found that the key knowledge area of vaccination status is not covered, relevant knowledge elements need to be supplemented. From the perspective of disease diagnosis, it checks whether the refined framework of the disease knowledge feature chain can be used for effective disease diagnosis based on this framework, and whether there are cases of missing key information or unreasonable association relationships in the refined framework of the disease knowledge feature chain. For example, if it is found that in the current refined framework, the association relationship between the breed of pigs and a certain disease is set incorrectly, and the susceptibility of Landrace pigs to a certain disease, which is originally low, is wrongly set to be high, this is a case of unreasonable association relationship. When it is determined that there are integrity problems in the refined framework of the disease knowledge feature chain, the refined framework of the disease knowledge feature chain is optimized. The optimization methods include supplementing missing knowledge elements, adjusting unreasonable association relationships, rearranging the order of knowledge subsets or elements. If it is found that knowledge elements related to the feed source are missing, they are supplemented; if there are unreasonable association relationships, such as the incorrect association between the breed and disease susceptibility mentioned above, they are adjusted; if the order of knowledge subsets or elements affects the diagnosis efficiency or accuracy, the order is rearranged. Through such an integrity check and optimization process, a disease knowledge feature chain is finally generated. This disease knowledge feature chain is used to reflect the disease knowledge features in the animal disease case data and the association relationships between these disease knowledge features, providing a comprehensive and accurate basis for subsequent operations such as disease analysis, diagnosis, and prevention and control.

[0062] In a possible implementation manner, the method further includes: Step A110: Obtain a sample epidemic disease case description data sequence. For each sample epidemic disease case description data in the sample epidemic disease case description data sequence, obtain the prior disease condition change trend of the sample epidemic disease case description data. The prior disease condition change trend refers to the disease condition change trend of the epidemic risk factors existing in the sample epidemic disease case description data.

[0063] Step A120: Obtain an epidemic disease condition description template, which is used to record whether there is a disease condition change trend of epidemic risk factors in the epidemic disease case description data.

[0064] Step A130: Based on the prior disease condition change trend of the sample epidemic disease case description data and the epidemic disease condition description template, generate sample epidemic disease condition description data. One prior disease condition change trend generates one sample epidemic disease condition description data, and the sample epidemic disease condition description data is used to record whether the prior disease condition change trend exists in the sample epidemic disease case description data.

[0065] Step A140: Configure animal epidemic disease case data as positive samples based on the sample epidemic disease case description data and the sample epidemic disease condition description data.

[0066] In this embodiment, first, a sample epidemic disease case description data sequence needs to be obtained. Taking the epidemic disease situation of pigs as an example, this sequence may include epidemic disease case description data of multiple different pig groups or individual pigs. For example, for pig group A, the epidemic disease case description data includes information such as the breed of pigs being Yorkshire pigs, the number of pigs in the group being 500, the temperature range of the breeding environment being 18 - 22 degrees Celsius, and the main feed ingredients being corn and soybean meal; for another pig group B, the epidemic disease case description data may be that the breed of pigs is Duroc pigs, the number of pigs in the group is 300, the breeding environment temperature is 20 - 23 degrees Celsius, and there has been a recent partial feed change situation. For each sample epidemic disease case description data in this sample epidemic disease case description data sequence, the server needs to obtain the prior disease condition change trend of the sample epidemic disease case description data. Here, the prior disease condition change trend refers to the disease condition change trend of the epidemic risk factors existing in the sample epidemic disease case description data. In the case of pig group A, if there is a situation of classical swine fever virus infection, then the prior disease condition change trend may be that the number of infected pigs increased from 5 to 10 in the past week, indicating an upward trend of classical swine fever virus infection in pig group A; for pig group B, if there is a certain bacterial epidemic disease, its prior disease condition change trend may be that the average duration of abnormal body temperature of diseased pigs has extended from 2 days to 3 days, reflecting an aggravation trend of this bacterial epidemic disease in pig group B.

[0067] Next, the server obtains an epidemic disease condition description template, which is used to record the trend of disease condition changes regarding whether there are epidemic disease risk factors in the epidemic disease case description data. For example, in the epidemic disease condition description template, there may be record items for multiple risk factors of swine diseases, such as the virus infection rate, the range of bacterial infection, the change in the severity of symptoms of diseased animals, etc., and it stipulates how to record the trend of disease condition changes of these factors, whether to record the increase or decrease in the number of infections according to a time series (such as daily, weekly), or to record the aggravation or alleviation of the disease condition according to the quantitative indicators of symptoms (such as the change in body temperature value, the frequency of coughing, etc.).

[0068] Then, based on the prior disease condition change trend of the sample epidemic disease case description data and the epidemic disease condition description template, the server generates sample epidemic disease condition description data. One prior disease condition change trend generates one sample epidemic disease condition description data, and the sample epidemic disease condition description data is used to record whether there is a prior disease condition change trend in the sample epidemic disease case description data. For the classical swine fever epidemic in swine herd A, according to the prior disease condition change trend (the number of infected pigs increased from 5 to 10 in the past week) and the epidemic disease condition description template (recording the change in the number of virus infections on a weekly basis), the generated sample epidemic disease condition description data will clearly record that the number of pigs infected with classical swine fever virus in swine herd A shows an upward trend this week; for the bacterial disease in swine herd B, based on the prior disease condition change trend (the duration of abnormal body temperature of diseased pigs extended from an average of 2 days to 3 days) and the epidemic disease condition description template (recording the disease condition change according to the duration of symptoms of diseased animals), the generated sample epidemic disease condition description data will record that the duration of abnormal body temperature of diseased pigs with this bacterial disease in swine herd B has a trend of extension.

[0069] Finally, based on the sample disease case description data and the sample epidemic disease condition description data, the server configures the animal disease case data as positive samples. For pig group A, the data of the Yorkshire pig breed it contains, the number of 500 pigs in the group, the breeding environment temperature of 18 - 22 degrees Celsius, the feed ingredients mainly composed of corn and soybean meal, and the sample epidemic disease condition description data showing an increase in the number of pigs infected with classical swine fever virus are combined to form the animal disease case data of a positive sample; for pig group B, the data of the Duroc pig breed, the number of 300 pigs in the group, the breeding environment temperature of 20 - 23 degrees Celsius, the recent partial feed replacement situation, and the sample epidemic disease condition description data showing an extended duration of abnormal body temperature in pigs suffering from bacterial diseases are integrated to also configure the animal disease case data of a positive sample. These animal disease case data of positive samples can accurately reflect the occurrence and development of diseases, contain various characteristics of disease cases and important information such as the trend of disease condition changes, etc., and provide a data basis with practical significance and value for subsequent operations such as disease data mining, analysis, diagnosis, and training and optimization of related models. For example, when training an animal disease data mining model, these positive samples can be used as correct examples to help the model learn the relationship between disease-related characteristics and the trend of disease condition changes, thereby improving the model's ability to identify and analyze diseases, so as to more accurately predict the development trend of diseases and take effective prevention and control measures in actual disease prevention and control work.

