Chuzhou crucian carp spore worm disease risk early warning method and system based on multiple environmental factors

Abstract

The present application relates to the technical field of aquaculture disease early warning, in particular to a Chuzhou crucian carp ciliates disease risk early warning method and system based on multiple environmental factors, comprising: collecting environmental data through multiple sensor nodes, generating a comprehensive state matrix representing the coordinated changes of different water layer factors through spatiotemporal correlation processing; calling a ciliates disease risk knowledge graph for pathogen suitability assessment, identifying specific risk periods and spatial coordinates; then combining physiological behavior data of the cultured objects, constructing a multi-dimensional risk propagation path model simulating pathogen spread. Finally, according to the model output, a risk early warning level and corresponding precise control schemes such as oxygenation, water flow disturbance and feeding adjustment are generated. This method breaks through the limitations of traditional single threshold alarm, realizing early, accurate and dynamic early warning of ciliates disease risk.

Description

Technical Field

[0001] This invention relates to the field of aquaculture disease early warning technology, and in particular to a method and system for early warning of crucian carp disease risk based on multiple environmental factors. Background Technology

[0002] In pond aquaculture, sporozoan diseases are a serious type of parasitic disease. Existing disease early warning technologies mainly rely on independent monitoring and threshold alarms of single or a few environmental factors. These methods involve deploying sensors at various water layers, triggering an alarm when a parameter exceeds a preset safety range. Current solutions typically perform isolated analysis or simple averaging of data collected from different locations and times, lacking a comprehensive evaluation of the overall state of the environmental system.

[0003] This monitoring method, based on independent parameter thresholds, has its limitations. It cannot capture and quantify the dynamic synergistic changes of key environmental factors such as temperature, dissolved oxygen, and pH across different water layers, and these synergistic patterns are often key triggers for pathogen proliferation. Furthermore, existing technologies struggle to combine static environmental conditions with the dynamic behavior of farmed organisms, making it impossible to simulate and predict the specific transmission paths and risk evolution processes of pathogens in the three-dimensional space and time dimensions of the pond. This results in delayed and untargeted early warnings, and control measures often have limited effectiveness.

[0004] A method is needed that can analyze the spatiotemporal synergistic changes of multiple factors in the pond environment and dynamically simulate and assess the spread risk of pathogens under the interaction between the environment and the host. This requires the technical solution to not only diagnose the overall state of the environment, but also to achieve a leap from "risk condition identification" to "risk transmission prediction". Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a risk warning method and system for crucian carp spore disease based on multiple environmental factors.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a risk early warning method for crucian carp spore disease based on multiple environmental factors, comprising:

[0007] Multiple environmental sensor nodes are deployed to continuously collect dynamic environmental factor data sets of the pond. The dynamic environmental factor data sets are then processed for spatiotemporal correlation to generate a comprehensive state matrix of the pond environment. The comprehensive state matrix is ​​used to characterize the coordinated change patterns of environmental factors in different water layers.

[0008] By calling a pre-constructed knowledge graph of spore-borne diseases, pathogen suitability assessment is performed on the comprehensive state matrix to identify potential risk periods and spatial coordinates of risk water layers that meet the conditions for spore proliferation.

[0009] Based on the potential risk period and the spatial coordinates of the risk water layer, and combined with the physiological behavior monitoring data of the pond aquaculture objects, a multidimensional risk transmission path model is constructed. The multidimensional risk transmission path model is used to simulate the spread of pathogens in the pond environment.

[0010] Based on the output of the multidimensional risk transmission path model, a risk warning level for spore-borne diseases and a corresponding environmental control scheme are generated. The environmental control scheme includes oxygenation strategies, water flow disturbance strategies, and feed feeding adjustment schemes.

[0011] As a further aspect of the present invention, the dynamic environmental factor data set is subjected to spatiotemporal correlation processing to generate a comprehensive state matrix of the pond environment, including:

[0012] The dynamic environmental factor dataset includes water temperature stratification data, dissolved oxygen concentration data, pH data, organic suspended matter concentration data, and time-series data on phytoplankton abundance.

[0013] The water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data from the same water depth are time-stamped and aligned to form a time series set of environmental parameters for each of the multiple water layers.

[0014] For each water layer, the cross-correlation function between water temperature and dissolved oxygen concentration, and the covariance matrix between pH and organic suspended matter concentration are calculated.

