A method for dynamically evaluating immunization effect by combining immunization performance rate with disease monitoring

Abstract

The application discloses an immune effect dynamic evaluation method combining immune execution rate and disease monitoring. The method comprises the following steps: constructing a stress behavior regression model and calculating individual behavior matching degree by analyzing the time series behavior data of live pigs after immunization, so as to dynamically evaluate the immune execution rate of the breeding farm. For the area with low immune execution rate, environmental samples are collected for pathogen detection, the pathogen diffusion trend is analyzed, and then the field area disease pressure index is calculated to evaluate the overall immune effect. When the immune effect is lower than the preset threshold, the system automatically performs immune vulnerability early warning and generates an immune supplement optimization scheme. The application realizes synchronous dynamic monitoring of immune execution and disease risk, can accurately identify weak immune links and guide intervention, and significantly improves the initiative and effectiveness of disease prevention and control.

Description

Technical Field

[0001] This invention relates to the field of smart farming and disease prevention and control technology, and in particular to a dynamic evaluation method for immunization effectiveness that combines immunization implementation rate and disease monitoring. Background Technology

[0002] With the intensive development of the livestock industry, swine disease prevention and control faces enormous challenges. Immunization is a core preventative measure, but traditional methods for evaluating immunization effectiveness suffer from shortcomings such as lag and incompleteness. Currently, the main reliance is on post-immunization antibody testing, which cannot reflect the quality of the immunization process in real time, such as vaccination coverage. Low immunization execution rates are a key risk leading to immunization failure, but current technologies struggle to dynamically assess this. Furthermore, routine disease monitoring is mostly based on static sampling, making it difficult to capture the real-time spread of pathogens within the farm and to dynamically correlate it with the actual protective status of the immunized population. When immunization loopholes exist, susceptible pigs can easily become entry points for disease transmission, and traditional methods struggle to locate and provide timely warnings.

[0003] In recent years, advancements in animal behavior monitoring and environmental pathogen detection technologies have provided new possibilities for dynamic assessment. By analyzing continuous behavioral data of immunized pigs, stress extinction patterns can be identified, thereby assessing immunization implementation. Combined with high-resolution environmental pathogen spread data, the disease pressure on farms can be quantified. However, current technologies have failed to effectively integrate these two types of dynamic information flows to form a closed-loop analysis method that moves from assessing implementation rates through behavior, to assessing disease pressure through pathogen spread, and finally to generating precise optimization plans. Therefore, there is an urgent need for a new method capable of real-time, dynamic assessment of immunization effectiveness and early risk warning, in order to achieve a shift from a passive response to a proactive prevention and control model. Summary of the Invention

[0004] To address at least one of the aforementioned technical problems, this invention proposes a dynamic evaluation method for immunization effectiveness that combines immunization execution rate with disease surveillance.

[0005] The first aspect of this invention provides a method for dynamically evaluating immunization effectiveness by combining immunization execution rate and disease surveillance, comprising: Obtain continuous temporal behavioral data of individual pigs in the target farm after historical preset immunization operations, extract the stress extinction behavior change feature vector of individual pigs after the execution of the immunization operation, and construct a stress behavior extinction model based on the stress extinction behavior change feature vector; Based on the stress behavior extinction model, the real-time behavioral changes of pigs after the target farm has completed the preset immunization operation are fitted to determine the matching degree between the real-time behavioral data of each pig and the stress behavior extinction pattern. The immunization execution rate of the target farm is determined based on the matching degree of each pig. Obtain pathogen detection data from environmental samples in areas where pig herds with immunization execution rates below a preset immunization coverage safety threshold are located, and determine pathogen spread trends based on the environmental sample pathogen detection data; The disease pressure index of the farm area is determined based on the pathogen spread trend, and the immunization effect of the target farm is evaluated based on the disease pressure index of the farm area. If the immune effect is lower than the preset efficacy threshold, an immune vulnerability warning is issued based on the pathogen spread trend, and an immune supplementation optimization plan is generated.

[0006] In this solution, the step of acquiring continuous temporal behavioral data of individual pigs in the target farm after historical preset immunization operations, extracting stress extinction behavior change feature vectors of individual pigs after the immunization operation, and constructing a stress extinction behavior model based on the stress extinction behavior change feature vectors are as follows: The video monitoring data of the target farm after the preset immunization operation was completed in the historical batches is obtained. The video monitoring data is used to identify individual pigs and extract the continuous time-series behavioral data of the individual pigs. The continuous time-series behavioral data includes movement trajectory, activity level, feeding behavior sequence, and lying posture change data. Based on the continuous temporal behavior data, the curvature change frequency and average movement speed of the movement trajectory, the decay and growth rate of activity over time, the delay time and feeding frequency changes of the first resumption of feeding in the feeding behavior sequence, and the posture flipping frequency of the lying posture change data are extracted to form a stress extinction behavior change feature vector characterizing the stress extinction process of individual pigs after the execution of immunization operations. A Gaussian mixture model is introduced as a clustering algorithm to perform clustering operations on the feature vector of stress extinction behavior changes, resulting in several categories representing different stress extinction behavior patterns; For feature vectors belonging to the same category, calculate the mean sequence that changes over time. Use the mean sequence as a typical pattern curve of stress extinction behavior. Combine the typical pattern curves of each category to construct a stress extinction behavior model.