[0070] In a possible implementation manner, the method further includes: Step B110, determining the negative sample disease condition of the sample disease case description data based on the prior disease condition change trend of the sample disease case description data, where the negative sample disease condition is the disease risk factor's disease condition change trend that does not exist in the sample disease case description data.

[0071] Step B120, generating sample epidemic disease condition description data based on the negative sample disease condition of the sample disease case description data and the epidemic disease condition description template, generating one sample epidemic disease condition description data for one negative sample disease condition, and the sample epidemic disease condition description data is used to record whether the negative sample disease condition exists in the sample disease case description data.

[0072] Step B130, configuring the animal disease case data as negative samples based on the sample disease case description data and the sample epidemic disease condition description data.

[0073] In this embodiment, first, the negative sample condition of the sample epidemic disease case description data is determined based on the prior disease condition change trend of the sample epidemic disease case description data. Here, the negative sample condition refers to the disease condition change trend of the epidemic disease risk factors that do not exist in the sample epidemic disease case description data. Taking the epidemic disease case of pigs as an example, for the sample epidemic disease case description data of a certain pig group, assuming that its prior disease condition change trend shows that the infection rate of classical swine fever virus is gradually increasing, the number of infected pigs is continuously increasing, and the abnormal increase in body temperature of the diseased pigs is relatively large. Based on this, the server determines the negative sample condition. One way to determine the negative sample condition is to determine multiple negative sample conditions of the sample epidemic disease case description data based on the prior disease condition change trend, and generate a negative sample condition sequence of the sample epidemic disease case description data. For example, for the classical swine fever epidemic, the negative sample condition opposite to the increase in virus infection rate can be that the virus infection rate remains stable or decreases; the opposite of the increase in the number of infected pigs is the decrease in the number of infected pigs; the opposite of the relatively large abnormal increase in body temperature of the diseased pigs is the decrease in the abnormal increase in body temperature or the return of body temperature to normal, etc. These negative sample conditions constitute the negative sample condition sequence. Then, a random extraction is performed on this negative sample condition sequence to generate a negative sample condition. For example, the negative sample condition of the decrease in the number of infected pigs is randomly extracted.

[0074] Alternatively, the server obtains the prior disease condition change trends of other sample epidemic disease case description data in the sample epidemic disease case description data sequence except the sample epidemic disease case description data, extracts the negative sample condition of the sample epidemic disease case description data from the prior disease condition change trends of other sample epidemic disease case description data, generates a negative sample condition sequence of the sample epidemic disease case description data, and then performs a random extraction on the negative sample condition sequence to generate a negative sample condition. For example, in the sample epidemic disease case description data of another pig group, the prior disease condition change trend shows that the infection rate of swine influenza is relatively low and remains stable. Then, the situation of the stable infection rate of swine influenza can be included as one of the negative sample conditions of the sample epidemic disease case description data of the classical swine fever epidemic in the current pig group in the negative sample condition sequence. After random extraction, if this negative sample condition is selected, it is determined as the negative sample condition of the sample epidemic disease case description data.

[0075] Alternatively, the server determines a shared disease condition change trend of the prior disease condition change trend of the sample disease case description data. The shared disease condition change trend is the disease condition change trend of the disease risk factors that do not exist in the sample disease case description data and have an associated feature connection with the prior disease condition change trend. Based on the shared disease condition change trend, a negative sample disease condition sequence of the sample disease case description data is generated, and a negative sample disease condition is generated by randomly sampling the negative sample disease condition sequence, which is used as the negative sample disease condition of the sample disease case description data. For example, for the sample disease case description data of the classical swine fever epidemic, the prior disease condition change trend is an increase in the infection rate of the classical swine fever virus. The infection of the classical swine fever virus may be related to the immune system status of pigs. In other diseases related to the immune system, there is a certain bacterial infection that can enhance the immune system of pigs and thus have a resistance effect on classical swine fever. Then, the increase in the infection rate of this bacterium (which has an associated feature connection with the prior disease condition change trend) can be used as the shared disease condition change trend and incorporated into the negative sample disease condition sequence. If this disease condition is selected after random sampling, it is used as the negative sample disease condition.

[0076] Next, based on the negative sample disease condition of the sample disease case description data and the epidemic disease condition description template, the server generates sample epidemic disease condition description data. One negative sample disease condition generates one sample epidemic disease condition description data, which is used to record whether there is a negative sample disease condition in the sample disease case description data. For example, for the negative sample disease condition of the decrease in the number of infected pigs in the previously determined classical swine fever epidemic, according to the epidemic disease condition description template (which stipulates the change in the number of infected pigs is statistically counted weekly), the generated sample epidemic disease condition description data will record that the number of pigs infected with the classical swine fever virus shows a decreasing trend in this sample disease case description data.

[0077] Finally, based on the sample disease case description data and the sample epidemic disease condition description data, the server configures animal disease case data as negative samples. For the example of the classical swine fever epidemic, the content of the sample disease case description data including the breed of pigs, the number of pig herds, the breeding environment, the feed situation, etc., and the sample epidemic disease condition description data recording the decreasing trend of the number of infected pigs are combined together to configure an animal disease case data of a negative sample. These animal disease case data of negative samples correspond to the animal disease case data of positive samples and are of great significance in subsequent operations such as training, optimizing, and analyzing the animal disease data mining model. For example, when training the model, positive samples can enable the model to learn the normal development patterns and characteristics of diseases, while negative samples can enable the model to better identify situations different from the normal pattern, thereby improving the accuracy and generalization ability of the model. In the assessment of disease risks, negative sample data can provide a basis for comparison with positive sample data, helping to more comprehensively analyze the development trend of diseases and providing a reference for formulating reasonable disease prevention and control strategies.

[0078] In a possible implementation, step B110 includes: Based on the prior disease condition change trend of the sample disease case description data, determine multiple negative sample disease conditions of the sample disease case description data, generate a negative sample disease condition sequence of the sample disease case description data, randomly select from the negative sample disease condition sequence, and generate a negative sample disease condition as the negative sample disease condition of the sample disease case description data. Or, Obtain the prior disease condition change trends of other sample disease case description data in the sample disease case description data sequence except the sample disease case description data, extract the negative sample disease condition of the sample disease case description data from the prior disease condition change trends of the other sample disease case description data, generate a negative sample disease condition sequence of the sample disease case description data, randomly select from the negative sample disease condition sequence, and generate a negative sample disease condition as the negative sample disease condition of the sample disease case description data. Or, Determine the shared disease condition change trend of the prior disease condition change trend of the sample disease case description data. The shared disease condition change trend is the disease condition change trend of the disease risk factors that do not exist in the sample disease case description data and have an associated feature connection with the prior disease condition change trend. Based on the shared disease condition change trend, generate a negative sample disease condition sequence of the sample disease case description data, randomly select from the negative sample disease condition sequence, and generate a negative sample disease condition as the negative sample disease condition of the sample disease case description data.