[0015] Extract the population change curves of specific indicative plankton species from the time-series plankton abundance data;

[0016] The cross-correlation function, covariance matrix and quantity change curve of each water layer are spliced ​​together as tensors to form a feature tensor describing the coupling relationship between the corresponding water layer environment and biological factors.

[0017] The feature tensors of all water layers are stacked along the depth dimension to form a comprehensive state matrix of the pond environment.

[0018] As a further aspect of the present invention, the step of calling a pre-constructed knowledge graph of spore-borne diseases to perform pathogen suitability assessment on the comprehensive state matrix, and identifying potential risk periods and spatial coordinates of risk water layers that meet the conditions for spore proliferation, includes:

[0019] Slice the comprehensive state matrix by time dimension and extract all water layer feature tensors corresponding to each time slice;

[0020] The feature tensor of each time slice is input into the matching engine of the spore disease risk knowledge graph and compared item by item with the environmental condition threshold range required for different life stages of spores stored in the knowledge graph.

[0021] When multiple environmental factors in a water layer feature tensor simultaneously fall within the corresponding spore proliferation threshold range, the water layer is determined to be in a risky state during the time slice.

[0022] Record the depth information and time slice start time of all water layers in a risky state to form an initial set of risky water layer time pairs;

[0023] Spatiotemporal clustering analysis is performed on the initial risk water layer time pair set to aggregate risk states that are temporally continuous and spatially adjacent into risk events. The core spatiotemporal coordinates of each risk event are the potential risk period and the spatial coordinates of the risk water layer.

[0024] As a further aspect of the present invention, based on the potential risk period and the spatial coordinates of the risk water layer, and combined with physiological behavior monitoring data of pond aquaculture objects, a multidimensional risk propagation path model is constructed, including:

[0025] Acquire physiological and behavioral monitoring data of pond-cultured organisms, including feeding activity heat map, cluster swimming trajectory map, and respiratory rate change curve;

[0026] The potential risk period and the spatial coordinates of the risk water layer are marked in the three-dimensional pond model and used as the initial distribution source of the pathogen.

[0027] Based on the feeding activity hotspot distribution map, determine the main water layer and horizontal area of ​​the cultured species during the potential risk period;

[0028] Computational fluid dynamics was used to simulate the water flow in the pond during the potential risk period, and to calculate the passive diffusion trajectory and concentration field of the pathogen from the initial pathogen distribution source to the main active water layer and horizontal area.

[0029] By combining the swimming trajectory map of the cluster with the respiratory rate change curve, the risk probability of farmed animals actively coming into contact with pathogens through breathing and feeding behavior is assessed. The risk of passive spread and active contact are coupled and calculated to complete the construction of the multidimensional risk transmission path model.

[0030] As a further aspect of the present invention, based on the output of the multidimensional risk transmission path model, a risk warning level for spore-borne diseases and a corresponding environmental control scheme are generated, including:

[0031] The output of the multidimensional risk transmission path model is analyzed to extract the spatial range, time duration, and proportion of farmed animals affected by the predicted pathogen exposure concentration exceeding the safety threshold;

[0032] Based on the spatial range, time duration, and proportion of affected aquaculture species, a preset risk level mapping table is consulted to determine the corresponding spore disease risk warning level;

[0033] For areas where the predicted exposure concentration of pathogens exceeds the standard, the dominant limiting environmental factors in the dynamic environmental factor dataset for the area are analyzed.

[0034] Based on the type of the dominant limiting environmental factor, basic control measures are matched from a preset control strategy library;

[0035] Based on the specific time period of the potential risk period, the activation time, intensity, and duration of the basic control measures are parameterized to generate a customized environmental control scheme.

[0036] As a further aspect of the present invention, the analysis of the dominant limiting environmental factors in the dynamic environmental factor dataset for areas where the predicted pathogen exposure concentration exceeds the standard includes:

[0037] Obtain fragments of historical dynamic environmental factor data for areas where pathogen predicted exposure concentrations exceed standards during periods of concentration exceedance;

[0038] Calculate the coefficient of variation for each of the following data in the historical dynamic environmental factor data set fragment: water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data;

[0039] The comprehensive impact index of each environmental factor is obtained by multiplying the coefficient of variation of each environmental factor data with its weight coefficient for its influence on the proliferation of sporozoites.