[0007] In this scheme, the introduction of a Gaussian mixture model as a clustering algorithm is used to cluster the feature vectors of stress extinction behavior changes, resulting in several categories representing different stress extinction behavior patterns, specifically: A Gaussian mixture model is introduced, and an initial value for the number of clusters in the Gaussian mixture model is set. The feature vector of the stress extinction behavior change is input into the Gaussian mixture model. The posterior probability of each feature vector belonging to each Gaussian mixture component is calculated iteratively through the expectation-maximization algorithm. The mean, covariance matrix and mixing coefficient of the Gaussian mixture component are updated until the log-likelihood function converges. Multiple cluster centers are obtained based on each Gaussian mixture component after convergence. Based on the membership degree between each stress extinction behavior change feature vector and each cluster center, the stress extinction behavior change feature vector is classified into the category represented by the cluster center with the largest membership degree, thus obtaining several categories that represent different stress extinction behavior patterns.

[0008] In this scheme, the real-time behavioral change data of pigs after the target farm completes the preset immunization operation based on the stress behavior extinction model is fitted to determine the matching degree between the real-time behavioral data of each individual pig and the stress behavior extinction pattern. The immunization execution rate of the target farm is determined based on the matching degree of each pig. Specifically: Acquire video monitoring data and individual pig identification data after the target farm has completed the preset immunization operation, perform individual trajectory association processing on the video monitoring data, and extract real-time behavioral change data of each individual pig within the preset monitoring period; Based on the real-time behavioral change data, the rate of change of curvature of the movement trajectory, the trend of activity fluctuation, the offset of feeding recovery time, and the frequency of posture flipping changes of each individual pig within a continuous time window are calculated and normalized to construct the real-time behavioral feature sequence of the corresponding individual pig. Obtain various typical pattern curves in the stress behavior extinction model, perform dynamic time warping matching on the real-time behavior feature sequence and the various typical pattern curves, and calculate the sequence offset distance and change trend similarity between the real-time behavior feature sequence and the various typical pattern curves. A behavior matching evaluation matrix is ​​constructed based on the sequence offset distance and the similarity of the change trend. The behavior matching evaluation matrix is ​​then subjected to weighted analysis to obtain the fitting probability value of each individual pig for each type of stress extinction behavior pattern. The target stress extinction behavior category corresponding to each pig individual is determined based on the fitted probability value, and the target stress extinction behavior category with the largest fitted probability value is taken as the behavior matching result of the current pig individual; Pigs in the target farm that match the preset normal stress extinction behavior pattern are marked as having completed immunization. The number of pigs that have completed immunization is counted. The immunization execution rate of the target farm is calculated based on the number of pigs and the total number of pigs that have completed the preset immunization procedure.

[0009] In this scheme, the step of obtaining pathogen detection data from environmental samples in areas where pig herds with immunization execution rates below a preset immunization coverage safety threshold are located, and determining the pathogen spread trend based on the environmental sample pathogen detection data, specifically involves: Based on immunization implementation rate and individual pig identification data, the pig pens where pig herds with immunization implementation rates below the preset immunization coverage safety threshold are located are designated as key monitoring areas. Environmental sample collection points are set up in the key monitoring area and the buffer area of ​​the adjacent preset range to collect environmental samples including air dust, wiping material from object surfaces and fecal samples. High-throughput pathogen nucleic acid detection is performed on the collected environmental samples to obtain multi-pathogen detection data and their load information at each sampling point. A spatiotemporal weighted regression model was introduced, with the three-dimensional spatial coordinates and sampling time of each sampling point as independent variables and the corresponding multi-pathogen detection data and their load information as dependent variables, to construct the spatiotemporal distribution surface of pathogen load. Gradient calculation is performed on the spatiotemporal distribution surface to obtain the gradient field of pathogen load change at each sampling point in the spatial and temporal dimensions. Based on the direction and magnitude of the gradient vector in the pathogen load change gradient field, the diffusion path and diffusion intensity of the pathogen are identified. The spread trend of pathogens within farms is determined based on their spread pathways and intensity.

[0010] In this plan, determining the disease pressure index of the farm area based on the pathogen spread trend, and evaluating the immunization effect of the target farm based on the disease pressure index of the farm area, specifically involves: Based on the pathogen spread trend, extract the spatiotemporal coordinates of each location point on the pathogen spread path and the corresponding pathogen load change gradient; The cumulative load of pathogens per unit time during the diffusion path from the starting point to the end point is calculated based on the gradient of pathogen load change. Combined with the spatial distribution density of pigs with immunization execution rate lower than the preset immunization coverage safety threshold at the corresponding location, the local disease pressure intensity is calculated. The local disease pressure intensity is integrated along the pathogen spread path, and the integrated result is weighted and integrated with the reciprocal of the preset basic biosafety coefficient per unit area of ​​the farm to calculate the farm disease pressure index. The immunization effect of the target farm is determined based on the disease pressure index of the farm area.

[0011] In this scheme, if the immunization effect is lower than a preset efficacy threshold, an immunization vulnerability warning is issued based on the pathogen spread trend, and an immunization supplementation optimization scheme is generated, specifically as follows: If the immunization effect is lower than the preset efficacy threshold, an immunization early warning operation is performed on the target farm. Based on the pathogen spread trend, the areas along the pathogen spread path with a pathogen load greater than a preset value and an immunization execution rate lower than the preset immunization coverage safety threshold are marked as immunization vulnerability areas. Based on the pathogen spread trend and the spatial distribution density of individual pigs within the immune vulnerability area, calculate the spread speed of the pathogen to each immune vulnerability area and calculate the estimated spread time of the pathogen to each immune vulnerability area. Obtain historical immunization time data and current immunization protection status of individual pigs corresponding to each immunization vulnerability area. Calculate the difference between the expected transmission time and the remaining effective immunization period corresponding to the current immunization protection status to determine the remaining safe immunization time window for each immunization vulnerability area. Based on the remaining security time window, determine the immune supplementation time point for each immune vulnerability area, and perform immune supplementation operation on the immune vulnerability area at the immune supplementation time point to form an immune supplementation optimization scheme.