[0079] In this embodiment, when the method of determining multiple negative sample disease conditions of the sample disease case description data based on the prior disease condition change trend of the sample disease case description data, generating a negative sample disease condition sequence, and then randomly selecting is adopted, the operation process is as follows.

[0080] Suppose there is a sample disease case description data of a pig herd, where the prior disease condition change trend shows that the infection rate of classical swine fever virus has been continuously increasing in the past month, rising from 5% to 15% initially. At the same time, the severity of the clinical symptoms of the diseased pigs is also increasing. For example, the average body temperature of the pigs with fever has increased from 40°C to 41°C, and the symptom of rapid breathing is more obvious, with the number of breaths per minute increasing from 40 times to 50 times.

[0081] Based on such a prior disease condition change trend, the server begins to determine the disease conditions of multiple negative samples. For the situation where the infection rate of the classical swine fever virus is increasing, the disease condition of its negative sample can be that the infection rate remains unchanged or decreases. For example, the infection rate remains at 5% unchanged, or decreases to 3%. For the situations of the increased body temperature and aggravated shortness of breath symptoms of diseased pigs, the disease conditions of the negative samples are that the body temperature remains at the original level or decreases, such as the body temperature remains at 40°C or decreases to 39.5°C; the shortness of breath symptoms are alleviated, such as the respiratory rate decreases to 35 times per minute or returns to the normal 30 times per minute.

[0082] The server aggregates these disease conditions of negative samples to generate a sequence of disease conditions of negative samples for the sample disease case description data. This sequence contains various possible situations contrary to the prior disease condition change trend, covering multiple aspects such as the infection rate of the classical swine fever virus and the symptoms of diseased pigs. Then, the server randomly selects from this sequence of disease conditions of negative samples. Suppose that after random selection, the group of disease conditions of negative samples where the infection rate of the classical swine fever virus decreases to 3%, the body temperature of the diseased pig remains at 40°C, and the shortness of breath symptoms are alleviated to 35 times per minute is selected, and it is taken as the disease condition of the negative sample for this sample disease case description data.

[0083] Alternatively, it is also possible to obtain the prior disease condition change trend of other sample disease case description data in the sample disease case description data sequence except for the sample disease case description data, extract the disease conditions of negative samples of the sample disease case description data from it, and then generate a sequence of disease conditions of negative samples and randomly select.

[0084] For example, there is a sequence of sample disease case description data of a group of pigs. The sample disease case description data of a pig group A is the object of current concern, and its prior disease condition change trend is that the infection rate of the swine influenza virus is gradually increasing, and the listless symptoms of the infected pigs are continuously aggravated. At the same time, in the sample disease case description data of pig group B, the prior disease condition change trend shows that after a period of time, the infection rate of the swine influenza virus begins to decline, and there are signs of gradual improvement in the mental state of the infected pigs.

[0085] The server will extract the disease conditions of negative samples related to pig group A from the prior disease condition change trend of pig group B. For the situations of the increasing infection rate of the swine influenza virus and the aggravated listless symptoms of the infected pigs in pig group A, the disease conditions of negative samples extracted from pig group B are that the infection rate of the swine influenza virus decreases and the mental state of the infected pigs improves. The server combines these disease conditions of negative samples extracted from other sample disease case description data to generate a sequence of disease conditions of negative samples for the sample disease case description data (pig group A). Then, a random selection is made from this sequence. Suppose that the group of disease conditions of negative samples where the infection rate of the swine influenza virus decreases and the mental state of the infected pigs improves is randomly selected, and it is determined as the disease condition of the negative sample for the sample disease case description data of pig group A.

[0086] In addition, the server can generate a negative sample disease condition sequence by determining a shared disease condition change trend of the prior disease condition change trend of the sample disease case description data and randomly select it.

[0087] Suppose in the sample disease case description data of a certain pig herd, the prior disease condition change trend is that the infection rate of porcine circovirus increases, and it is accompanied by a decrease in the immune function of pigs, manifested as a decrease in the lymphocyte ratio and a decrease in the white blood cell count. The server needs to determine the shared disease condition change trend, that is, the disease risk factor's disease condition change trend that does not exist in this sample disease case description data and has an associated feature connection with the prior disease condition change trend.

[0088] After analyzing the disease knowledge system and relevant data, it is found that the quantity of a certain probiotic in the pig intestine is closely related to the immune function of pigs. In the research data of other pig herds, when the quantity of this probiotic in the pig intestine increases, the immune function of pigs is enhanced, and it can effectively resist the infection of porcine circovirus. Therefore, the increase in the quantity of this probiotic is the shared disease condition change trend that has an associated feature connection with the prior disease condition change trend (the increase in the porcine circovirus infection rate and the decrease in the pig immune function).

[0089] Based on this shared disease condition change trend, the server generates a negative sample disease condition sequence of the sample disease case description data. For example, different degrees of increase in the quantity of the probiotic can constitute this sequence, such as situations at different levels of increasing by 10%, 20%, etc. Then, randomly select from this negative sample disease condition sequence. Suppose the negative sample disease condition of a 20% increase in the quantity of the probiotic is drawn, and it is used as the negative sample disease condition of this sample disease case description data.

[0090] Through the above three methods, the negative sample disease condition can be accurately determined according to the prior disease condition change trend of the sample disease case description data, which lays a foundation for subsequent generating the sample epidemic disease condition description data and configuring the animal disease case data of the negative sample, and has important significance in aspects such as the training, optimization of the animal disease data mining model, and disease risk analysis. For example, in model training, the introduction of the negative sample disease condition can enable the model to better identify the differences between different disease condition development patterns, thereby improving the accuracy and comprehensiveness of the model's judgment of the disease situation.

[0091] In a possible implementation manner, the method further includes: Step C110, obtaining the prior disease condition change trend of each sample disease case description data in the sample disease case description data sequence.

[0092] Step C120: Calculate the correlation degree between any two prior disease condition change trends based on the sample disease case description data sequence. The simultaneous occurrence of two different prior disease condition change trends in the same sample disease case description data is regarded as one association.

[0093] Step C130: When the correlation degree between two different prior disease condition change trends meets the target requirements, determine that there is an associated feature connection between the two different prior disease condition change trends.

[0094] In this embodiment, first, obtain the prior disease condition change trends of each sample disease case description data in the sample disease case description data sequence. Taking the disease case of pigs as an example, in the sample disease case description data sequence, for the sample disease case description data of pig group A, it may contain information related to swine fever. Its prior disease condition change trend is that the swine fever virus infection rate has increased from 5% to 10% in the past week, and at the same time, the abnormal increase in the body temperature of diseased pigs has risen from an average of 39.5°C - 40°C to 40°C - 40.5°C; for the sample disease case description data of pig group B, if it is about swine influenza, the prior disease condition change trend may be that the number of infected pigs has increased from 20 to 30 in a recent period, and the severity of coughing in diseased pigs has increased, with the coughing frequency increasing from 10 times per hour to 15 times per hour. The server will obtain the prior disease condition change trends in the sample disease case description data of these different pig groups one by one.

