A Big Data-Based Epidemiological Investigation System and Method for Swine Epidemic Prevention and Control

Abstract

This invention discloses a big data-based epidemiological investigation system and method for swine epidemic prevention and control, belonging to the field of animal health information technology. The system includes: identifying abnormal swine signals and trajectory intersections in structured epidemic prevention characteristic data using intelligent algorithms; comparing viral gene differences; constructing a virus transmission network map; and outputting a high-risk transmission node map. Using the super-transmission hub node of the high-risk transmission node map as the origin, a three-dimensional aerosol dynamics model is constructed in conjunction with breeding environment parameters to simulate the airborne diffusion path of the virus and generate an environmental transmission heat map. The high-risk transmission node map and the environmental transmission heat map are integrated, and a multi-dimensional risk superposition model is used to identify cross-infection areas for dynamic risk classification, outputting a graded early warning report. This invention achieves a breakthrough improvement in swine epidemic prevention and control capabilities, accurately identifying key nodes and potential risk areas in epidemic transmission, and improving the targeting of prevention and control measures.

Description

Technical Field

[0001] This invention relates to the field of animal health information technology, and in particular to a big data-based epidemiological investigation system and method for swine epidemic prevention and control. Background Technology

[0002] In the field of animal disease prevention and control, intelligent monitoring systems based on the Internet of Things (IoT) architecture have gradually become the industry standard technical solution. Current mainstream technologies deploy distributed sensor networks to achieve real-time monitoring of the entire pig farming process, integrating multiple types of data acquisition terminals to construct a comprehensive monitoring network. Optimized machine learning classification algorithms are used to process the collected physiological characteristic data in real time, and geographic information technology is used to correlate the analysis results with spatial location information to generate risk heat maps with tiered early warning capabilities. This type of technical solution has formed a complete technology chain from data perception, transmission and processing to decision output, providing modern animal husbandry with digital and intelligent disease monitoring methods.

[0003] Current animal disease control and prevention systems still have limitations in analyzing virus transmission pathways. Existing technologies mainly rely on static environmental parameter collection and simplified transmission models, which cannot accurately track the dynamic diffusion process of virus aerosols in complex environments. This technical deficiency leads to a significant discrepancy between risk assessment results and actual transmission situations, resulting in a lack of targeted basis for the formulation of prevention and control measures. Because it is impossible to accurately quantify the transmission range and risk intensity of virus particles under different environmental conditions, it is difficult to accurately delineate isolation areas and set disinfection frequencies. Static analysis models cannot reflect the spatiotemporal characteristics of virus transmission, resulting in a mismatch between the allocation of prevention and control resources and the actual risk distribution. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a big data-based epidemiological investigation method for swine epidemic prevention and control to address the problem of insufficient quantitative analysis of virus transmission paths in the environment, which leads to a lack of targeted prevention and control measures.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a big data-based method for epidemiological investigation of swine epidemic prevention and control, which includes collecting swine epidemic prevention characteristic data and preprocessing it to output structured epidemic prevention characteristic data; the swine epidemic prevention characteristic data includes swine vital signs data, breeding environment parameters and logistics trajectory data;

[0008] By using intelligent algorithms to identify the intersection features of abnormal pig signals and trajectories in structured epidemic prevention characteristic data, comparing the differences in viral genes, constructing a virus transmission network map, and outputting a map of high-risk transmission nodes;

[0009] Using the super-spreading hub node of the high-risk transmission node map as the origin, and combining aquaculture environment parameters, a three-dimensional spatial model of aerosol dynamics is constructed to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map.

[0010] By integrating high-risk transmission node maps with environmental transmission heat maps, a multi-dimensional risk overlay model is used to identify cross-infection areas, dynamically classify risks, and output graded early warning reports.

[0011] Based on the tiered early warning report, a strategy knowledge base is invoked to generate prevention and control strategies, and machine learning is used to optimize these strategies, outputting the optimal prevention and control strategy.

[0012] As a preferred embodiment of the big data-based epidemiological investigation method for swine epidemic prevention and control described in this invention, the method involves: filtering outliers, filling in missing values, correcting trace drift, and fusing multi-source time labels and spatial location labels on swine epidemic prevention characteristic data to output structured epidemic prevention characteristic data.

[0013] As a preferred embodiment of the big data-based swine epidemic prevention and control epidemiological investigation method described in this invention, the steps of identifying abnormal swine signals and trajectory intersection features in structured epidemic prevention characteristic data using intelligent algorithms, comparing viral gene differences, constructing a virus transmission network graph, and outputting a high-risk transmission node map are as follows.

[0014] A deep convolutional neural network was trained using historical structured epidemic prevention feature data to obtain the trained deep convolutional neural network.

[0015] By scanning the structured epidemic prevention feature data of pigs with a trained deep convolutional neural network, abnormal body temperature and sudden drop in activity level of pigs are identified, and abnormal pig signals are obtained.

[0016] Based on the logistics trajectory data of structured epidemic prevention feature data, the DBSCAN algorithm is used to calculate the degree of overlap of the stops of transport vehicles at transport nodes and output the trajectory intersection features.

[0017] Viral gene sequences were extracted from swine vital signs data from structured disease prevention feature data, compared with historical gene feature databases, and the viral evolutionary distance was calculated to output viral gene difference.

[0018] By integrating abnormal pig signals, trajectory intersection features, and viral gene differences, a graph theory algorithm is used to construct a virus transmission network graph, identify super-spreading hub nodes, and output a high-risk transmission node map.

[0019] As a preferred embodiment of the big data-based swine epidemic prevention and control epidemiological investigation method described in this invention, the following steps are taken: A three-dimensional spatial model of aerosol dynamics is constructed using the super-spreading hub node of the high-risk transmission node map as the origin, combined with breeding environment parameters, to simulate the airborne diffusion path of the virus and generate an environmental transmission heat distribution map.