[0012] This invention discloses a dynamic evaluation method for immunization effectiveness that combines immunization implementation rate and disease monitoring. The method includes: constructing a stress behavior extinction model by analyzing time-series behavioral data of pigs after immunization, and calculating individual behavior matching degree to dynamically evaluate the immunization implementation rate of the farm. For areas with low immunization implementation rates, environmental samples are collected for pathogen detection, and pathogen spread trends are analyzed to calculate the farm's disease pressure index to evaluate the overall immunization effectiveness. When the immunization effectiveness falls below a preset threshold, the system automatically issues an immunization vulnerability warning and generates an optimized immunization supplementation plan. This invention achieves simultaneous dynamic monitoring of immunization implementation and disease risk, accurately identifies weak links in immunization and guides intervention, significantly improving the initiative and effectiveness of disease prevention and control. Attached Figure Description

[0013] Figure 1 A flowchart of a dynamic evaluation method for immunization effectiveness combining immunization execution rate and disease surveillance according to the present invention is shown; Figure 2 The flowchart illustrating the construction of the stress behavior extinction model of this invention is shown; Figure 3 The flowchart illustrating the present invention for determining the immunization compliance rate of a target farm is shown; Detailed Implementation

[0014] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0015] 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 therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0016] Figure 1 The flowchart of the present invention is shown as a dynamic evaluation method for immunization effectiveness that combines immunization execution rate and disease surveillance.

[0017] like Figure 1As shown, the first aspect of the present invention provides a method for dynamic evaluation of immunization effectiveness by combining immunization execution rate and disease surveillance, comprising: Obtain continuous temporal behavioral data of individual pigs in the target farm after historical preset immunization operations, extract the stress extinction behavior change feature vector of individual pigs after the execution of the immunization operation, and construct a stress behavior extinction model based on the stress extinction behavior change feature vector; Based on the stress behavior extinction model, the real-time behavioral changes of pigs after the target farm has completed the preset immunization operation are fitted to determine the matching degree between the real-time behavioral data of each pig and the stress behavior extinction pattern. The immunization execution rate of the target farm is determined based on the matching degree of each pig. Obtain pathogen detection data from environmental samples in areas where pig herds with immunization execution rates below a preset immunization coverage safety threshold are located, and determine pathogen spread trends based on the environmental sample pathogen detection data; The disease pressure index of the farm area is determined based on the pathogen spread trend, and the immunization effect of the target farm is evaluated based on the disease pressure index of the farm area. If the immune effect is lower than the preset efficacy threshold, an immune vulnerability warning is issued based on the pathogen spread trend, and an immune supplementation optimization plan is generated.

[0018] It should be noted that by analyzing the continuous behavioral time-series data of pigs after historical immunization, typical behavioral patterns of stress reduction can be extracted, thereby establishing an assessment benchmark model. Based on this model, the real-time behavioral data of the pig herd after immunization is fitted to determine whether each pig has truly experienced an effective immune stress and recovery process, thus accurately calculating the true immunization execution rate of the current batch, overcoming the misjudgment of coverage caused by relying on records or spot checks. For areas where the execution rate is not up to standard, pathogen data in the environment is further collected and analyzed to construct a spatiotemporal diffusion trend map of pathogens within the farm area. Combining this diffusion trend with the spatial distribution of immunization gaps, a real-time disease pressure index faced by the entire farm area is calculated. This index comprehensively reflects the superimposed effect of immunization protection gaps and pathogen transmission risks. When the assessment effect is unsatisfactory, the immunization gap area can be accurately located and the risk window period can be predicted based on the pathogen diffusion path and speed. Then, an optimized plan for supplementary immunization in specific areas at specific times can be automatically generated, ultimately realizing a closed-loop management of disease prevention and control from passive treatment to active early warning, and from herd coverage to precise intervention.

[0019] Figure 2 The flowchart illustrating the construction of the stress behavior extinction model of the present invention is shown.

[0020] According to an embodiment of the present invention, the step of obtaining continuous temporal behavioral data of individual pigs in the target farm after historical preset immunization operations, extracting stress extinction behavior change feature vectors of individual pigs after the execution of the immunization operation, and constructing a stress extinction behavior model based on the stress extinction behavior change feature vectors specifically includes: The video monitoring data of the target farm after the preset immunization operation was completed in the historical batches is obtained. The video monitoring data is used to identify individual pigs and extract the continuous time-series behavioral data of the individual pigs. The continuous time-series behavioral data includes movement trajectory, activity level, feeding behavior sequence, and lying posture change data. Based on the continuous temporal behavior data, the curvature change frequency and average movement speed of the movement trajectory, the decay and growth rate of activity over time, the delay time and feeding frequency changes of the first resumption of feeding in the feeding behavior sequence, and the posture flipping frequency of the lying posture change data are extracted to form a stress extinction behavior change feature vector characterizing the stress extinction process of individual pigs after the execution of immunization operations. A Gaussian mixture model is introduced as a clustering algorithm to perform clustering operations on the feature vector of stress extinction behavior changes, resulting in several categories representing different stress extinction behavior patterns; For feature vectors belonging to the same category, calculate the mean sequence that changes over time. Use the mean sequence as a typical pattern curve of stress extinction behavior. Combine the typical pattern curves of each category to construct a stress extinction behavior model.