[0095] Next, calculate the correlation degree between any two prior disease condition change trends based on the sample disease case description data sequence. Suppose we want to calculate the correlation degree between the prior disease condition change trend of swine fever in pig group A and the prior disease condition change trend of swine influenza in pig group B. Since the definition of association is that the simultaneous occurrence of two different prior disease condition change trends in the same sample disease case description data is regarded as one association, it is necessary to traverse the entire sample disease case description data sequence. For example, in the sample disease case description data of some pig groups infected with both swine fever and swine influenza, if there are 10 such pig group data, and the situation where the increase in the swine fever virus infection rate and the increase in the number of swine influenza-infected pigs occur simultaneously in these pig groups is 3 times, then according to the calculation method of the correlation degree (the number of associations divided by the total number of samples), the correlation degree between these two prior disease condition change trends is 3 / 10 = 0.3. Another example is to calculate the correlation degree between the prior disease condition change trend of the increase in the swine fever virus infection rate and the abnormal increase in the body temperature of diseased pigs in pig group A and the prior disease condition change trend of the weight loss of pigs infected with a certain bacterial disease in pig group C. By traversing the sample disease case description data sequence, it is found that there are 2 pig group data with these three disease condition change trends at the same time. Assuming that there are a total of 8 pig group data containing relevant diseases, then the correlation degree between them is 2 / 8 = 0.25.

[0096] Finally, when the correlation degree between the prior disease condition change trends of two different prior disease conditions meets the target requirement, the server determines that there is an associated feature connection between these two different prior disease condition change trends. For example, when the target requirement is set that there is an associated feature connection when the correlation degree is greater than 0.2. For the correlation degree between the prior disease condition change trend of classical swine fever in pig group A and the prior disease condition change trend of swine influenza in pig group B calculated previously, which is 0.3, since 0.3 is greater than 0.2, the server determines that there is an associated feature connection between the two prior disease condition change trends of the increase in the classical swine fever virus infection rate and the increase in the number of pigs infected with swine influenza. Similarly, for the correlation degree between the prior disease condition change trend of the increase in the classical swine fever virus infection rate and the increase in the abnormal rise in body temperature of diseased pigs in pig group A and the prior disease condition change trend of the weight loss of pigs infected with a certain bacterial disease in pig group C, which is 0.25, because 0.25 is greater than 0.2, it is also determined that there is an associated feature connection between them. The determination of this associated feature connection helps to deeply understand the internal relationship between different diseases or between different disease condition change trends of the same disease, and provides an important basis for the comprehensive analysis, diagnosis, and formulation of prevention and control strategies for diseases. For example, in disease prevention and control, if it is found that there is an associated feature connection between the increase in the classical swine fever virus infection rate and the increase in the number of pigs infected with swine influenza, then more attention needs to be paid to the prevention and control of swine influenza while preventing classical swine fever to prevent the two diseases from affecting each other and aggravating the epidemic situation.

[0097] In a possible implementation manner, step S110 may further include: Obtain the mining result of the animal disease data mining model for the animal disease case data, where the mining result is one of a first mining result and a second mining result.

[0098] Based on the category of the mining result and the label data of the animal disease case data, disassemble the cross-domain attention knowledge vector into a sample attention knowledge vector of the first type of risk diffusion trend, a sample attention knowledge vector of the second type of risk diffusion trend, a sample attention knowledge vector of the first type of non-risk diffusion trend, or a sample attention knowledge vector of the second type of non-risk diffusion trend.

[0099] Step S130 includes: Step S131, extract from the sample attention knowledge vectors loaded into the first risk diagnosis network the sample attention knowledge vectors with the attention label attribute of the first type of risk diffusion trend, and continue to load them into the second risk diagnosis network of the first network label to generate the second risk diagnosis result of the second risk diagnosis network of the first network label for the sample attention knowledge vectors. Based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vectors, optimize the second risk diagnosis network of the first network label.

[0100] Step S132: From the sample attention knowledge vectors loaded into the first risk diagnosis network, extract the sample attention knowledge vectors whose attention label attributes are the second type of risk diffusion trend, and continue to load them into the second risk diagnosis network with the second network label, generate the second risk diagnosis result of the second risk diagnosis network with the second network label for the sample attention knowledge vectors, and optimize the second risk diagnosis network with the second network label based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vectors.

[0101] Step S140 includes: Step S141: Connect the optimized first risk diagnosis network with the second risk diagnosis network of the first network label to generate a target risk diagnosis model, which is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model obtains the first mining result during risk diagnosis. And Step S142: Connect the optimized first risk diagnosis network with the second risk diagnosis network of the second network label to generate a target risk diagnosis model, which is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model obtains the second mining result during risk diagnosis.

[0102] In this embodiment, first, taking swine diseases as an example, assume that the first mining result indicates that the overall infection situation of classical swine fever in a certain area is in the initial development stage, the number of infected pigs is relatively small but shows a slow upward trend, and the virus transmission range is limited to several farms; the second mining result may indicate the outbreak situation of swine influenza in this area, the number of infected pigs has increased significantly in a short time, the transmission range involves multiple farms and has a tendency to spread to the surrounding areas. The label data of the animal disease case data contains pre-annotated information such as the accurate type of the disease, the severity of the condition, and whether there is a risk of spread. For example, for classical swine fever case data, the label data may be annotated as classical swine fever type, medium severity of the condition, and currently there is a low risk of spread; for swine influenza case data, the label data is annotated as swine influenza type, high severity of the condition, and there is a high risk of spread.

[0103] Furthermore, for the case of classical swine fever, if the mining result is the first mining result (preliminary development stage), according to the low-risk diffusion trend in the label data, the server will disassemble from the cross-domain attention knowledge vector the sample attention knowledge vector of the first type of risk diffusion trend related to this low-risk diffusion trend. This vector may contain knowledge elements related to factors such as the slow spread speed of the classical swine fever virus within the farm and the limited activity range of infected pigs. At the same time, the sample attention knowledge vector of the first type of non-risk diffusion trend will also be disassembled, which may contain knowledge elements related to factors that inhibit virus diffusion, such as good epidemic prevention measures in the farm and relatively high immunity of the pig herd. For the case of swine influenza, if the mining result is the second mining result (outbreak and spread), according to the high-risk diffusion trend in the label data, the server disassembles the sample attention knowledge vector of the second type of risk diffusion trend, which may contain knowledge elements related to high-risk diffusion such as the high infectivity of the swine influenza virus and the high-density contact between pig herds, as well as the sample attention knowledge vector of the second type of non-risk diffusion trend, such as knowledge elements related to the fact that local prevention and control measures may have played a certain protective role for some pig herds.