[0020] Analyze the latitude and longitude information of super propagation hub nodes in the high-risk propagation node map to generate the location coordinates of super propagation hub nodes;

[0021] The coordinates of the super-propagation hub node are combined with real-time aquaculture environment parameters, and a three-dimensional spatial model of aerosol dynamics is constructed using fluid dynamics control equations.

[0022] Based on the three-dimensional spatial model of aerosol dynamics, the virus movement trajectory was numerically analyzed in three-dimensional space using the finite volume method to obtain the virus aerosol diffusion trajectory.

[0023] Spatial kernel density analysis and risk concentration labeling were performed on the viral aerosol diffusion trajectory to obtain an environmental transmission thermal distribution map.

[0024] As a preferred embodiment of the big data-based swine epidemic prevention and control epidemiological investigation method described in this invention, the method involves integrating high-risk transmission node maps and environmental transmission heat maps, identifying cross-infection areas through a multi-dimensional risk overlay model, dynamically classifying risks, and outputting a graded early warning report. The specific steps are as follows:

[0025] Biological transmission factors were extracted from the high-risk transmission node map, and environmental transmission factors were analyzed from the environmental transmission heat map.

[0026] Biological transmission factors are geographically weighted using spatial overlay analysis to obtain a geographically weighted risk grid. An improved analytic hierarchy process is then used to dynamically calculate and optimize the consistency of environmental transmission factors, generating a dynamic weight matrix for environmental risk.

[0027] Spatial weighted fusion and dynamic coupling analysis were performed on the geographic weighted risk grid and the environmental risk dynamic weight matrix to obtain a multi-dimensional risk superposition model.

[0028] By aligning the topological data of the high-risk propagation node map with the spatial grid of the environmental propagation heat map in the coordinate system, composite risk data is generated.

[0029] By using a multi-dimensional risk overlay model, spatial probability density analysis and risk threshold classification assessment are performed on composite risk data, outputting a risk concentration matrix and performing risk level visualization rendering to obtain a graded early warning report.

[0030] As a preferred embodiment of the big data-based swine epidemic prevention and control epidemiological investigation method of the present invention, the specific steps for generating prevention and control strategies based on the hierarchical early warning report calling strategy knowledge base are as follows:

[0031] Feature extraction and strategy association mapping are performed on historical structured epidemic prevention characteristic data to obtain a strategy knowledge base;

[0032] By utilizing tiered early warning reports, historical epidemic response cases can be retrieved from the strategy knowledge base to output prevention and control strategies.

[0033] As a preferred embodiment of the big data-based swine epidemic prevention and control epidemiological investigation method of the present invention, the step of optimizing the prevention and control strategy through machine learning and outputting the optimal prevention and control strategy includes the following specific steps.

[0034] By utilizing historical prevention and control strategies, a domain knowledge-enhanced XGBoost framework is trained to obtain a model for predicting the effectiveness of prevention and control strategies.

[0035] The prevention and control strategy is input into the prevention and control strategy effect prediction model for multi-dimensional effect evaluation, and the strategy execution effect prediction data is obtained. The optimal prevention and control strategy is then output through multi-objective genetic algorithm optimization iteration.

[0036] Secondly, this invention provides a big data-based epidemiological investigation system for swine epidemic prevention and control, comprising:

[0037] The data acquisition module is used to collect swine disease prevention characteristic data and preprocess it to output structured disease prevention characteristic data; the swine disease prevention characteristic data includes swine physical condition data, breeding environment parameters and logistics trajectory data;

[0038] The transmission analysis module is used to identify abnormal pig signals and trajectory intersection features in structured epidemic prevention characteristic data through intelligent algorithms, compare viral gene differences, construct a virus transmission network diagram, and output a high-risk transmission node map.

[0039] The diffusion simulation module is used to construct a three-dimensional spatial model of aerosol dynamics based on the super-spreading hub node of the high-risk transmission node map as the origin and combined with aquaculture environment parameters to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map.

[0040] The decision-making module is used to integrate high-risk transmission node maps with environmental transmission heat maps, identify cross-infection areas through a multi-dimensional risk overlay model, dynamically classify risks, and output graded early warning reports.

[0041] The optimization module is used to generate prevention and control strategies based on the tiered early warning reports by calling the strategy knowledge base, and to optimize the prevention and control strategies through machine learning, outputting the optimal prevention and control strategy.

[0042] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the big data-based swine epidemic prevention and control epidemiological investigation method as described in the first aspect of the present invention.

[0043] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the big data-based swine epidemic prevention and control epidemiological investigation method as described in the first aspect of the present invention.

[0044] The beneficial effects of this invention are as follows: By integrating virus transmission network analysis and aerosol diffusion simulation, a breakthrough improvement in the ability to prevent and control swine epidemics has been achieved. It can accurately identify key nodes and potential risk areas in epidemic transmission, improving the targeting of prevention and control measures. Simultaneously, by dynamically simulating virus transmission paths, it effectively solves the lag problem inherent in static assessment methods. This multi-dimensional analysis technology not only improves the accuracy of epidemic tracing and early warning but also optimizes the allocation efficiency of prevention and control resources, making the prevention and control response more timely and effective. The resulting tiered early warning and optimized strategy output mechanism provides decision support for livestock enterprises, realizing a shift from passive response to proactive prevention and control, and improving the level of prevention and control of major animal epidemics. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart of a big data-based epidemiological investigation method for swine epidemic prevention and control.

[0047] Figure 2 This is a schematic diagram of a big data-based swine epidemic prevention and control epidemiological investigation system.

[0048] Figure 3 A flowchart for outputting a high-risk propagation node map.

[0049] Figure 4 This is a flowchart for outputting the optimal prevention and control strategy. Detailed Implementation

[0050] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0051] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0052] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0053] Reference Figures 1-4 This is one embodiment of the present invention, which provides a big data-based method for epidemiological investigation of swine epidemic prevention and control, including the following steps:

[0054] S1. Collect swine disease prevention characteristic data and preprocess it to output structured disease prevention characteristic data; the swine disease prevention characteristic data includes swine physical condition data, breeding environment parameters and logistics trajectory data.