[0040] The environmental factor with the highest comprehensive impact index is selected and identified as the dominant limiting environmental factor.

[0041] As a further aspect of the present invention, it also includes the steps of early warning verification and knowledge graph updating:

[0042] After issuing a risk warning level for spore disease and implementing an environmental control plan, the actual monitoring procedure for pathogens in ponds is initiated, and water samples are collected for quantitative detection of spore cysts or live organisms.

[0043] The spatiotemporal consistency of the actual density data of spores obtained by quantitative detection was compared with the pathogen exposure concentration data predicted by the multidimensional risk transmission path model.

[0044] The accuracy of this warning will be assessed based on the deviation results of the comparative analysis.

[0045] If the warning is accurate, the comprehensive state matrix features that triggered the warning, the corresponding potential risk spatiotemporal coordinates, and the final actual density data will be stored as a positive sample in the training sample library of the spore disease risk knowledge graph.

[0046] The threshold ranges and matching rules in the knowledge graph of sporidiosis risk are retrained and optimized regularly using an updated training sample library.

[0047] As a further aspect of the present invention, the step of performing a spatiotemporal consistency comparison analysis between the quantitatively detected actual density data of sporozoites and the pathogen exposure concentration data predicted by the multidimensional risk transmission path model includes:

[0048] The actual density data of sporozoites obtained from the quantitative detection is mapped to the corresponding spatiotemporal grid of the three-dimensional pond model according to the location of the sampling point and the sampling time.

[0049] From the output of the multidimensional risk transmission path model, extract the pathogen prediction exposure concentration data on the same spatiotemporal grid;

[0050] Calculate the absolute and relative errors between the actual density data and the predicted exposure concentration data for each spatiotemporal grid.

[0051] The average absolute error, average relative error, and proportion of grids whose predicted and measured concentration trends are consistent across all spatiotemporal grids were statistically analyzed.

[0052] Based on the mean absolute error, mean relative error, and the proportion of grid cells with consistent trends, a quantitative evaluation report on the model's predictive performance is generated.

[0053] As a further aspect of the present invention, it also includes a step of fusing the individual susceptibility differences of the cultured subjects:

[0054] Biosensor tags were attached to representative cultured individuals in the pond to collect real-time physiological and behavioral parameters of the representative cultured individuals;

[0055] Establish an individual health status assessment model for aquaculture subjects, and calculate the real-time health score for each representative aquaculture subject based on the real-time physiological and behavioral parameters.

[0056] By combining the real-time health scores of individual aquaculture subjects with their real-time location information in the pond, a spatial distribution map of the susceptibility of individual aquaculture subjects is constructed.

[0057] When calculating risk using the multidimensional risk transmission path model, the spatial distribution map of individual susceptibility of the farmed objects is introduced as a weight field, and the pathogen exposure concentration is adjusted by susceptibility weighting to generate a differentiated risk distribution map for farmed objects with different immunity levels.

[0058] The establishment of an individual health status assessment model for farmed animals, based on the real-time physiological and behavioral parameters, calculates a real-time health score for each representative farmed animal individual, specifically including:

[0059] The baseline physiological parameters and standard behavioral patterns of individual farmed animals are obtained as a reference benchmark.

[0060] The collected real-time physiological parameters are compared with the baseline physiological parameter range in the reference benchmark, and the deviation of each physiological parameter is calculated.

[0061] The collected real-time behavioral parameters are matched with the standard behavioral patterns in the reference benchmark to calculate the behavioral anomaly index.

[0062] The deviations of various physiological parameters and behavioral abnormality indices were standardized and assigned different weights for their health impacts.

[0063] The weighted summation of all standardized deviations and anomaly indices yields a preliminary health score. Then, a preset mapping function is used to convert the preliminary health score into a real-time health score that falls within a standard range.

[0064] As a further aspect of the present invention, the present invention also includes a pond sporozoan disease risk early warning system based on multiple environmental factors. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-described Chuzhou crucian carp sporozoan disease risk early warning method based on multiple environmental factors.

[0065] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0066] By performing spatiotemporal correlation processing on dynamic data collected from multiple environmental sensor nodes, a comprehensive state matrix was generated that characterizes the coordinated change patterns of environmental factors in different water layers. This technique transforms discrete, independent environmental parameters into a model describing the overall dynamic relationships of the system. This allows environmental risk assessment to move beyond assessing whether a single parameter exceeds a certain limit, and instead base it on the specific ecological state formed by the interaction of multiple environmental factors. This enables more accurate identification of the complex environmental conditions truly suitable for sporozoite proliferation, improving the accuracy and foresight of risk identification.