[0104] Next, for the sample attention knowledge vector of the first type of risk diffusion trend, the server extracts the sample attention knowledge vector with the attention label attribute of the first type of risk diffusion trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, and continues to load it into the second risk diagnosis network of the first network label. For example, after the sample attention knowledge vector related to the first type of risk diffusion trend of classical swine fever is loaded into the second risk diagnosis network of the first network label, this network will analyze and process it. Suppose the second risk diagnosis network of the first network label is a diagnosis network constructed based on a neural network, and it has been pre-trained with a large amount of swine disease data. This network analyzes the sample attention knowledge vector related to classical swine fever, and the knowledge elements contained therein, such as the transmission speed of classical swine fever virus and the activity range of infected pigs, will be processed by the neurons in the network. The network generates the second risk diagnosis result of the second risk diagnosis network of the first network label for the sample attention knowledge vector according to its own algorithm and parameters. For example, the diagnosis result may indicate that the risk diffusion trend of classical swine fever may remain at a relatively low level in the short term, but if certain factors (such as changes in the breeding environment and introduction of new pig groups) change, the risk may increase. Then, based on this second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector, the second risk diagnosis network of the first network label is optimized. The sample training annotation data contains an accurate assessment of the risk diffusion trend of classical swine fever, such as the risk diffusion speed and influence range annotated according to the actual epidemic development data. If there is a deviation between the second risk diagnosis result and the sample training annotation data, for example, the diagnosis result underestimates the risk diffusion speed, the server will adjust the parameters in the second risk diagnosis network of the first network label according to a specific optimization algorithm (such as the backpropagation algorithm). For example, the neuron weights related to the transmission speed of classical swine fever virus may be appropriately increased to improve the diagnostic accuracy of the network for the risk diffusion trend of classical swine fever.

[0105] Similarly, for the sample attention knowledge vector of the second type of risk diffusion trend, the server extracts the sample attention knowledge vector with the attention label attribute of the second type of risk diffusion trend from the sample attention knowledge vector loaded into the first risk diagnosis network, and continues to load it into the second risk diagnosis network with the second network label. Taking swine flu as an example, when the sample attention knowledge vector of the second type of risk diffusion trend related to swine flu is loaded into the second risk diagnosis network with the second network label, this network starts to diagnose. The neurons in the network process knowledge elements such as the high infectivity of the swine flu virus and the high-density contact of the pig population, and generate the second risk diagnosis result of the second risk diagnosis network of the second network label for the sample attention knowledge vector. For example, the diagnosis result may indicate that the risk diffusion trend of swine flu is very high under the current conditions and may further expand the spread range in a short time. Then, according to this second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector, the second risk diagnosis network of the second network label is optimized. The sample training annotation data contains detailed annotations of the risk diffusion trend of swine flu, such as the accurate value of the transmission speed and the infection range in different time periods. If the diagnosis result is not completely consistent with the sample training annotation data, for example, the diagnosis result overestimates the transmission speed, the server will adjust the parameters in the second risk diagnosis network of the second network label through an optimization algorithm. For example, the neuron weights related to the infectivity of the swine flu virus may be appropriately reduced to enable the network to more accurately diagnose the risk diffusion trend of swine flu.

[0106] Finally, the server connects the optimized first risk diagnosis network and the second risk diagnosis network to generate a target risk diagnosis model.

[0107] For the case of connecting the optimized first risk diagnosis network and the second risk diagnosis network of the first network label to generate a target risk diagnosis model, this target risk diagnosis model is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model obtains the first mining result during risk diagnosis. For example, in the case of classical swine fever, when the animal disease data mining model obtains the first mining result (the initial development stage), the target risk diagnosis model can comprehensively consider the preliminary judgment of the overall situation of classical swine fever by the first risk diagnosis network and the further analysis of the risk diffusion trend of classical swine fever by the second risk diagnosis network of the first network label. The first risk diagnosis network may have already made a preliminary assessment of the basic condition and infection range of classical swine fever, while the second risk diagnosis network of the first network label focuses on the diagnosis of the first type of risk diffusion trend of classical swine fever. After connecting them, the target risk diagnosis model can more accurately judge whether there is a risk diffusion trend of classical swine fever when the animal disease data mining model obtains the first mining result.

[0108] For the case of connecting the optimized first risk diagnosis network with the second risk diagnosis network of the second network label to generate a target risk diagnosis model, this target risk diagnosis model is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model obtains a second mining result during risk diagnosis. Taking swine flu as an example, when the animal disease data mining model obtains the second mining result (outbreak and spread), the target risk diagnosis model combines the preliminary judgment of the overall situation of swine flu by the first risk diagnosis network and the in-depth analysis of the second risk diagnosis network of the second network label on the risk diffusion trend of the second type of swine flu. The first risk diagnosis network may have determined the basic situation such as the severity of swine flu and the number of infected pigs, while the second risk diagnosis network of the second network label focuses on analyzing the knowledge elements related to the high-risk diffusion trend of swine flu. The connected target risk diagnosis model can more accurately judge whether there is a risk diffusion trend of swine flu when the animal disease data mining model obtains the second mining result, thereby providing a more accurate and targeted decision-making basis for the prevention and control of swine diseases.

[0109] In a possible implementation manner, the method further includes: Step D110, obtaining a target mining result generated by the animal disease data mining model for risk diagnosis of target animal disease case data, where the target animal disease case data includes a target disease case description data and a target epidemic situation and condition description data, and the target epidemic situation and condition description data is used to record the trend of the change in the disease risk factors in the target disease case description data.

[0110] Step D120, obtaining a cross-domain attention knowledge vector of the disease knowledge feature chain of the target animal disease case data by the animal disease data mining model, loading the cross-domain attention knowledge vector into the target risk diagnosis model, where the target risk diagnosis model is generated by connecting the optimized first risk diagnosis network and the second risk diagnosis network, and using the first risk diagnosis network to generate a first risk diagnosis result.

[0111] Step D130, when the first risk diagnosis result indicates that there is a risk diffusion trend when the animal disease data mining model conducts data mining, loading the cross-domain attention knowledge vector into the second risk diagnosis network to generate a second risk diagnosis result.

[0112] Step D140, when the second risk diagnosis result indicates that there is a risk diffusion trend when the animal disease data mining model conducts data mining, determining that there is a risk diffusion trend when the animal disease data mining model conducts data mining on the target animal disease case data.

[0113] And, step D150, when the first risk diagnosis result indicates that there is no risk diffusion trend in the data mining of the animal disease data mining model, or when the second risk diagnosis result indicates that there is no risk diffusion trend in the data mining of the animal disease data mining model, it is determined that there is no risk diffusion trend in the data mining of the animal disease data mining model for the target animal disease case data.