[0021] It should be noted that after immunization, individual pigs enter a short-term, observable state of stress due to the physical stimulus and the reaction to biological agents. Behavioralally, this state manifests as a series of behavioral changes deviating from baseline, including an abnormally increased curvature of movement trajectories, significant fluctuations in overall activity levels, delayed or interrupted feeding behavior, and an increased frequency of turning over while lying down. These behavioral parameters are not continuously abnormal; rather, they exhibit a continuous temporal process of gradually declining from their peak and eventually returning to normal homeostasis as the stress physiologically subsides. This process is the dynamic decay pattern of stress behavior.

[0022] However, the behavioral responses of different individuals to immunization procedures exhibit natural variations in intensity, duration, and specific manifestations. This variation is a key information carrier for assessing the quality of immunization procedures and individual response status. To extract universally applicable assessment benchmarks from population behavioral data, mathematical modeling of the temporal behavioral characteristics of a large number of historical post-immunization individuals is necessary. Constructing stress extinction models aims to identify and abstract several typical stress extinction behavioral trajectories from multidimensional behavioral feature vectors using machine learning methods such as cluster analysis. These trajectory models can quantitatively characterize standard patterns of behavioral parameter evolution over time, such as the rate of recovery of activity levels from low points to normal, and the typical delayed time window for the recovery of feeding behavior.

[0023] According to an embodiment of the present invention, the introduction of a Gaussian mixture model as a clustering algorithm to cluster the feature vector of stress extinction behavior changes yields several categories representing different stress extinction behavior patterns, specifically: A Gaussian mixture model is introduced, and an initial value for the number of clusters in the Gaussian mixture model is set. The feature vector of the stress extinction behavior change is input into the Gaussian mixture model. The posterior probability of each feature vector belonging to each Gaussian mixture component is calculated iteratively through the expectation-maximization algorithm. The mean, covariance matrix and mixing coefficient of the Gaussian mixture component are updated until the log-likelihood function converges. Multiple cluster centers are obtained based on each Gaussian mixture component after convergence. Based on the membership degree between each stress extinction behavior change feature vector and each cluster center, the stress extinction behavior change feature vector is classified into the category represented by the cluster center with the largest membership degree, thus obtaining several categories that represent different stress extinction behavior patterns.

[0024] It's important to note that the feature vector of stress extinction behavior changes contains multi-dimensional temporal information about behavior, with individual responses exhibiting continuous distribution and natural variation in intensity, speed, and morphology. Gaussian mixture models, as a soft clustering method based on probability distributions, can model this complex data structure as a weighted combination of multiple Gaussian distributions. Each Gaussian component represents a potential typical behavioral pattern; its mean defines the pattern center, the covariance matrix characterizes the correlation and variation range of each feature dimension within the pattern, and the mixture coefficient reflects the prevalence of the pattern in the population. Through iterative optimization using the expectation-maximization algorithm, the model can automatically learn the parameters of these distributions from the data, thereby quantifying the posterior probability of each sample belonging to each category. This probability-based soft classification method accurately handles biological behavioral data with ambiguous category boundaries. Ultimately, by assigning the feature vector to the component with the highest posterior probability, several categories representing different stress extinction behavior patterns can be identified.

[0025] Figure 3 The flowchart illustrating the present invention for determining the immunization compliance rate of a target farm is shown.

[0026] According to an embodiment of the present invention, the fitting operation of the real-time behavioral change data of the pig herd after the target farm completes the preset immunization operation based on the stress behavior extinction model determines the matching degree between the real-time behavioral data of each individual pig and the stress behavior extinction pattern, and determines the immunization execution rate of the target farm based on the matching degree of each pig, specifically as follows: Acquire video monitoring data and individual pig identification data after the target farm has completed the preset immunization operation, perform individual trajectory association processing on the video monitoring data, and extract real-time behavioral change data of each individual pig within the preset monitoring period; Based on the real-time behavioral change data, the rate of change of curvature of the movement trajectory, the trend of activity fluctuation, the offset of feeding recovery time, and the frequency of posture flipping changes of each individual pig within a continuous time window are calculated and normalized to construct the real-time behavioral feature sequence of the corresponding individual pig. Obtain various typical pattern curves in the stress behavior extinction model, perform dynamic time warping matching on the real-time behavior feature sequence and the various typical pattern curves, and calculate the sequence offset distance and change trend similarity between the real-time behavior feature sequence and the various typical pattern curves. A behavior matching evaluation matrix is ​​constructed based on the sequence offset distance and the similarity of the change trend. The behavior matching evaluation matrix is ​​then subjected to weighted analysis to obtain the fitting probability value of each individual pig for each type of stress extinction behavior pattern. The target stress extinction behavior category corresponding to each pig individual is determined based on the fitted probability value, and the target stress extinction behavior category with the largest fitted probability value is taken as the behavior matching result of the current pig individual; Pigs in the target farm that match the preset normal stress extinction behavior pattern are marked as having completed immunization. The number of pigs that have completed immunization is counted. The immunization execution rate of the target farm is calculated based on the number of pigs and the total number of pigs that have completed the preset immunization procedure.