[0067] A pre-constructed knowledge graph of spore-borne diseases is used to evaluate the comprehensive state matrix, and a multi-dimensional risk transmission path model is constructed by combining physiological and behavioral data of farmed organisms. The knowledge graph structures professional knowledge such as pathogen biological characteristics and environmental requirements, enabling the system to understand the semantic relationships between complex conditions and risks. Based on this, a transmission model constructed using host behavioral data is introduced to simulate the dynamic diffusion process of pathogens in the "environment-host" coupled system. This makes the early warning output no longer a simple danger signal, but a dynamic risk picture that includes specific risk periods, spatial coordinates of risk water layers, and potential transmission paths, providing a direct basis for taking stratified, time-based, and spatially targeted precise control and intervention measures. Attached Figure Description

[0068] Figure 1 This is a flowchart of the pond spore disease risk early warning method based on multiple environmental factors described in this invention;

[0069] Figure 2 A flowchart for constructing a multidimensional risk propagation path model. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0071] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0072] See Figure 1Multiple environmental sensor nodes are deployed in the pond to continuously collect various data that constitute a dynamic environmental factor dataset. The dynamic environmental factor dataset is processed for spatiotemporal correlation, integrating and analyzing environmental data from different locations and times to generate a comprehensive state matrix of the pond environment that characterizes the collaborative change patterns of environmental factors at different water layers. A pre-constructed knowledge graph of spore-borne diseases is used, leveraging the knowledge of pathogen growth and environmental relationships stored in the graph, to assess the pathogen suitability of the comprehensive state matrix, thereby identifying potential risk periods and spatial coordinates of risk water layers that meet the conditions for spore proliferation. After identifying potential risks, combined with synchronously acquired physiological behavior monitoring data of pond-cultured organisms, a multidimensional risk transmission path model is constructed based on the potential risk periods and spatial coordinates of risk water layers. This model is used to simulate and extrapolate the diffusion process of pathogens under the combined influence of pond water and the activities of cultured organisms. Finally, based on the output of the multidimensional risk transmission path model, a specific risk warning level for spore-borne diseases is calculated and generated. At the same time, based on the risk causes revealed by the model, an environmental control scheme including oxygenation strategy, water flow disturbance strategy, and feed feeding adjustment scheme is matched and generated.

[0073] In one embodiment of the present invention, the dynamic environmental factor dataset includes water temperature stratification data, dissolved oxygen concentration data, pH data, organic suspended matter concentration data, and phytoplankton abundance time-series data. The water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data from the same water depth are aligned based on precise timestamps to form multiple independent environmental parameter time series sets for each water layer. It can be understood that each water layer generates an independent environmental parameter time series set. For each water layer's environmental parameter time series set, it is necessary to calculate the cross-correlation function between the water temperature and dissolved oxygen concentration time series, and the covariance matrix between the pH and organic suspended matter concentration time series. The calculated cross-correlation function is used to describe the degree of temporal correlation between water temperature changes and dissolved oxygen concentration changes, while the covariance matrix is ​​used to quantify the synergistic change relationship between pH and organic suspended matter concentration. In some embodiments, the cross-correlation function can characterize the correlation strength at a specific time lag. Simultaneously, from the phytoplankton abundance time-series data, a curve showing the change in the quantity of a specific indicative phytoplankton population over time is extracted, referred to as the quantity change curve. The population change curves of specific indicator plankton reflect the response of biological factors to changes in the pond environment.

[0074] In practical implementation, the cross-correlation function and covariance matrix calculated for each water layer, along with the corresponding indicator phytoplankton abundance change curve, are integrated into a feature tensor through tensor concatenation. This feature tensor is a multidimensional data structure describing the coupling relationship between environmental and biological factors in the corresponding water layer. It can be understood that the feature tensor integrates the statistical correlation characteristics of physicochemical parameters and the dynamic change information of biological indicators. The feature tensors of all water layers are stacked according to their corresponding depth dimensions, from surface water to deep water, ultimately forming a comprehensive state matrix of the pond environment. In some embodiments, the stacking operation can construct a three-dimensional data structure, where the three dimensions represent time, water layer depth, and the fused features, respectively.