[0114] In this embodiment, taking the disease situation of pigs as an example, the target disease case description data in the target animal disease case data may include information such as the breed of pigs being Landrace pigs, the scale of the pig herd being 300 heads, a partial replacement of the recent feed source, and the breeding environment temperature being between 18 - 22 degrees Celsius; the target epidemic disease condition description data records the trend of disease condition changes of whether there are disease risk factors in the target disease case description data, such as the infection rate of classical swine fever virus rising from 3% to 5% in the past week, and the abnormal body temperature ratio of diseased pigs increasing from 10% to 15%, etc. The animal disease data mining model conducts risk diagnosis based on these data to generate the target mining result. Suppose the target mining result indicates that there is a risk of classical swine fever infection in the pig herd, and the number of infected pigs may continue to increase, and the spread range has a tendency to expand, etc.

[0115] Next, the server obtains the cross - domain attention knowledge vector of the disease knowledge feature chain of the target animal disease case data by the animal disease data mining model. This cross - domain attention knowledge vector synthesizes various association information in the disease knowledge feature chain and the disease condition change trend information in the disease condition trend knowledge graph. For example, the association relationships between factors such as pig breed, age, weight, and breeding density in the disease knowledge feature chain, combined with information such as the rising trend of classical swine fever virus infection rate and the increasing proportion of abnormal body temperature of infected pigs in the disease condition trend knowledge graph, generate a cross - domain attention knowledge vector through a specific algorithm. Then, the server loads this cross - domain attention knowledge vector into the target risk diagnosis model, which is generated by connecting the optimized first risk diagnosis network and the second risk diagnosis network. The first risk diagnosis network is used to generate the first risk diagnosis result. After receiving the cross - domain attention knowledge vector, the first risk diagnosis network processes it according to its own structure and algorithm. For example, the neurons in the first risk diagnosis network analyze each element in the cross - domain attention knowledge vector, such as the relationship between pig breed and classical swine fever susceptibility, the impact of pig herd size on disease transmission, and the rising amplitude of classical swine fever virus infection rate. Based on these analyses, the first risk diagnosis network generates the first risk diagnosis result. Suppose the first risk diagnosis result indicates that there is a risk diffusion trend in the data mining of the animal disease data mining model. For example, the result shows that there is a high possibility of classical swine fever spreading to other pig herds in the current pig herd because of factors such as the rising classical swine fever virus infection rate and a certain possibility of contact between pig herds.

[0116] When the first risk diagnosis result indicates that there is a risk of spread during the data mining of the animal disease data mining model, the server loads the cross-domain attention knowledge vector into the second risk diagnosis network to generate the second risk diagnosis result. The second risk diagnosis network is a network that conducts a more in-depth and detailed analysis of the risk of spread. When it receives the cross-domain attention knowledge vector, it further mines the information therein. For example, the second risk diagnosis network will analyze in more detail the transmission pattern of the classical swine fever virus, whether the risk of spread is caused by direct contact transmission or indirect contact (such as through contaminated feed, water sources, etc.), and the role of pigs of different ages and weights in this transmission process. Suppose the second risk diagnosis result indicates that there is a risk of spread during the data mining of the animal disease data mining model. For example, it is determined that classical swine fever is mainly transmitted through direct contact between pig herds, and due to the large size and high breeding density of the pig herds, this transmission method will cause classical swine fever to spread rapidly to the surrounding pig herds in a short time.

[0117] In this case, the server determines that there is a risk of spread during the data mining of the animal disease data mining model for the target animal disease case data. This result indicates that there is a risk of spread of classical swine fever in the pig herd, and corresponding prevention and control measures need to be taken, such as isolating the infected pigs, thoroughly disinfecting the pigsty, and restricting the movement of the pig herd.

[0118] On the other hand, if the first risk diagnosis result indicates that there is no risk of spread during the data mining of the animal disease data mining model. For example, the result shows that although there is an infection with the classical swine fever virus, due to factors such as good sanitary conditions in the breeding environment and high immunity of the pig herd, the spread of the classical swine fever virus has been effectively controlled and will not spread to other pig herds. Or, when the second risk diagnosis result indicates that there is no risk of spread during the data mining of the animal disease data mining model. For example, the second risk diagnosis network analyzes and finds that although there is an infection with the classical swine fever virus, the transmission route has been effectively blocked, and the contact between pig herds is strictly controlled, and there will be no spread. In these two cases, the server determines that there is no risk of spread during the data mining of the animal disease data mining model for the target animal disease case data. This result means that the current situation of the pig herd disease is under control, and large-scale prevention and control measures do not need to be taken, but the development of the disease still needs to be continuously monitored to prevent the situation from changing.

[0119] Thus, it is possible to use the target risk diagnosis model to further judge the risk of spread of the mining result of the animal disease data mining model for the target animal disease case data, provide an accurate basis for the prevention and control decision-making of animal diseases, and help to timely and effectively control the spread and development of animal diseases.

[0120] Figure 2The figure shows a hardware structure diagram of an animal disease remote diagnosis system 100 for implementing the data mining method applied to the animal disease remote diagnosis center platform provided by the embodiments of the present application, as Figure 2 shown, the animal disease remote diagnosis system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0121] In a possible design, the animal disease remote diagnosis system 100 may be a single server or a server group. The server group may be centralized or distributed (for example, the animal disease remote diagnosis system 100 may be a distributed system). In some embodiments, the animal disease remote diagnosis system 100 may be local or remote. For example, the animal disease remote diagnosis system 100 may access information and / or data stored in the machine-readable storage medium 120 via a network. Also, for example, the animal disease remote diagnosis system 100 may be directly connected to the machine-readable storage medium 120 to access the stored information and / or data. In some embodiments, the animal disease remote diagnosis system 100 may be implemented on the animal disease remote diagnosis system. By way of example only, the animal disease remote diagnosis system may include a private cloud, a semantic-related cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc. or any aggregation thereof.

[0122] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store the data and / or instructions that the animal disease remote diagnosis system 100 uses to execute or use to complete the exemplary methods described in the present application.

[0123] In a specific implementation process, one or more processors 110 execute computer-executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the data mining method applied to the animal disease remote diagnosis center platform in the above method embodiments. The processor 110, the machine-readable storage medium 120, and the communication unit 140 are connected through the bus 130, and the processor 110 may be used to control the transceiver actions of the communication unit 140.

[0124] For the specific implementation process of the processor 110, reference may be made to the various method embodiments executed by the above animal disease remote diagnosis system 100. The implementation principles and technical effects are similar, and will not be elaborated herein in this embodiment.

[0125] In addition, an embodiment of the present application further provides a readable storage medium, in which computer-executable instructions are set. When a processor executes the computer-executable instructions, the data mining method applied to the animal disease remote diagnosis center platform as described above is implemented.

[0126] It should be noted that, in order to simplify the description of the present application disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, sometimes multiple features are merged into one embodiment, drawing or description thereof. Similarly, it should be noted that, in order to simplify the description of the present application disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, sometimes multiple features are merged into one embodiment, drawing or description thereof.