[0027] It should be noted that the stress extinction model is essentially a mathematical benchmark based on historical data from successfully immunized pig herds, establishing how behavioral stress normally subsides after effective immunization. This model incorporates multiple typical behavioral trajectory patterns that smoothly transition from a high-stress state to a normal state. When monitoring pig herds in real time after immunization, the real-time behavioral sequence of each pig is extracted as a feature sequence. Using time-series matching algorithms such as dynamic time warping, the overall morphological similarity and sequence offset distance between the real-time sequence and each preset typical pattern curve in the model are calculated. By constructing a matching evaluation matrix and performing weighted analysis, a quantified fitting probability value is finally output. This probability value objectively reflects the degree of agreement between the current individual behavioral evolution process and the benchmark pattern of "normal stress extinction after immunization." If an individual's behavioral sequence closely matches the typical pattern representing normal stress expiration in the model (i.e., a high fitting probability), it indicates that the individual likely underwent a complete immune stress and physiological recovery process, thus qualifying as "effectively immunized." Conversely, if the individual's behavioral pattern is closer to the "no stress" or "abnormal stress" pattern, it suggests that the individual may not have been vaccinated or the vaccination procedure was ineffective. By quantitatively comparing an individual's real-time behavioral sequence with the typical pattern in the stress expiration model, contactless and accurate determination of each pig's immunization status can be achieved. This allows for real-time and objective identification of individuals whose behavior does not exhibit the normal post-immunization stress expiration pattern, thus accurately pinpointing gaps in the effective execution of immunization procedures. The final calculated immunization execution rate accurately reflects the actual coverage of the expected physiological response elicited by the vaccine. The real-time behavioral change data includes real-time movement trajectory, real-time activity level changes, real-time feeding behavior data, and real-time lying posture changes. The preset normal stress extinction behavior pattern is a standardized temporal trajectory of individual pigs after effective vaccination, showing a gradual return from an abnormal stress state to a normal steady state according to typical patterns, characterized by the curvature of their movement trajectory, activity level, feed recovery time, and postural turning frequency. While some pigs may exhibit no significant stress extinction behavior despite successful vaccination due to differences in constitution or immune tolerance, such cases are extremely rare in actual production. This invention assesses the overall immunization execution rate at the population level; individual exceptions have minimal impact on the statistical results and do not affect the accuracy of the method. Therefore, this invention does not require further processing for pigs that have been successfully vaccinated but exhibit no significant stress extinction behavior.

[0028] According to an embodiment of the present invention, the step of obtaining pathogen detection data from environmental samples in areas where pig herds with immunization execution rates below a preset immunization coverage safety threshold are located, and determining the pathogen spread trend based on the environmental sample pathogen detection data, specifically involves: Based on immunization implementation rate and individual pig identification data, the pig pens where pig herds with immunization implementation rates below the preset immunization coverage safety threshold are located are designated as key monitoring areas. Environmental sample collection points are set up in the key monitoring area and the buffer area of ​​the adjacent preset range to collect environmental samples including air dust, wiping material from object surfaces and fecal samples. High-throughput pathogen nucleic acid detection is performed on the collected environmental samples to obtain multi-pathogen detection data and their load information at each sampling point. It is important to note that focusing on monitoring and analyzing the spread trends of areas with low immunization implementation rates is crucial because, within the semi-closed biological system of a farm, pigs with insufficient immunization coverage constitute a weak link and a susceptible cluster in the overall herd immunity barrier. Pigs in these areas are more likely to become sources of pathogen infection, replication, and shedding. Once the pathogen establishes itself in a location, it will spread from that area to surrounding regions via airflow, personnel and equipment movement, and manure removal. Homogeneous or random environmental monitoring alone will fail to capture this localized spread originating from a "vulnerable area," which may initially be limited but poses a significant threat. By intensifying monitoring of these specific high-risk areas and their surroundings and analyzing their spread trends, true early warning can be achieved. Before the pathogen becomes a widespread epidemic across the farm, its "origin" and "transmission corridor" can be identified, providing crucial decision-making support for accurately calculating disease pressure and implementing timely and targeted interventions such as containment, disinfection, or supplemental immunization.

[0029] A spatiotemporal weighted regression model was introduced, with the three-dimensional spatial coordinates and sampling time of each sampling point as independent variables and the corresponding multi-pathogen detection data and their load information as dependent variables, to construct the spatiotemporal distribution surface of pathogen load. Gradient calculation is performed on the spatiotemporal distribution surface to obtain the gradient field of pathogen load change at each sampling point in the spatial and temporal dimensions. Based on the direction and magnitude of the gradient vector in the pathogen load change gradient field, the diffusion path and diffusion intensity of the pathogen are identified. The spread trend of pathogens within farms is determined based on their spread pathways and intensity.

[0030] It should be noted that by automatically identifying areas with substandard immunization implementation rates as key monitoring areas, and introducing a spatiotemporally weighted regression model to perform high-order analysis on environmental pathogen data, a distribution surface showing the continuous spatial and temporal variation of pathogen load within the production area is constructed. By calculating the gradient of this surface, the direction, path, and intensity of pathogen spread can be accurately determined, thus revealing the true transmission trend of pathogens within the production area in a visualized and quantitative manner. This provides a dynamic, environmentally evidence-based objective data foundation for assessing disease pressure.

[0031] According to an embodiment of the present invention, the step of determining the disease pressure index of the farm area based on the pathogen spread trend, and assessing the immunization effect of the target farm based on the disease pressure index of the farm area, specifically includes: Based on the pathogen spread trend, extract the spatiotemporal coordinates of each location point on the pathogen spread path and the corresponding pathogen load change gradient; The cumulative load of pathogens per unit time during the diffusion path from the starting point to the end point is calculated based on the gradient of pathogen load change. Combined with the spatial distribution density of pigs with immunization execution rate lower than the preset immunization coverage safety threshold at the corresponding location, the local disease pressure intensity is calculated. The local disease pressure intensity is integrated along the pathogen spread path, and the integrated result is weighted and integrated with the reciprocal of the preset basic biosafety coefficient per unit area of ​​the farm to calculate the farm disease pressure index. The immunization effect of the target farm is determined based on the disease pressure index of the farm area.