[0075] In practice, the comprehensive state matrix of the pond environment is sliced ​​along the time dimension, and all water layer feature tensors corresponding to each time slice are extracted. Each time slice can represent a specific moment or a short period of data integration. The feature tensors of each time slice are input into the matching engine of the spore disease risk knowledge graph. The matching engine compares the input feature tensors with the environmental condition threshold ranges required for different life stages of spores stored in the spore disease risk knowledge graph. When multiple environmental factors in a certain water layer feature tensor, such as the statistical characteristics of water temperature and dissolved oxygen, simultaneously fall within the corresponding spore proliferation threshold range stored in the spore disease risk knowledge graph, the water layer is determined to be in a risk state in that time slice. The depth information and the start time of the time slice corresponding to all water layers in a risk state are recorded to form an initial risk water layer time pair set. Each element of the initial risk water layer time pair set records a location identified as having risk at a specific time and depth. Optionally, the initial risk water layer time pair set can be stored in the form of a list or an array.

[0076] In practice, spatiotemporal clustering analysis is performed on the initial risk water layer time pair set. Spatiotemporal clustering analysis can identify risk points that are continuous in time and adjacent in space. In essence, it aggregates discrete risk points into continuous risk events. Specifically, spatiotemporal clustering analysis analyzes the occurrence time and depth coordinates of each point in the initial risk water layer time pair set, merging multiple risk points with time intervals and depth differences less than preset thresholds into a single risk event. The core time period covered by each aggregated risk event is the potential risk period, and the spatial range of water layer depth involved in its core is the risk water layer spatial coordinates. The potential risk period and the risk water layer spatial coordinates are the core parameters characterizing the spatiotemporal location of risk events.

[0077] See Figure 2In one embodiment of the present invention, physiological behavior monitoring data of pond-cultured organisms are acquired. This data includes a feeding activity heatmap, a cluster swimming trajectory map, and a respiratory rate variation curve. The feeding activity heatmap is generated by analyzing the duration and frequency of the organisms' stay in the feeding area. The cluster swimming trajectory map is drawn using acoustic or optical marker tracking technology, and the respiratory rate variation curve is collected by a wearable sensor. Potential risk periods and risk water layer spatial coordinates are marked in a three-dimensional pond model, and these coordinates are used as the initial pathogen distribution source. The initial pathogen distribution source defines the starting spatiotemporal location of the risk simulation. It can be understood that the three-dimensional pond model is a digital representation of the actual pond topography and volume. In a specific implementation, the main active water layer and horizontal region of the cultured organisms during the potential risk period are determined based on the feeding activity heatmap. The main active water layer and horizontal region characterize the concentrated range of the spatial distribution of the cultured organisms during this period. Using computational fluid dynamics methods, based on the pond's three-dimensional geometric model, boundary conditions, and inputs such as wind force and water temperature distribution recorded during the potential risk period, the water flow movement in the pond during the potential risk period is simulated. Computational fluid dynamics simulations can solve the equations of motion for water bodies and predict velocity and flow field distributions. Combining the simulated water flow results, the passive diffusion trajectory and concentration field of pathogens, originating from their initial distribution source, can be calculated under the influence of water flow. The passive diffusion trajectory describes the possible path the pathogen may take with the water flow, while the concentration field describes the spatial distribution and concentration of the pathogen in the water body.

[0078] In practical implementation, the risk probability of farmed organisms actively encountering pathogens is assessed by combining the swarm swimming trajectory map and the respiratory rate change curve. The swarm swimming trajectory map reflects the movement pattern of the farmed organism group, and the respiratory rate change curve is related to the volume of water passing through the gills per unit time. It can be understood that the active swimming behavior of farmed organisms affects their probability of encountering pathogen-carrying water masses, while respiratory behavior is the main route of pathogen invasion through the gills. The calculation of the risk probability of active pathogen contact considers the spatial overlap between the movement path of farmed organisms and the pathogen diffusion path, and also considers the water filtration volume per unit time represented by the respiratory rate. In some embodiments, the risk probability of active pathogen contact can be quantified by analyzing the product of the dwell time of farmed organisms on the pathogen diffusion path and the respiratory intensity. The risk of passive diffusion and active contact are coupled for calculation. The passive diffusion calculation provides the background field of pathogen concentration in the environment, and the active contact calculation superimposes the behavior of farmed organisms onto this background field. The coupled calculation of the risks of passive diffusion and active contact can integrate the physical transport process of pathogens with the biological behavior process of farmed organisms, thereby completing the construction of a multidimensional risk transmission path model. Multidimensional risk transmission path models can output predicted exposure concentrations of pathogens in the three-dimensional spatial and temporal dimensions of ponds, and correlate them with specific activity areas of aquaculture organisms. Optionally, the coupling of passive diffusion and active contact can be achieved through a model that includes diffusion and contact terms, for example:

[0079]

[0080] in: Representing a point in time and space The overall risk value at the location, This represents the pathogen passive diffusion concentration field obtained from computational fluid dynamics simulation. This represents the probability intensity of active contact with pathogens per unit time, derived from the swimming trajectory diagram of the swarm and the curve of changes in respiratory rate. This represents the contact probability intensity caused by feeding behavior per unit time, determined jointly by the feeding activity hotspot distribution map and the swarm swimming trajectory map. and It is a coefficient used to balance the weights of respiratory and ingestion pathways.

[0081] In one embodiment of the present invention, the output of the multidimensional risk propagation path model is analyzed to extract the spatial range, duration, and proportion of affected farmed organisms where the predicted pathogen exposure concentration exceeds the safety threshold. It can be understood that the spatial range where the predicted pathogen exposure concentration exceeds the safety threshold is a three-dimensional spatial region description, the duration is the duration of the exceeding state, and the proportion of affected farmed organisms is calculated based on the time and number of farmed organisms within the exceeding spatial range. Based on the spatial range, duration, and proportion of affected farmed organisms where the predicted pathogen exposure concentration exceeds the safety threshold, a preset risk level mapping table is consulted to determine the corresponding spore-borne disease risk warning level. The preset risk level mapping table is a lookup table that defines the risk level values ​​corresponding to different combinations of spatial range, duration, and proportion of impact. In some embodiments, the risk level mapping table can be divided into multiple levels, such as low, medium, high, and urgent. In specific implementations, for areas where the predicted pathogen exposure concentration exceeds the standard, the dominant limiting environmental factors in the dynamic environmental factor dataset for these areas are analyzed. It can be understood that the dominant limiting environmental factor refers to the single environmental factor that has the most significant impact on sporozoite proliferation and is in an abnormal state during the period and space when the predicted pathogen exposure concentration exceeds the standard. Based on the type of the dominant limiting environmental factor, basic control measures are matched from a pre-set control strategy library. This library stores a list of control operations corresponding to different abnormal states of environmental factors; for example, low dissolved oxygen concentration corresponds to an oxygenation strategy, and excessively high organic suspended matter concentration corresponds to a water flow disturbance strategy. Combined with the specific time of the potential risk period, the activation time, intensity, and duration of the basic control measures are parameterized to generate a customized environmental control scheme. This customized scheme explicitly specifies when to activate the oxygenation equipment, at what power and for how long, or when to turn on the water flow disturbance equipment, how to set the disturbance intensity, how long to run it, and whether the feeding time, location, or quantity needs to be adjusted.

[0082] In specific implementation, a historical dynamic environmental factor data set fragment is obtained for the area where the predicted pathogen exposure concentration exceeds the standard during the period of concentration exceedance. This historical dynamic environmental factor data set fragment contains time-series data of all environmental factors recorded within the area where the predicted pathogen exposure concentration exceeds the standard during the period of concentration exceedance. The coefficients of variation (COPs) for water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data within the historical dynamic environmental factor data set fragment are calculated. The COP is the ratio of the standard deviation to the mean, used to measure the relative dispersion of the data. It can be understood that the COP reflects the degree of fluctuation of each environmental factor during the period of concentration exceedance. The COP of each environmental factor data is multiplied by its weighting coefficient for its influence on sporozoite proliferation to obtain the comprehensive influence index of each environmental factor. The weighting coefficient for the influence on sporozoite proliferation is pre-set based on historical data or expert knowledge and is used to quantify the importance of different environmental factors to sporozoite growth. The environmental factor with the highest comprehensive influence index is selected and identified as the dominant limiting environmental factor. In some embodiments, the formula for calculating the comprehensive influence index can be expressed as:

[0083]

[0084] in: Indicates environmental factors The comprehensive impact index Indicates environmental factors Weighting coefficients for the influence on sporozoan proliferation. This indicates the environmental factors during the period when the predicted exposure concentration of pathogens exceeded the standard. The standard deviation of the data measured in the area exceeding the standard This indicates the environmental factors during the period when the predicted exposure concentration of pathogens exceeded the standard. The average value of the data measured in the area exceeding the standard is obtained. Optionally, the degree to which the factors deviate from the normal range can also be considered when calculating the comprehensive impact index; the specific calculation logic depends on the definition of the weighting coefficients.