[0055] S1.1 Perform outlier filtering, missing value imputation, trace drift correction, and multi-source time label and spatial location label fusion on the swine disease prevention feature data to output structured disease prevention feature data.

[0056] Specifically, outlier filtering is performed on body temperature monitoring records in the pig vital signs data; records with body temperature values ​​exceeding the exemplary upper limit of 42 degrees Celsius or the exemplary lower limit of 37 degrees Celsius are marked as invalid data; missing value imputation is performed on exercise monitoring records; error correction is performed on temperature and humidity sensor readings in the breeding environment parameters; and trajectory drift correction is performed on GPS coordinate points in the logistics trajectory data.

[0057] The timestamps of swine disease prevention characteristic data, including swine vital signs data, breeding environment parameters, and logistics trajectory data, are aligned with a fixed time window; body temperature values, activity levels, temperature and humidity values, ammonia concentration values, and GPS coordinate points are spatially correlated and mapped according to the aligned timestamps; and a structured disease prevention characteristic data table containing timestamps, spatial coordinates, and values ​​of each monitoring indicator is output.

[0058] S2. Using intelligent algorithms, identify the intersection features of abnormal pig signals and trajectories in structured epidemic prevention characteristic data, compare the differences in virus genes, construct a virus transmission network map, and output a high-risk transmission node map.

[0059] S2.1. Train the deep convolutional neural network using historical structured epidemic prevention feature data to obtain the trained deep convolutional neural network.

[0060] Specifically, the process involves extracting swine vital signs data, breeding environment parameters, and logistics trajectory data from historical structured disease prevention characteristic data; performing feature engineering on swine vital signs data such as body temperature monitoring records, activity monitoring records, and breeding environment parameters such as temperature and humidity readings and ammonia concentration to generate a training sample set; and then standardizing the training sample set.

[0061] The standardized training sample set is input into a deep convolutional neural network (DCNN), configured with alternating stacked convolutional and pooling layers. The convolutional layers use an exemplary 3×3 kernel for local feature extraction, and a ReLU activation function combined with a non-linear transformation is applied immediately after each convolution operation. The pooling layers employ max pooling, with an exemplary 2×2 sampling window and a downsampling parameter of stride 2. A fully connected layer is configured at the end of the DCNN, with the output dimension matching the classification task requirements. The training process uses a mini-batch gradient descent algorithm, with a batch size of 32 samples as shown. The prediction error is calculated using the cross-entropy loss function, and the weight parameters of the DCNN are adjusted layer by layer using backpropagation. The learning rate is initialized to an exemplary 0.001 with an exponential decay strategy, and the training epochs are set to an exemplary 100 complete epochs. After each training epoch, the performance of the DCNN is evaluated on a validation set. Training is terminated early when the validation loss does not decrease for five consecutive exemplary epochs. Finally, the optimal weight parameters of the DCNN are saved to obtain the trained DCNN.

[0062] It should be noted that the feature engineering processing of the body temperature monitoring records, exercise monitoring records, and temperature and humidity readings and ammonia concentration in the breeding environment parameters of pigs specifically involves: segmenting and discretizing the body temperature monitoring records; calculating the sliding window statistic for the exercise monitoring records; extracting time series features from the temperature and humidity sensor readings; normalizing the ammonia concentration detection values; and then splicing the processed features to form a training sample set.

[0063] S2.2. By scanning the structured epidemic prevention feature data of pigs with the trained deep convolutional neural network, abnormal body temperature and abnormal sudden drop in exercise volume of pigs are identified, and abnormal pig signals are obtained.

[0064] Specifically, the structured epidemic prevention feature data of pigs is input into a trained deep convolutional neural network. Pig body temperature monitoring records and pig activity monitoring records are processed by convolutional layers to extract spatiotemporal features, which are then processed by the ReLU activation function and passed to the pooling layer for feature compression. The fully connected layer outputs abnormal body temperature probability values ​​and abnormal activity probability values. An exemplary abnormal probability threshold of 0.85 is set. When the abnormal body temperature probability value of a pig exceeds the abnormal probability threshold, it is determined to be an abnormal body temperature. When the abnormal activity probability value of a pig exceeds the abnormal probability threshold and decreases by more than 40% compared to the previous period, it is determined to be an abnormal activity drop. Pigs with abnormal body temperature and abnormal activity drops are marked as abnormal pig signals, and an abnormal pig signal containing the abnormality type, occurrence time, and individual identifier is output.

[0065] It should be noted that the abnormality probability threshold is set by analyzing the distribution of prediction results of the trained deep convolutional neural network on the validation set, balancing the recognition accuracy and false alarm rate. For example, an abnormality probability threshold of 0.85 keeps the recognition rate of real abnormal cases at a high level, while controlling the proportion of healthy pigs misclassified as abnormal to a low range.

[0066] S2.3 Based on the logistics trajectory data of structured epidemic prevention feature data, the DBSCAN algorithm is used to calculate the overlap of the stops of transport vehicles at transport nodes and output the trajectory intersection features; the transport nodes include farms and trading markets.

[0067] Specifically, the process involves extracting logistics trajectory data from structured epidemic prevention feature data to obtain the GPS coordinates and timestamps of transport vehicles; performing spatial clustering analysis on the GPS coordinates, setting the neighborhood radius of the DBSCAN algorithm to an exemplary 100 meters and the minimum number of neighborhood points to an exemplary 3, to identify the areas where transport vehicles stop around transport nodes in farms and trading markets; obtaining the overlap of different transport vehicles' stay time windows in the same farm and trading market, setting an exemplary time overlap threshold of 30 minutes, and determining a trajectory intersection event when the time periods of two or more transport vehicles staying in the same farm and trading market overlap for more than the time overlap threshold; and outputting trajectory intersection features including the location of transport nodes, the identifiers of the intersecting vehicles, and the overlapping time periods.