[0032] It should be noted that the farm disease pressure index, calculated by analyzing the spatiotemporal dynamics of pathogen spread and combining it with the distribution density of susceptible animals in areas with immune gaps, quantifies the potential risk intensity of disease occurrence and transmission within the population under the current immunization coverage status. This index comprehensively reflects the interaction between environmental pathogen load, transmission dynamics, and herd immunity vulnerability, measuring the degree to which existing immune barriers offset the actual disease transmission pressure within the farm. This allows the assessment conclusions to go beyond statistical data on immunization implementation, directly linking them to the realistic possibility of disease outbreaks. It provides a risk-based decision-making basis for determining whether the immunization program has built an effective defense and whether enhanced intervention measures are needed. The farm disease pressure index characterizes the degree of pathogen threat currently faced by the entire target farm; the higher the farm disease pressure index, the worse the immunization effect.

[0033] According to an embodiment of the present invention, if the immunization effect is lower than a preset efficacy threshold, an immunization vulnerability warning is issued based on the pathogen spread trend, and an immunization supplementation optimization plan is generated, specifically as follows: If the immunization effect is lower than the preset efficacy threshold, an immunization early warning operation is performed on the target farm. Based on the pathogen spread trend, the areas along the pathogen spread path with a pathogen load greater than a preset value and an immunization execution rate lower than the preset immunization coverage safety threshold are marked as immunization vulnerability areas. Based on the pathogen spread trend and the spatial distribution density of individual pigs within the immune vulnerability area, calculate the spread speed of the pathogen to each immune vulnerability area and calculate the estimated spread time of the pathogen to each immune vulnerability area. Obtain historical immunization time data and current immunization protection status of individual pigs corresponding to each immunization vulnerability area. Calculate the difference between the expected transmission time and the remaining effective immunization period corresponding to the current immunization protection status to determine the remaining safe immunization time window for each immunization vulnerability area. Based on the remaining security time window, determine the immune supplementation time point for each immune vulnerability area, and perform immune supplementation operation on the immune vulnerability area at the immune supplementation time point to form an immune supplementation optimization scheme.

[0034] It should be noted that, based on pathogen spread pathways and immunization implementation rate data, high-risk immunization vulnerability areas are accurately identified. Subsequently, by calculating the estimated time for pathogen transmission to each vulnerability area and comparing it with the remaining immunization protection period of individual pigs within that area, a specific and differentiated remaining safe immunization time window is calculated for each vulnerability area. Ultimately, the scheme generates a clearly defined immunization supplementation optimization plan, which can generate a differentiated optimal supplemental immunization time point for each high-risk area based on the timeliness of risk transmission. This ensures that immunization intervention measures can be completed within the critical time window before the pathogen reaches the susceptible population, thereby establishing a timely and effective immune barrier along the disease transmission path and maximizing the protective efficacy of immunization resources. The specific execution method of the immunization supplementation optimization plan generated by this invention is as follows: based on the remaining safe time window of each immunization vulnerability area, differentiated supplemental immunization instructions are automatically pushed to the breeding management system, and managers administer targeted supplemental immunization to pigs in designated pens at the specified time points.

[0035] According to an embodiment of the present invention, it further includes: Acquire multimodal synchronous data of the target farm after the preset immunization operation is completed. The multimodal synchronous data includes continuous environmental sound data collected by audio sensors deployed inside and outside the pig house, and continuous time-series video data of pig herd behavior collected by the video monitoring system. The environmental sound data is analyzed in real time to identify and mark sudden high-decibel noise events, and the temporal energy burst characteristics, frequency distribution characteristics and duration of the noise events are extracted. The noise type is determined based on the characteristics, and the corresponding interfered video time period is located in the continuous temporal behavior video data based on the start and end timestamps of the noise events. Extract the behavior data of the pig herd within the time period of the interfered video, and calculate the overall behavioral consistency index and spatial clustering mutation index of the pig herd within this time period. Combined with the noise type, construct an interference pattern vector describing the coupling relationship between noise and herd behavior response. When extracting the stress extinction behavior change feature vector of individual pigs, the individual behavior data within the time period of the interfered video is identified and marked according to the interference pattern vector. An adaptive filtering algorithm based on the interference pattern vector is applied to the real-time behavior change data within the time period to filter out the instantaneous and highly consistent behavioral mutation signals directly caused by sudden noise, and a corrected individual behavior data sequence is obtained. Based on the corrected individual behavior data sequence, the frequency of movement trajectory curvature change, activity decay rate, feed recovery delay time and posture reversal frequency of individual pigs within the corresponding time window are recalculated to generate a corrected stress extinction behavior change feature vector, which is used to subsequently construct a stress behavior extinction model and calculate the immunization execution rate.

[0036] It is important to note that sudden environmental noises inside and outside the pigsty, such as equipment malfunctions, thunderstorms, or construction noises, can instantly cause panic and abnormal group behavior in the pigs. This transient, highly consistent behavioral disturbance triggered by external factors overlaps with the behavioral monitoring window after immunization, severely contaminating the collected continuous temporal behavioral data of individual pigs and distorting the characteristic vector of stress-induced behavioral changes. Behavioral models constructed based on this distorted vector and calculations of immunization execution rates will exhibit systematic biases. Therefore, by simultaneously collecting and fusing audio and video data, we can accurately identify and label sudden noise events and their corresponding periods of interference in the pig herd. The core principle is that it doesn't simply remove interfering data, but intelligently analyzes the noise characteristics and the overall response pattern of the pig herd, constructing a "noise-behavioral response" model, and then using this model to perform targeted filtering and correction on the individual behavioral data during the affected periods. This process effectively removes the transient, herd-wide behavioral noise caused by external noise, thereby restoring individual behavioral signals that truly reflect the process of immunization stress reduction. Finally, the corrected data was used to generate accurate stress extinction behavior feature vectors, laying a solid foundation for the subsequent construction of a reliable behavior extinction model and the calculation of the true immune execution rate. This significantly improved the anti-interference ability of the entire dynamic evaluation system of immune effect in complex real-world environments and the accuracy of the evaluation results.