Claims

1. A data mining method applied to an animal disease remote diagnosis center platform, characterized in that: The method comprises: Obtaining a cross-domain attention knowledge vector of a disease knowledge feature chain of the animal disease case data when the animal disease data mining model performs data mining on the animal disease case data, and based on the mining result of the animal disease data mining model on the animal disease case data, decomposing the cross-domain attention knowledge vector into a sample attention knowledge vector of a risk diffusion trend or a sample attention knowledge vector of a non-risk diffusion trend; Loading the sample attention knowledge vector into a first risk diagnosis network, generating a first risk diagnosis result of the first risk diagnosis network for the sample attention knowledge vector, and optimizing the first risk diagnosis network based on the first risk diagnosis result and the attention label attribute of the sample attention knowledge vector; Extracting sample attention knowledge vectors whose attention label attribute is risk diffusion trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, and continuously loading them into the second risk diagnosis network, generating a second risk diagnosis result of the second risk diagnosis network for the sample attention knowledge vector, and optimizing the second risk diagnosis network based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector; The optimized first risk diagnosis network is connected to the second risk diagnosis network to generate a target risk diagnosis model, which is used to diagnose whether there is a risk diffusion trend in the animal disease data mining model when performing data mining.

2. The data mining method applied to the animal disease remote diagnosis center platform according to claim 1 is characterized in that: The cross-domain attention knowledge vector of the disease knowledge feature chain of the animal disease case data when the animal disease data mining model is used to perform data mining on the animal disease case data includes: Acquire animal disease case data, wherein the animal disease case data includes disease case description data and epidemic condition description data, wherein the epidemic condition description data is used to record whether there is a disease risk factor condition change trend in the disease case description data; The animal disease data mining model is used to obtain a disease knowledge feature chain based on the disease case description data, and a disease trend knowledge graph is obtained based on the epidemic condition description data. Based on the disease knowledge feature chain and the disease trend knowledge graph, a cross-domain attention knowledge vector of the disease knowledge feature chain is generated.

3. The data mining method applied to the animal disease remote diagnosis center platform according to claim 2 is characterized in that: The animal disease data mining model includes a disease case feature extraction unit, a cross-domain association unit and a knowledge reasoning unit, and the knowledge reasoning unit includes an epidemic condition feature extraction subunit and an epidemic condition feature restoration subunit; The method of using the animal disease data mining model to obtain a disease knowledge feature chain based on the disease case description data, and obtaining a disease trend knowledge graph based on the epidemic condition description data, and generating a cross-domain attention knowledge vector of the disease knowledge feature chain based on the disease knowledge feature chain and the disease trend knowledge graph, includes: Using the disease case feature extraction unit in the animal disease data mining model, feature extraction is performed on the disease case description data to generate a disease case feature vector; using the cross-domain association unit of the animal disease data mining model, cross-domain knowledge association is performed on the disease case feature vector to generate a disease knowledge feature chain of the animal disease case data; Utilizing the epidemic condition feature extraction subunit of the animal epidemic disease data mining model, the epidemic condition description data is represented by graph features to generate a disease trend knowledge graph of the animal epidemic disease case data; Utilizing the epidemic disease characteristic restoration subunit of the animal disease data mining model, characteristic restoration is performed based on the cross-domain characteristic path data generated by fusing the disease knowledge characteristic chain with the disease trend knowledge graph, generating a mining result of the animal disease data mining model on the animal disease case data, and obtaining a cross-domain attention knowledge vector of the disease knowledge characteristic chain of the animal disease case data by the animal disease data mining model when generating the mining result; The step of using the cross-domain association unit of the animal disease data mining model to perform cross-domain knowledge association on the disease case feature vector to generate a disease knowledge feature chain of the animal disease case data includes: Using the cross-domain association unit, according to the pre-set classification standard of the knowledge field related to the epidemic, each element in the characteristic vector of the epidemic case is divided into different knowledge subsets, and the knowledge subsets are divided based on the type of epidemic, the category of symptom manifestation, and the source of epidemic infection; Based on the structure of the disease knowledge system and the prior requirements for animal disease diagnosis, a cross-domain knowledge association rule set of the cross-domain association unit is set, and the setting of the cross-domain knowledge association rule revolves around the potential connections between different knowledge subsets; For each knowledge subset, traverse the elements in the knowledge subset, and determine whether the element has an association relationship with the elements in other knowledge subsets according to the cross-domain knowledge association rule set. If so, mark the element with the name or identifier of the other knowledge subset associated with it, so that the elements in each knowledge subset have a clear association relationship with other knowledge subsets, and generate a knowledge subset with an association mark; Based on the knowledge subsets carrying association marks, a preliminary disease knowledge feature chain framework is constructed. The construction process of the preliminary disease knowledge feature chain framework starts from the core knowledge subset and gradually connects the related knowledge subsets along the association marks of the elements. The specific connection method is to arrange the knowledge subsets with association relationships in a logical order to form a chain structure framework. In the chain structure framework, the position of each knowledge subset is determined according to the association relationship between the knowledge subset and other knowledge subsets and the importance weight in disease diagnosis; For each knowledge subset in the preliminary disease knowledge feature chain framework, the elements in the knowledge subset are again arranged according to the advanced and refined association relationships based on the association tags to generate a refined disease knowledge feature chain framework; The detailed disease knowledge feature chain framework is checked for integrity and optimized, wherein, from the perspective of the disease knowledge system, it is checked whether the detailed disease knowledge feature chain framework covers various key knowledge fields of the disease; from the perspective of disease diagnosis, it is checked whether the detailed disease knowledge feature chain framework can perform effective disease diagnosis based on the detailed disease knowledge feature chain framework; whether the detailed disease knowledge feature chain framework has key information missing or unreasonable association relationships; when it is determined that the detailed disease knowledge feature chain framework has integrity problems, the detailed disease knowledge feature chain framework is optimized, and the optimization method includes supplementing missing knowledge elements, adjusting unreasonable association relationships, and rearranging the order of knowledge subsets or elements, thereby generating the disease knowledge feature chain, which is used to reflect the disease knowledge features in the animal disease case data and the association relationships between the disease knowledge features.

4. The data mining method applied to the animal disease remote diagnosis center platform according to claim 1 is characterized in that: The method further comprises: Obtain a sample disease case description data sequence, and for each sample disease case description data in the sample disease case description data sequence, obtain a priori disease condition change trend of the sample disease case description data, wherein the prior disease condition change trend refers to a disease condition change trend of a disease risk factor existing in the sample disease case description data; Obtain an epidemic condition description template, where the epidemic condition description template is used to record whether there is a condition change trend of epidemic risk factors in the epidemic case description data; Based on the prior condition change trend of the sample epidemic case description data and the epidemic condition description template, sample epidemic condition description data is generated, one prior condition change trend generates one sample epidemic condition description data, and the sample epidemic condition description data is used to record whether the prior condition change trend exists in the sample epidemic case description data; Based on the sample epidemic case description data and the sample epidemic condition description data, animal epidemic case data is configured as a positive sample.