[0068] It should be noted that the process of setting the time overlap threshold involves collecting loading and unloading operation records from typical farms and trading markets, and statistically analyzing the distribution of the actual dwell time of transport vehicles at transportation nodes. The data sample covers farms of different sizes (exemplary stock ranging from 500 to 5000 head) and three types of trading markets (origin-based, destination-based, and transit-based) to ensure the universality of the time overlap threshold setting; approximately 85% of the routine loading and unloading operations take place within the exemplary 30-45 minute range, so the exemplary time overlap threshold is set at 30 minutes.

[0069] S2.4 Extract the viral gene sequence from the structured disease prevention feature data of pigs, perform multi-sequence alignment with the historical gene feature database, calculate the viral evolutionary distance, and output the viral gene difference degree.

[0070] Specifically, viral gene sequences are extracted from the swine physical condition data of structured disease prevention feature data to obtain the nucleotide sequences of specific segments of the viral genome; the extracted viral gene sequences are compared with reference sequences in the historical gene feature library using multiple sequence alignment, and a position-by-position alignment algorithm is used to identify insertion, deletion, and substitution mutation sites; the number of differentially expressed nucleotides in the compared viral gene sequences is counted, and the ratio of the number of differentially expressed nucleotides to the total length of the viral gene sequence is obtained as the original difference rate. When the original difference rate exceeds the example difference rate of 0.05, it is determined to be a significant viral gene difference; the viral gene difference degree, which includes the number of differentially expressed sites, mutation type, and original difference rate, is output.

[0071] S2.5 integrates abnormal pig signals, trajectory intersection features, and viral gene differences, uses graph theory algorithms to construct a virus transmission network graph, identifies super-spreading hub nodes, and outputs a high-risk transmission node map.

[0072] Specifically, the process involves associating and matching individual pig identifiers from abnormal pig signals with transport vehicle identifiers from trajectory intersection features to establish a network of contact relationships between individual pigs and transport nodes. An initial propagation network graph is constructed using individual pigs as vertices and transport vehicle trajectory intersection events as edges. Viral gene diversity is used as edge weights; the greater the viral gene diversity, the higher the edge weight. The betweenness centrality algorithm in graph theory is used to calculate the centrality score of nodes in the initial propagation network graph, with an exemplary centrality threshold of 0.75. Nodes whose centrality scores exceed the threshold are identified as super-propagation hub nodes. The output is a high-risk propagation node map containing node locations, centrality scores, and associated propagation paths.

[0073] It should be noted that the betweenness centrality algorithm in graph theory is used to calculate the centrality scores of nodes in the initial propagation network graph:

[0074] ;

[0075] in, It is a node in the initial propagation network graph. The betweenness centrality score, It is a node in the initial propagation network graph. To the nodes in the initial propagation network graph The shortest path length, It is a distance function. It must pass through the nodes in the initial propagation network graph. Nodes in the initial propagation network graph To the nodes in the initial propagation network graph The shortest path length, It is a node in the initial propagation network graph. To the nodes in the initial propagation network graph The path, It is a node in the initial propagation network graph. To the nodes in the initial propagation network graph Edge weights.

[0076] S3. Using the super-spreading hub node of the high-risk transmission node map as the origin, and combining it with aquaculture environment parameters, construct a three-dimensional spatial model of aerosol dynamics to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map.

[0077] S3.1 Analyze the latitude and longitude information of the super propagation hub node in the high-risk propagation node map and generate the location coordinates of the super propagation hub node.

[0078] Specifically, the process involves extracting transportation node information marked as super-propagation hub nodes from the high-risk propagation node map, obtaining the GPS positioning coordinates of the transportation node information of super-propagation hub nodes recorded in the logistics trajectory data, verifying the validity of the GPS coordinates, and removing signal drift points. An example drift judgment criterion is: speed > 120 km / h or horizontal accuracy > 10 meters. Valid GPS coordinates are converted into latitude and longitude values ​​in the WGS84 coordinate system, retaining 6 decimal places. The output is the location coordinates of the super-propagation hub nodes, including node identifiers, longitude, and latitude fields.

[0079] S3.2 Combine the location coordinates of the super propagation hub node with real-time aquaculture environment parameters, and construct a three-dimensional spatial model of aerosol dynamics through a set of fluid dynamic control equations.

[0080] Specifically, the location coordinates of the super-propagation hub nodes, from which latitude and longitude information is extracted, are used as the spatial origin for aerosol diffusion simulation. Real-time aquaculture environmental parameters, including temperature, humidity, wind speed, wind direction monitoring data, and ammonia concentration detection values, are simultaneously acquired. The location coordinates of the super-propagation hub nodes are aligned with the environmental parameters by timestamps to establish a three-dimensional data cube containing spatial coordinates, environmental parameters, and time stamps. Based on the Navier-Stokes control equations, fluid dynamics control equations are constructed, with exemplary initial conditions of wind speed 2 m / s, relative humidity 60%, and temperature 25℃. The solution domain is discretized using the finite volume method. In the computational grid, the super-propagation hub nodes are defined as point pollution sources. The location coordinates of the super-propagation hub nodes and real-time aquaculture environmental parameters are substituted into the fluid dynamics control equations and iteratively solved to obtain the aerosol concentration distribution in the three-dimensional data cube, outputting a three-dimensional spatial model of aerosol dynamics.

[0081] S3.3 Based on the three-dimensional spatial model of aerosol dynamics, the virus movement trajectory is numerically analyzed in three-dimensional space using the finite volume method to obtain the virus aerosol diffusion trajectory.

[0082] Specifically, the discretized numerical solutions of the fluid dynamics control equations are extracted from the three-dimensional spatial model of aerosol dynamics to obtain aerosol concentration distribution data on the three-dimensional data cube. Using the Lagrange particle tracking algorithm, 10,000 virtual virus particles are released within the computational domain. Each virtual virus particle carries a viral load proportional to the local aerosol concentration. The particle trajectory is calculated using a fourth-order Runge-Kutta integral, with a time step set to an exemplary 0.1 seconds. Tracking is terminated when the particle movement exceeds the computational boundary or after an exemplary 4 hours of simulation. The movement paths of all particles in the three-dimensional data cube are statistically analyzed, and the virus aerosol diffusion trajectory, including the coordinates of the trajectory start point, the sequence of movement path points, and the endpoint position, is output.