[0037] The adaptive filtering algorithm based on the interference pattern vector first constructs a multi-dimensional reference signal containing noise type, group behavior consistency intensity, and spatial clustering abrupt change amplitude. Next, within the time period of the interfered video, the algorithm performs sliding window analysis on the real-time behavioral change data of individual pigs (including instantaneous activity, posture flipping frequency, and abrupt movement trajectory changes), calculating the correlation coefficient between the individual behavioral data and the reference signal in terms of trend and intensity within each time window. Based on the correlation coefficient, the algorithm dynamically generates a time-varying filtering weight, which is proportional to the correlation. For high-correlation periods, the algorithm determines that individual behavior is mainly driven by noise, and applies strong filtering to the original behavioral data sequence, significantly suppressing abrupt signal changes during that period through local weighted regression smoothing. For low-correlation periods, most features of the original behavioral data sequence are retained. Finally, the algorithm outputs a corrected individual behavioral data sequence that filters out non-specific behavioral abrupt changes highly synchronized with noise events.

[0038] This invention discloses a dynamic evaluation method for immunization effectiveness that combines immunization implementation rate and disease monitoring. The method includes: constructing a stress behavior extinction model by analyzing time-series behavioral data of pigs after immunization, and calculating individual behavior matching degree to dynamically evaluate the immunization implementation rate of the farm. For areas with low immunization implementation rates, environmental samples are collected for pathogen detection, and pathogen spread trends are analyzed to calculate the farm's disease pressure index to evaluate the overall immunization effectiveness. When the immunization effectiveness falls below a preset threshold, the system automatically issues an immunization vulnerability warning and generates an optimized immunization supplementation plan. This invention achieves simultaneous dynamic monitoring of immunization implementation and disease risk, accurately identifies weak links in immunization and guides intervention, significantly improving the initiative and effectiveness of disease prevention and control.

[0039] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0040] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for dynamically evaluating immunization effectiveness by combining immunization execution rate and disease surveillance, characterized in that, Includes the following steps: Obtain continuous temporal behavioral data of individual pigs in the target farm after historical preset immunization operations, extract the stress extinction behavior change feature vector of individual pigs after the execution of the immunization operation, and construct a stress behavior extinction model based on the stress extinction behavior change feature vector; Based on the stress behavior extinction model, the real-time behavioral changes of pigs after the target farm has completed the preset immunization operation are fitted to determine the matching degree between the real-time behavioral data of each pig and the stress behavior extinction pattern. The immunization execution rate of the target farm is determined based on the matching degree of each pig. Obtain pathogen detection data from environmental samples in areas where pig herds with immunization execution rates below a preset immunization coverage safety threshold are located, and determine pathogen spread trends based on the environmental sample pathogen detection data; The disease pressure index of the farm area is determined based on the pathogen spread trend, and the immunization effect of the target farm is evaluated based on the disease pressure index of the farm area. If the immune effect is lower than the preset efficacy threshold, an immune vulnerability warning is issued based on the pathogen spread trend, and an immune supplementation optimization plan is generated.

2. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 1, characterized in that, The process involves acquiring continuous temporal behavioral data of individual pigs in the target farm after historically preset immunization operations, extracting stress extinction behavior change feature vectors of individual pigs after the immunization operation, and constructing a stress extinction behavior model based on the stress extinction behavior change feature vectors. Specifically: The video monitoring data of the target farm after the preset immunization operation was completed in the historical batches is obtained. The video monitoring data is used to identify individual pigs and extract the continuous time-series behavioral data of the individual pigs. The continuous time-series behavioral data includes movement trajectory, activity level, feeding behavior sequence, and lying posture change data. Based on the continuous temporal behavior data, the curvature change frequency and average movement speed of the movement trajectory, the decay and growth rate of activity over time, the delay time and feeding frequency changes of the first resumption of feeding in the feeding behavior sequence, and the posture flipping frequency of the lying posture change data are extracted to form a stress extinction behavior change feature vector characterizing the stress extinction process of individual pigs after the execution of immunization operations. A Gaussian mixture model is introduced as a clustering algorithm to perform clustering operations on the feature vector of stress extinction behavior changes, resulting in several categories representing different stress extinction behavior patterns; For feature vectors belonging to the same category, calculate the mean sequence that changes over time. Use the mean sequence as a typical pattern curve of stress extinction behavior. Combine the typical pattern curves of each category to construct a stress extinction behavior model.

3. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 2, characterized in that, The Gaussian mixture model is introduced as a clustering algorithm to cluster the feature vectors of stress extinction behavior changes, resulting in several categories representing different stress extinction behavior patterns, specifically: A Gaussian mixture model is introduced, and an initial value for the number of clusters in the Gaussian mixture model is set. The feature vector of the stress extinction behavior change is input into the Gaussian mixture model. The posterior probability of each feature vector belonging to each Gaussian mixture component is calculated iteratively through the expectation-maximization algorithm. The mean, covariance matrix and mixing coefficient of the Gaussian mixture component are updated until the log-likelihood function converges. Multiple cluster centers are obtained based on each Gaussian mixture component after convergence. Based on the membership degree between each stress extinction behavior change feature vector and each cluster center, the stress extinction behavior change feature vector is classified into the category represented by the cluster center with the largest membership degree, thus obtaining several categories that represent different stress extinction behavior patterns.

4. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 1, characterized in that, The process involves fitting real-time behavioral changes in pigs after completing a pre-defined immunization procedure at the target farm using a stress extinction model. This determines the matching degree between the real-time behavioral data of each individual pig and the stress extinction pattern. Based on the matching degree of each pig, the immunization execution rate of the target farm is determined. Specifically: Acquire video monitoring data and individual pig identification data after the target farm has completed the preset immunization operation, perform individual trajectory association processing on the video monitoring data, and extract real-time behavioral change data of each individual pig within the preset monitoring period; Based on the real-time behavioral change data, the rate of change of curvature of the movement trajectory, the trend of activity fluctuation, the offset of feeding recovery time, and the frequency of posture flipping changes of each individual pig within a continuous time window are calculated and normalized to construct the real-time behavioral feature sequence of the corresponding individual pig. Obtain various typical pattern curves in the stress behavior extinction model, perform dynamic time warping matching on the real-time behavior feature sequence and the various typical pattern curves, and calculate the sequence offset distance and change trend similarity between the real-time behavior feature sequence and the various typical pattern curves. A behavior matching evaluation matrix is ​​constructed based on the sequence offset distance and the similarity of the change trend. The behavior matching evaluation matrix is ​​then subjected to weighted analysis to obtain the fitting probability value of each individual pig for each type of stress extinction behavior pattern. The target stress extinction behavior category corresponding to each pig individual is determined based on the fitted probability value, and the target stress extinction behavior category with the largest fitted probability value is taken as the behavior matching result of the current pig individual; Pigs in the target farm that match the preset normal stress extinction behavior pattern are marked as having completed immunization. The number of pigs that have completed immunization is counted. The immunization execution rate of the target farm is calculated based on the number of pigs and the total number of pigs that have completed the preset immunization procedure.

5. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 1, characterized in that, The process of obtaining pathogen detection data from environmental samples in areas where immunization implementation rates are lower than a preset immunization coverage safety threshold, and determining pathogen spread trends based on this environmental sample pathogen detection data, specifically involves: Based on immunization implementation rate and individual pig identification data, the pig pens where pig herds with immunization implementation rates below the preset immunization coverage safety threshold are located are designated as key monitoring areas. Environmental sample collection points are set up in the key monitoring area and the buffer area of ​​the adjacent preset range to collect environmental samples including air dust, wiping material from object surfaces and fecal samples. High-throughput pathogen nucleic acid detection is performed on the collected environmental samples to obtain multi-pathogen detection data and their load information at each sampling point. A spatiotemporal weighted regression model was introduced, with the three-dimensional spatial coordinates and sampling time of each sampling point as independent variables and the corresponding multi-pathogen detection data and their load information as dependent variables, to construct the spatiotemporal distribution surface of pathogen load. Gradient calculation is performed on the spatiotemporal distribution surface to obtain the gradient field of pathogen load change at each sampling point in the spatial and temporal dimensions. Based on the direction and magnitude of the gradient vector in the pathogen load change gradient field, the diffusion path and diffusion intensity of the pathogen are identified. The spread trend of pathogens within farms is determined based on their spread pathways and intensity.

6. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 1, characterized in that, The process of determining the disease pressure index of the farm area based on the pathogen spread trend, and assessing the immunization effect of the target farm based on the disease pressure index, specifically involves: Based on the pathogen spread trend, extract the spatiotemporal coordinates of each location point on the pathogen spread path and the corresponding pathogen load change gradient; The cumulative load of pathogens per unit time during the diffusion path from the starting point to the end point is calculated based on the gradient of pathogen load change. Combined with the spatial distribution density of pigs with immunization execution rate lower than the preset immunization coverage safety threshold at the corresponding location, the local disease pressure intensity is calculated. The local disease pressure intensity is integrated along the pathogen spread path, and the integrated result is weighted and integrated with the reciprocal of the preset basic biosafety coefficient per unit area of ​​the farm to calculate the farm disease pressure index. The immunization effect of the target farm is determined based on the disease pressure index of the farm area.

7. The method for dynamic evaluation of immunization effectiveness combining immunization execution rate and disease surveillance as described in claim 1, characterized in that, If the immunization effect is lower than a preset efficacy threshold, an immunization vulnerability warning is issued based on the pathogen spread trend, and an immunization supplementation optimization plan is generated, specifically as follows: If the immunization effect is lower than the preset efficacy threshold, an immunization early warning operation is performed on the target farm. Based on the pathogen spread trend, the areas along the pathogen spread path with a pathogen load greater than a preset value and an immunization execution rate lower than the preset immunization coverage safety threshold are marked as immunization vulnerability areas. Based on the pathogen spread trend and the spatial distribution density of individual pigs within the immune vulnerability area, calculate the spread speed of the pathogen to each immune vulnerability area and calculate the estimated spread time of the pathogen to each immune vulnerability area. Obtain historical immunization time data and current immunization protection status of individual pigs corresponding to each immunization vulnerability area. Calculate the difference between the expected transmission time and the remaining effective immunization period corresponding to the current immunization protection status to determine the remaining safe immunization time window for each immunization vulnerability area. Based on the remaining security time window, determine the immune supplementation time point for each immune vulnerability area, and perform immune supplementation operation on the immune vulnerability area at the immune supplementation time point to form an immune supplementation optimization scheme.

Images