[0085] In one embodiment of the present invention, after issuing a risk warning level for spore-borne diseases and implementing an environmental control plan, a pond pathogen monitoring program is initiated to collect water samples for quantitative detection of spore-borne cysts or motile organisms. Quantitative detection of spore-borne cysts or motile organisms is achieved through microscopic counting, molecular biological detection, or flow cytometry, generating actual spore density data for specific sampling locations. The quantitatively obtained actual spore density data is then compared with the pathogen exposure concentration data predicted by a multidimensional risk transmission path model for spatiotemporal consistency. This spatiotemporal consistency comparison analysis requires matching the geographical location and sampling time of the sampling points with the spatial grid and time nodes of the multidimensional risk transmission path model prediction results, comparing the measured values ​​with the predicted values ​​at the same spatiotemporal location. Based on the deviation results of the comparison analysis, the accuracy of the warning is evaluated. The deviation results include the degree of difference between the predicted and measured values ​​in terms of quantity and trend. If the warning is accurate, the comprehensive state matrix features that triggered the warning, the corresponding potential risk spatiotemporal coordinates, and the final actual density data are stored as a positive sample in the training sample library of the spore-borne disease risk knowledge graph. It is understandable that each sample stored in the training sample library of the sporidiosis risk knowledge graph contains environmental data, spatiotemporal risk assessment results, and the final actual pathogen density, forming a complete learning record. The threshold ranges and matching rules in the sporidiosis risk knowledge graph are periodically retrained and optimized using the updated training sample library. The retraining and optimization process can utilize machine learning algorithms to adjust the boundaries of the environmental threshold parameters in the knowledge graph based on newly added samples, or to optimize the weights of feature matching.

[0086] In practice, the actual density data of sporozoites obtained from quantitative detection is mapped to the corresponding spatiotemporal grid of a 3D pond model based on the location and sampling time of the sampling points. The mapping process matches the 3D geographic coordinates and sampling time of the sampling points to pre-divided grid cells and time slices in the 3D pond model. Pathogen predicted exposure concentration data from the same spatiotemporal grid is extracted from the output of the multidimensional risk transmission path model. The absolute and relative errors between the actual density data and the predicted exposure concentration data are calculated for each spatiotemporal grid. The absolute error is the absolute value of the difference between the predicted and actual density data, and the relative error is the ratio of the absolute error to the actual density data. The average absolute error, average relative error, and the proportion of grids where the predicted and measured concentration trends are consistent across all spatiotemporal grids are statistically analyzed. The average absolute error is the average of the absolute errors of all grids, the average relative error is the average of the relative errors of all grids, and the proportion of grids where the predicted and measured concentration trends are consistent refers to the proportion of grids where the predicted and measured values ​​change in the same direction at adjacent time points out of the total number of grids. A quantitative evaluation report of the model's predictive performance is generated based on the mean absolute error, mean relative error, and the proportion of grid cells with consistent trends. The quantitative evaluation report displays the degree of agreement between the model's predictive results and the measured data in numerical and graphical form. In some embodiments, refer to Table 1 for the quantitative evaluation report of the model's predictive performance.

[0087] Table 1: Summary of Model Prediction Performance Evaluation Metrics

[0088] Evaluation indicators Calculate numerical values illustrate Mean Absolute Error (MAE) Calculated values The average level of the absolute difference between predicted and measured values Mean Relative Error (MRE) Calculated percentage The average level of the relative difference between predicted and measured values Trend Consistency Ratio (TCR) Calculated percentage The percentage of grid cells whose predicted trends match the actual trends. Assessment time range Start and end times The time period covered by this assessment Assessment of spatial scope Area Description The pond area covered by this assessment

[0089] It is understood that Table 1 is an optional format for presenting the core indicators of a quantitative assessment report. Optionally, the assessment report may also include a spatial distribution map or a time series comparison chart of the error.