[0083] It should be noted that the expression for calculating the particle trajectory using the fourth-order Runge-Kutta method integral is as follows:

[0084] ;

[0085] in, It is the slope of stage 1. It is the velocity field function. Particles in the first Spatial coordinate vectors at each time step It is the first The moment value of each time step;

[0086] ;

[0087] in, It is the slope of stage 2. It is the time step;

[0088] ;

[0089] in, It is the slope of stage 3;

[0090] ;

[0091] in, It is the slope of stage 4;

[0092] ;

[0093] in, It is the viral aerosol particles completing the current time step. The calculated trajectory of the moving particle.

[0094] S3.4. Spatial kernel density analysis and risk concentration labeling of the viral aerosol diffusion trajectory are performed to obtain an environmental transmission thermal distribution map.

[0095] Specifically, the process involves extracting the motion path sequence of all particles from the viral aerosol diffusion trajectory, obtaining the three-dimensional spatial coordinates and timestamp of each path point; dividing the three-dimensional data cube into aerosol concentration grid cells of an exemplary 0.5m × 0.5m × 0.2m, counting the number of particles passing through each aerosol concentration grid cell, and obtaining the product of particle residence time and local aerosol concentration as the risk density value; using a Gaussian kernel density estimation algorithm to spatially smooth the risk density value, with the kernel function bandwidth set to an exemplary 1.0m; and classifying the viral aerosol concentration into five exemplary levels (low, medium, high, low, high ... The virus aerosol concentration levels are categorized into low, medium, medium-high, and high, each marked with a different color. For example, the colors for marking virus aerosol concentration levels are: Low (green RGB(0,128,0)), Low-medium (yellow RGB(255,255,0)), Medium (orange RGB(255,165,0)), Medium-high (red RGB(255,0,0)), and High (dark red RGB(139,0,0)), corresponding to the 0-20%, 20-40%, 40-60%, 60-80%, and 80-100% percentile concentration ranges, respectively. The output is an environmental transmission heat map containing grid coordinates, risk level, and aerosol concentration values.

[0096] S4. Integrate high-risk transmission node maps with environmental transmission heat maps, identify cross-infection areas through a multi-dimensional risk overlay model, dynamically classify risks, and output graded early warning reports.

[0097] S4.1 Extract biological transmission factors from the high-risk transmission node map and analyze environmental transmission factors from the environmental transmission heat distribution map.

[0098] Specifically, the betweenness centrality score, number of associated propagation paths, and weight data of super propagation hub nodes are extracted from the high-risk propagation node map as biological propagation factors; the risk level and aerosol concentration value of each aerosol concentration grid cell are read from the environmental propagation heat distribution map, and the mean aerosol concentration, spatial distribution range, and diffusion direction of high-risk areas (marked in red and dark red) are extracted as environmental propagation factors; the biological propagation factors and environmental propagation factors are aligned according to spatial coordinates.

[0099] S4.2. Geographically weighted calculation of biological transmission factors is performed using spatial overlay analysis to obtain a geographically weighted risk grid. Then, dynamic weight calculation and consistency optimization of environmental transmission factors are performed using an improved hierarchical analysis method to generate a dynamic weight matrix of environmental risks.

[0100] Specifically, the process involves extracting betweenness centrality scores, the number of associated propagation paths, and the weights of super-spreading hub nodes from biological propagation factors, converting them into a spatial raster format with an exemplary resolution of 0.5m × 0.5m; spatial interpolation of biological propagation factors using an inverse distance weighting method, with a weight attenuation coefficient set to an exemplary 2.0, to generate a geographically weighted risk raster; extracting the mean aerosol concentration, spatial distribution range, and diffusion direction of high-risk areas from environmental propagation factors to construct a judgment matrix, with exemplary element values ​​ranging from 1 to 9; performing consistency correction on the judgment matrix using an improved analytic hierarchy process, and extracting eigenvectors of the judgment matrix using eigenvalue optimization, which serve as the initial relative weights for each environmental propagation factor; performing a consistency check on the initial relative weights of each environmental propagation factor, outputting the environmental propagation factor weights when the consistency ratio is less than the exemplary 0.1, otherwise adjusting the judgment matrix until the condition is met; and multiplying the optimized environmental propagation factor weights with the corresponding environmental propagation factors to generate a dynamic environmental risk weight matrix.

[0101] It should be noted that the improved analytic hierarchy process (AHP) has improved upon the traditional AHP by adding a dynamic weight adjustment mechanism and a real-time consistency optimization function. Specific improvements include replacing static expert scoring with automatic judgment matrix generation based on real-time monitoring data, such as updating matrix elements hourly based on environmental sensor readings.

[0102] Machine learning algorithms are used to automatically correct the consistency ratio. For example, when the CR (consistency ratio) > 0.1, the judgment matrix is ​​adjusted using gradient descent.

[0103] By adding time-sensitivity constraints to the weights and setting an exemplary weight decay coefficient of 0.95 / hour, the contribution rate of older data is ensured to decrease over time. These improvements enable the weight allocation to dynamically respond to changes in aquaculture environment parameters, addressing the problem of insufficient adaptability of traditional AHP in epidemic transmission scenarios.

[0104] S4.3. Spatial weighted fusion and dynamic coupling analysis are performed on the geographic weighted risk grid and the environmental risk dynamic weight matrix to obtain a multi-dimensional risk superposition model.

[0105] Specifically, the geographic weighted risk grid is aligned with the environmental risk dynamic weight matrix according to spatial coordinates to ensure that each geographic weighted risk grid cell corresponds one-to-one with the element of the environmental risk dynamic weight matrix; the biological transmission factor value and the environmental transmission factor weight at each spatial location are multiplied to obtain the comprehensive risk value.