5. The data mining method applied to the animal disease remote diagnosis center platform according to claim 4 is characterized in that: The method further comprises: Determine a negative sample condition of the sample disease case description data based on a priori condition change trend of the sample disease case description data, wherein the negative sample condition is a condition change trend of a disease risk factor that does not exist in the sample disease case description data; Based on the negative sample condition of the sample epidemic case description data and the epidemic condition description template, sample epidemic condition description data is generated, one negative sample condition generates one sample epidemic condition description data, and the sample epidemic condition description data is used to record whether the negative sample condition exists in the sample epidemic case description data; Based on the sample disease case description data and the sample epidemic condition description data, animal disease case data is configured as a negative sample.

6. The data mining method applied to the animal disease remote diagnosis center platform according to claim 5 is characterized in that: The determining of negative sample disease conditions of the sample disease case description data based on the prior disease condition change trend of the sample disease case description data includes: Determine multiple negative sample conditions of the sample disease case description data based on the prior condition change trend of the sample disease case description data, generate a negative sample condition sequence of the sample disease case description data, randomly extract the negative sample condition sequence, and generate a negative sample condition as the negative sample condition of the sample disease case description data; or Obtaining the prior condition change trends of other sample disease case description data in the sample disease case description data sequence except the sample disease case description data, extracting the negative sample condition of the sample disease case description data from the prior condition change trends of the other sample disease case description data, generating a negative sample condition sequence of the sample disease case description data, randomly sampling the negative sample condition sequence, generating a negative sample condition as the negative sample condition of the sample disease case description data; or, Determine a shared condition change trend of the prior condition change trend of the sample epidemic case description data, wherein the shared condition change trend is a condition change trend of an epidemic risk factor that does not exist in the sample epidemic case description data and has associated characteristics with the prior condition change trend; based on the shared condition change trend, generate a negative sample condition sequence of the sample epidemic case description data; randomly extract the negative sample condition sequence to generate a negative sample condition as the negative sample condition of the sample epidemic case description data.

7. The data mining method applied to the animal disease remote diagnosis center platform according to claim 6 is characterized in that: The method further comprises: Obtaining a priori disease condition change trend of each sample disease case description data in the sample disease case description data sequence; Based on the sample disease case description data sequence, calculate the correlation between any two prior disease condition change trends, and the simultaneous appearance of two different prior disease condition change trends in the same sample disease case description data is regarded as one correlation; When the correlation between two different prior disease condition change trends meets the target requirement, it is determined that the two different prior disease condition change trends have a correlation characteristic connection.

8. The data mining method applied to the animal disease remote diagnosis center platform according to claim 1 is characterized in that: The mining result of the animal disease case data based on the animal disease data mining model, decomposing the cross-domain attention knowledge vector into a sample attention knowledge vector of risk diffusion trend or a sample attention knowledge vector of non-risk diffusion trend, comprises: Obtaining a mining result of the animal disease data mining model on the animal disease case data, wherein the mining result is one of a first mining result and a second mining result; Based on the category of the mining result and the label data of the animal disease case data, the cross-domain attention knowledge vector is decomposed into a sample attention knowledge vector of a first type of risk diffusion trend, a sample attention knowledge vector of a second type of risk diffusion trend, a sample attention knowledge vector of a first type of non-risk diffusion trend, or a sample attention knowledge vector of a second type of non-risk diffusion trend; The method extracts the sample attention knowledge vector whose attention label attribute is the risk diffusion trend from the sample attention knowledge vector loaded into the first risk diagnosis network, continues to load it into the second risk diagnosis network, generates a second risk diagnosis result of the second risk diagnosis network for the sample attention knowledge vector, and optimizes the second risk diagnosis network based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector, including: Extracting sample attention knowledge vectors with attention label attributes of the first type of risk diffusion trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, and continuously loading them into the second risk diagnosis network of the first network label, generating a second risk diagnosis result of the second risk diagnosis network of the first network label for the sample attention knowledge vector, and optimizing the second risk diagnosis network of the first network label based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector; Extracting sample attention knowledge vectors with attention label attributes of the second type of risk diffusion trend from the sample attention knowledge vectors loaded into the first risk diagnosis network, and continuously loading them into the second risk diagnosis network of the second network label, generating a second risk diagnosis result of the second risk diagnosis network of the second network label for the sample attention knowledge vector, and optimizing the second risk diagnosis network of the second network label based on the second risk diagnosis result and the sample training annotation data of the sample attention knowledge vector; The step of connecting the optimized first risk diagnosis network with the second risk diagnosis network to generate a target risk diagnosis model includes: The optimized first risk diagnosis network is connected with the second risk diagnosis network of the first network label to generate a target risk diagnosis model, wherein the target risk diagnosis model is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model performs risk diagnosis to obtain the first mining result; The optimized first risk diagnosis network is connected to the second risk diagnosis network of the second network label to generate a target risk diagnosis model, which is used to diagnose whether there is a risk diffusion trend when the animal disease data mining model performs risk diagnosis to obtain the second mining result.

9. The data mining method applied to the animal disease remote diagnosis center platform according to any one of claims 1 to 8, characterized in that: The method further comprises: Obtaining a target mining result generated by the animal disease data mining model performing risk diagnosis on target animal disease case data, wherein the target animal disease case data includes a target disease case description data and a target epidemic condition description data, wherein the target epidemic condition description data is used to record whether there is a disease condition change trend of the disease risk factor in the target disease case description data; Obtaining a cross-domain attention knowledge vector of the disease knowledge feature chain of the target animal disease case data from the animal disease data mining model, and loading the cross-domain attention knowledge vector into the target risk diagnosis model, wherein the target risk diagnosis model is generated by connecting the optimized first risk diagnosis network with the second risk diagnosis network, and using the first risk diagnosis network to generate a first risk diagnosis result; When the first risk diagnosis result indicates that the animal disease data mining model has a risk diffusion trend when performing data mining, the cross-domain concern knowledge vector is loaded into the second risk diagnosis network to generate a second risk diagnosis result; When the second risk diagnosis result indicates that the animal disease data mining model has a risk diffusion trend when performing data mining, determining that the animal disease data mining model has a risk diffusion trend when performing data mining on the target animal disease case data; Furthermore, when the first risk diagnosis result represents that the animal disease data mining model does not have a risk diffusion trend when performing data mining, or when the second risk diagnosis result represents that the animal disease data mining model does not have a risk diffusion trend when performing data mining, it is determined that the animal disease data mining model does not have a risk diffusion trend when performing data mining on the target animal disease case data.

10. A remote diagnosis system for animal diseases, characterized in that: The animal disease remote diagnosis system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the data mining method applied to the animal disease remote diagnosis center platform as described in any one of claims 1 to 9 above.

Images