[0106] The contribution ratio of biological transmission factor and environmental transmission factor is adjusted by using a dynamic coupling coefficient, with an exemplary initial weight of 0.6 for biological transmission factor and 0.4 for environmental transmission factor. Local data noise is eliminated by using a sliding window analysis method (exemplary window size 3×3 grid). The fused comprehensive risk value is normalized and converted into a relative risk index in the 0-1 range. The output is a multi-dimensional risk overlay model containing the coordinates of each geographic weighted risk grid cell, the comprehensive risk value, and the risk level.

[0107] S4.4 Align the topological data of the high-risk propagation node map with the spatial grid of the environmental propagation heat distribution map to generate composite risk data.

[0108] Specifically, the process involves extracting super-spreading hub node topology data from a high-risk propagation node map, including super-spreading hub node identifiers, edge weights, and spatial coordinates; reading the center coordinates and risk values ​​of geographically weighted risk raster cells from an environmental propagation heat map; uniformly converting the super-spreading hub node topology data and the center coordinates and risk values ​​of geographically weighted risk raster cells to the WGS84 geographic coordinate system, with UTM partitioning used for planar projection; performing nearest neighbor matching between node coordinates and geographically weighted risk raster center coordinates, setting an exemplary matching tolerance of 0.5 meters; when a super-spreading hub node falls into a geographically weighted risk raster cell, associating the super-spreading hub node attributes with the geographically weighted risk raster risk value; and outputting composite risk data containing the super-spreading hub node identifier, topology features, matched geographically weighted risk raster coordinates, and corresponding risk values.

[0109] S4.5. Using a multi-dimensional risk superposition model, spatial probability density analysis and risk threshold classification assessment are performed on composite risk data, outputting a risk concentration matrix and performing risk level visualization rendering to obtain a graded early warning report.

[0110] Specifically, the system extracts super-propagation hub node attributes and geographically weighted risk raster risk values ​​from composite risk data, inputs them into a multi-dimensional risk overlay model for spatial kernel density analysis, and sets the kernel function bandwidth to an exemplary 1.0 meter. The multi-dimensional risk overlay model integrates temporal evolution characteristics, spatial adjacency relationships, and propagation factor weight information to achieve joint modeling and response intensity assessment of risk factors from different sources. The kernel density function uses a Gaussian kernel function to smoothly weight risk events within a local space, highlighting high-frequency clustering areas. The system obtains the risk probability density value for each spatial location and generates a risk concentration matrix.

[0111] The probability density values ​​in the risk concentration matrix are classified into percentiles, with the 20%, 40%, 60%, and 80% percentiles set as examples, dividing them into five risk levels: low (0-20%), low-medium (20-40%), medium (40-60%), medium-high (60-80%), and high (80-100%). The risk probability densities of different regions are standardized and sorted to ensure that the distribution of each risk level is reasonable and spatially distinguishable. The classification results are mapped to a spatial coordinate system, and color-coded rendering is performed using regional boundaries and environmental layers to form a risk level visualization layer.

[0112] The color gradation scheme can use a standard five-segment continuous color spectrum, from cool to warm colors: blue (low risk), light green (low to medium risk), yellow (medium risk), orange (medium to high risk), and red (high risk), enabling intuitive risk identification; and outputting graded early warning reports.

[0113] S5. Generate prevention and control strategies based on the strategy knowledge base invoked from the hierarchical early warning report.

[0114] S5.1. Extract features and map strategies to historical structured epidemic prevention data to obtain a strategy knowledge base.

[0115] Specifically, the process involves extracting swine vital signs data (body temperature, activity level), breeding environment parameters, and logistics trajectory data from historical structured epidemic prevention feature data; performing anomaly detection on the swine vital signs data, marking those with body temperature exceeding the exemplary 39.5℃ or activity levels below the exemplary 200 units / hour as abnormal; aligning the breeding environment parameters and logistics trajectory data by timestamp to generate a spatiotemporal correlated dataset; using the FP-Growth algorithm to mine high-frequency prevention and control strategy combinations, with a minimum support set to the exemplary 0.3; obtaining the effectiveness score of the prevention and control strategy, storing it in the strategy knowledge base when the accuracy exceeds the exemplary 90%; and outputting a strategy knowledge base containing strategy rules, implementation conditions, and effectiveness scores.

[0116] S5.2 Utilize the tiered early warning report to retrieve historical epidemic response cases from the strategy knowledge base and output prevention and control strategies.

[0117] Specifically, the risk level, spatial location, and time range information are extracted from the graded early warning report. Historical epidemic response cases that match the conditions are retrieved from the strategy knowledge base. When the risk level, spatial distribution pattern, and viral genotype similarity all exceed the exemplary 0.8, the case is determined to be a valid match. The prevention and control strategy records in the matched cases are extracted and sorted in descending order of implementation effect score. The prevention and control strategy containing the strategy name, applicable conditions, implementation steps, and expected effect is output.

[0118] S6. Optimize prevention and control strategies through machine learning and output the optimal prevention and control strategy.

[0119] S6.1. Using historical prevention and control strategies, the domain knowledge-enhanced XGBoost framework is trained to obtain a prediction model for the effectiveness of prevention and control strategies.

[0120] Specifically, the process involves extracting feature vectors from historical prevention and control strategies, including implementation conditions (risk level, environmental parameters), measures (disinfection intensity, isolation scope), and effectiveness indicators (control period, loss rate). These feature vectors are then converted into numerical matrices, with missing values ​​filled using the mean of the same strategy category. Hyperparameters of the domain knowledge-enhanced XGBoost framework are configured, with the maximum tree depth set to an exemplary 6 layers and the learning rate set to an exemplary 0.01. The training and validation sets are divided in a 7:3 ratio, and the domain knowledge-enhanced XGBoost framework is trained using five-fold cross-validation. Training stops when the mean absolute error on the validation set falls below the exemplary 0.15, and the structure and parameters of the domain knowledge-enhanced XGBoost framework are saved. Finally, a model predicting the effectiveness of prevention and control strategies, including feature importance ranking and prediction error indicators, is output.

[0121] S6.2 Input the prevention and control strategy into the prevention and control strategy effect prediction model for multi-dimensional effect evaluation, obtain the strategy execution effect prediction data, and optimize and iterate through a multi-objective genetic algorithm to output the optimal prevention and control strategy.

[0122] Specifically, the process involves extracting risk level data and environmental parameter data from the prevention and control strategy, obtaining data on the content of measures, and recording the radius of the isolation fence and the frequency of quarantine. Non-numerical data is converted into standardized numerical features, missing fields are filled with the historical average of similar strategies, and outliers are removed using box plot rules. The processed prevention and control strategy is input into the prevention and control strategy effect prediction model, and multi-dimensional effect evaluation calculations are performed, including forward propagation to calculate the predicted control cycle value, backpropagation to optimize the loss function gradient, and cross-validation to evaluate the stability of the prevention and control strategy effect prediction model. The model outputs an estimated number of days for the epidemic control cycle (example range: 3 to 15 days), an estimated economic loss in the tens of thousands of yuan (example accuracy: ±50,000 yuan), and resource consumption (human and material costs, example unit: tens of thousands of yuan). Time-related indicators are logarithmically transformed to the 0-1 range, economic indicators are standardized to the 0-1 range using Min-Max, and resource indicators are converted to budget percentages from 0% to 100%. Finally, the process generates prediction data for the strategy implementation effect.

[0123] A three-objective optimization function is established to minimize the control cycle and economic loss, while constraining resource consumption to not exceed the budget hard constraint. The genetic algorithm parameters are configured using the NSGA-II framework, with a population size of 100 individuals (exemplary). The crossover operator uses a simulated binary crossover probability of 0.9 (exemplary), the mutation operator uses a polynomial mutation probability of 0.1, and the selection mechanism uses a tournament selection size of 3. An initial population satisfying the constraints is randomly generated, and the three-objective function value for each individual is calculated. Non-dominated sorting is performed to classify the frontier levels, and crowding distance is obtained to maintain diversity. The hypervolume index convergence is checked to terminate the iteration. Feasibility verification is performed on the Pareto front solution set, eliminating schemes with an isolation radius exceeding the exemplary 5 km. A three-dimensional target space scatter plot visualization output is generated. Finally, a structured optimal strategy report is output.

[0124] This embodiment also provides a big data-based swine epidemic prevention and control epidemiological investigation system, including:

[0125] The data acquisition module is used to collect swine disease prevention characteristic data and preprocess it to output structured disease prevention characteristic data; the swine disease prevention characteristic data includes swine physical condition data, breeding environment parameters and logistics trajectory data;

[0126] The transmission analysis module is used to identify abnormal pig signals and trajectory intersection features in structured epidemic prevention characteristic data through intelligent algorithms, compare viral gene differences, construct a virus transmission network diagram, and output a high-risk transmission node map.

[0127] The diffusion simulation module is used to construct a three-dimensional spatial model of aerosol dynamics based on the super-spreading hub node of the high-risk transmission node map as the origin and combined with aquaculture environment parameters to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map.

[0128] The decision-making module is used to integrate high-risk transmission node maps with environmental transmission heat maps, identify cross-infection areas through a multi-dimensional risk overlay model, dynamically classify risks, and output graded early warning reports.

[0129] The optimization module is used to generate prevention and control strategies based on the tiered early warning reports by calling the strategy knowledge base, and to optimize the prevention and control strategies through machine learning, outputting the optimal prevention and control strategy.

[0130] This embodiment also provides a computer device applicable to the epidemiological investigation method for swine epidemic prevention and control based on big data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the epidemiological investigation method for swine epidemic prevention and control based on big data as proposed in the above embodiment.

[0131] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0132] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the big data-based swine epidemic prevention and control epidemiological investigation method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0133] In summary, this invention achieves a breakthrough in swine epidemic prevention and control capabilities by integrating virus transmission network analysis and aerosol diffusion simulation. It can accurately identify key nodes and potential risk areas in epidemic transmission, improving the targeting of prevention and control measures. Simultaneously, by dynamically simulating virus transmission paths, it effectively solves the lag problem inherent in static assessment methods. This multi-dimensional analysis technology not only improves the accuracy of epidemic tracing and early warning but also optimizes the allocation efficiency of prevention and control resources, making the prevention and control response more timely and effective. The resulting tiered early warning and optimized strategy output mechanism provides decision support for livestock enterprises, enabling a shift from passive response to proactive prevention and control, and improving the level of prevention and control of major animal epidemics.

[0134] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A big data-based epidemiological investigation method for swine epidemic prevention and control, characterized in that: include, Collect swine disease prevention characteristic data, and preprocess it to output structured disease prevention characteristic data; the swine disease prevention characteristic data includes swine physical condition data, breeding environment parameters and logistics trajectory data; The following steps are taken to identify abnormal pig signals and trajectory intersections in structured epidemic prevention data using intelligent algorithms, compare viral gene differences, construct a virus transmission network diagram, and output a high-risk transmission node map: A deep convolutional neural network was trained using historical structured epidemic prevention feature data to obtain the trained deep convolutional neural network. By scanning structured disease prevention feature data with a trained deep convolutional neural network, abnormal pig body temperature and sudden drops in activity levels are identified, thus obtaining abnormal pig signals. Based on structured epidemic prevention feature data and logistics trajectory data, the DBSCAN algorithm is used to calculate the overlap degree of vehicle dwell times at transportation nodes and output trajectory intersection features. Viral gene sequences were extracted from swine vital signs data from structured disease prevention feature data, compared with historical gene feature databases, and viral evolutionary distance was calculated to output viral gene dissimilarity. By integrating abnormal pig signals, trajectory intersection features, and viral gene differences, a graph theory algorithm is used to construct a virus transmission network graph, identify super-spreading hub nodes, and output a high-risk transmission node map. Using the super-spreading hub node of the high-risk transmission node map as the origin, and combining aquaculture environment parameters, a three-dimensional spatial model of aerosol dynamics is constructed to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map. By integrating high-risk transmission node maps with environmental transmission heat maps, a multi-dimensional risk overlay model is used to identify cross-infection areas, dynamically classify risks, and output graded early warning reports. Based on the tiered early warning report, a strategy knowledge base is invoked to generate prevention and control strategies, and machine learning is used to optimize these strategies, outputting the optimal prevention and control strategy.

2. The method for epidemiological investigation of swine epidemic prevention and control based on big data as described in claim 1, characterized in that: Outlier filtering, missing value imputation, trajectory drift correction, and fusion of multi-source time labels and spatial location labels are performed on the swine disease prevention feature data to output structured disease prevention feature data.

3. The method for epidemiological investigation of swine epidemic prevention and control based on big data as described in claim 1, characterized in that: The process involves constructing a three-dimensional aerosol dynamics model using the super-spreading hub node of the high-risk transmission node map as the origin, combined with aquaculture environment parameters, to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map. The specific steps are as follows: Analyze the latitude and longitude information of super propagation hub nodes in the high-risk propagation node map to generate the location coordinates of super propagation hub nodes; The coordinates of the super-propagation hub node are combined with real-time aquaculture environment parameters, and a three-dimensional spatial model of aerosol dynamics is constructed using fluid dynamics control equations. Based on the three-dimensional spatial model of aerosol dynamics, the virus movement trajectory was numerically analyzed in three-dimensional space using the finite volume method to obtain the virus aerosol diffusion trajectory. Spatial kernel density analysis and risk concentration labeling were performed on the viral aerosol diffusion trajectory to obtain an environmental transmission thermal distribution map.

4. The method for epidemiological investigation of swine epidemic prevention and control based on big data as described in claim 3, characterized in that: The method integrates high-risk transmission node maps and environmental transmission heat maps, identifies cross-infection areas through a multi-dimensional risk overlay model, dynamically classifies risks, and outputs graded early warning reports. The specific steps are as follows. Biological transmission factors were extracted from the high-risk transmission node map, and environmental transmission factors were analyzed from the environmental transmission heat map. Biological transmission factors are geographically weighted using spatial overlay analysis to obtain a geographically weighted risk grid. An improved analytic hierarchy process is then used to dynamically calculate and optimize the consistency of environmental transmission factors, generating a dynamic weight matrix for environmental risk. Spatial weighted fusion and dynamic coupling analysis were performed on the geographic weighted risk grid and the environmental risk dynamic weight matrix to obtain a multi-dimensional risk superposition model. By aligning the topological data of the high-risk propagation node map with the spatial grid of the environmental propagation heat map in the coordinate system, composite risk data is generated. By using a multi-dimensional risk overlay model, spatial probability density analysis and risk threshold classification assessment are performed on composite risk data, outputting a risk concentration matrix and performing risk level visualization rendering to obtain a graded early warning report.

5. The method for epidemiological investigation of swine epidemic prevention and control based on big data as described in claim 4, characterized in that: The specific steps for generating prevention and control strategies based on the hierarchical early warning report and the strategy knowledge base are as follows. Feature extraction and strategy association mapping are performed on historical structured epidemic prevention characteristic data to obtain a strategy knowledge base; By utilizing tiered early warning reports, historical epidemic response cases can be retrieved from the strategy knowledge base to output prevention and control strategies.

6. The method for epidemiological investigation of swine epidemic prevention and control based on big data as described in claim 5, characterized in that: The steps for optimizing the prevention and control strategy using machine learning and outputting the optimal strategy are as follows. By utilizing historical prevention and control strategies, a domain knowledge-enhanced XGBoost framework is trained to obtain a model for predicting the effectiveness of prevention and control strategies. The prevention and control strategy is input into the prevention and control strategy effect prediction model for multi-dimensional effect evaluation, and the strategy execution effect prediction data is obtained. The optimal prevention and control strategy is then output through multi-objective genetic algorithm optimization iteration.

7. A big data-based swine epidemic prevention and control epidemiological investigation system, based on the big data-based swine epidemic prevention and control epidemiological investigation method according to any one of claims 1 to 6, characterized in that: include, The data acquisition module is used to collect swine disease prevention characteristic data and preprocess it to output structured disease prevention characteristic data; the swine disease prevention characteristic data includes swine physical condition data, breeding environment parameters and logistics trajectory data; The transmission analysis module is used to identify abnormal pig signals and trajectory intersections in structured epidemic prevention characteristic data through intelligent algorithms, compare viral gene differences, construct a virus transmission network diagram, and output a high-risk transmission node map. The specific steps are as follows: A deep convolutional neural network was trained using historical structured epidemic prevention feature data to obtain the trained deep convolutional neural network. By scanning structured disease prevention feature data with a trained deep convolutional neural network, abnormal pig body temperature and sudden drops in activity levels are identified, thus obtaining abnormal pig signals. Based on structured epidemic prevention feature data and logistics trajectory data, the DBSCAN algorithm is used to calculate the overlap degree of vehicle dwell times at transportation nodes and output trajectory intersection features. Viral gene sequences were extracted from swine vital signs data from structured disease prevention feature data, compared with historical gene feature databases, and viral evolutionary distance was calculated to output viral gene dissimilarity. By integrating abnormal pig signals, trajectory intersection features, and viral gene differences, a graph theory algorithm is used to construct a virus transmission network graph, identify super-spreading hub nodes, and output a high-risk transmission node map. The diffusion simulation module is used to construct a three-dimensional spatial model of aerosol dynamics based on the super-spreading hub node of the high-risk transmission node map as the origin and combined with aquaculture environment parameters to simulate the airborne diffusion path of the virus and generate an environmental transmission thermal distribution map. The decision-making module is used to integrate high-risk transmission node maps with environmental transmission heat maps, identify cross-infection areas through a multi-dimensional risk overlay model, dynamically classify risks, and output graded early warning reports. The optimization module is used to generate prevention and control strategies based on the tiered early warning report by calling the strategy knowledge base, and to optimize the prevention and control strategies through machine learning, outputting the optimal prevention and control strategy.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the big data-based swine epidemic prevention and control epidemiological investigation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the big data-based swine epidemic prevention and control epidemiological investigation method as described in any one of claims 1 to 6.

Images