[0090] In one embodiment of the invention, biosensor tags are attached to representative aquaculture subjects in a pond to collect their real-time physiological and behavioral parameters. The biosensor tags can record real-time physiological parameters such as heart rate and gill movement frequency, as well as real-time behavioral parameters such as three-dimensional motion trajectory and acceleration. A health status assessment model for the aquaculture subjects is established, and a real-time health score is calculated for each representative aquaculture subject based on the real-time physiological and behavioral parameters. The health status assessment model is an algorithmic model that maps multi-source sensor data into a comprehensive health score. The real-time health score can be understood as a quantitative value reflecting the current health status of the aquaculture subject. The real-time health score of the aquaculture subject is combined with its real-time location information in the pond to construct a spatial distribution map of the aquaculture subject's susceptibility. The real-time location information is provided by the positioning module built into the biosensor tag. The spatial distribution map of the aquaculture subject's susceptibility reflects the average health or susceptibility level of the aquaculture subject at different spatial locations. When there are a sufficient number of representative individuals, a continuous distribution map can be generated using spatial interpolation methods. When calculating risk using a multidimensional risk propagation path model, the spatial distribution map of individual susceptibility of farmed animals is introduced as a weighting field. Pathogen exposure concentrations are then weighted and adjusted for susceptibility, generating differentiated risk distribution maps for farmed animals with different immunity levels. These differentiated risk distribution maps display risk values ​​adjusted for individual susceptibility; the actual risk in high-susceptibility areas is amplified, while the risk in low-susceptibility areas is reduced. In some embodiments, the weighting operation using the spatial distribution map of individual susceptibility of farmed animals can be performed after the multidimensional risk propagation path model has calculated the baseline pathogen exposure concentration field.

[0091] In practical implementation, an individual health status assessment model for farmed animals is established. Based on real-time physiological and behavioral parameters, a real-time health score is calculated for each representative farmed animal individual. The specific process includes the following steps: 1. Obtaining the baseline physiological parameter range and standard behavioral pattern of the farmed animal individual as a reference benchmark. The baseline physiological parameter range and standard behavioral pattern are established using statistical data from long-term monitoring of healthy farmed animal groups. 2. Comparing the collected real-time physiological parameters with the baseline physiological parameter range in the reference benchmark to calculate the deviation of each physiological parameter. The deviation indicates the degree to which the real-time physiological parameter value deviates from its baseline normal range. 3. Performing a matching degree analysis between the collected real-time behavioral parameters and the standard behavioral pattern in the reference benchmark to calculate the behavioral abnormality index. The behavioral abnormality index indicates the degree of difference between the real-time behavioral pattern and the standard behavioral pattern. 4. Standardizing the deviation of each physiological parameter and the behavioral abnormality index. Standardization converts indicators with different dimensions and ranges to a unified numerical scale. 5. Assigning different health impact weights to each standardized deviation and abnormality index. The health impact weights are pre-set based on the contribution of each physiological parameter and behavioral pattern to the health status. 6. Weighted summing of all standardized deviations and abnormal indices to obtain a preliminary health score. The preliminary health score is a weighted sum of multiple indicators. A pre-defined mapping function converts this preliminary health score into a real-time health score within a standard range. This mapping function can be a linear scaling function or a sigmoid function, its purpose being to normalize the preliminary health score to a fixed, easily interpretable range, such as 0 to 100. The real-time health score can be calculated using the following formula:

[0092]

[0093] in: This indicates a real-time health score. This represents the preset mapping function. Indicates the number of physiological parameters, Indicates the first Standardized deviation of physiological parameters Indicates the first Health impact weights of deviations from physiological parameters Indicates the number of behavioral parameters. Indicates the first Standardized anomaly index of the behavioral parameter, Indicates the first The health impact weight of the abnormality index of the behavioral parameter. Optional, mapping function. The specific format must ensure that the output results fall within the preset standard range. It can be understood that the lower the calculated real-time health score, the worse the health of the individual farmed object, and the higher its susceptibility to sporozoan diseases.

[0094] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A risk early warning method for Chuzhou carp sporozoonosis based on multiple environmental factors, characterized in that, The method includes: Multiple environmental sensor nodes are deployed to continuously collect dynamic environmental factor data sets of the pond. The dynamic environmental factor data sets are then processed for spatiotemporal correlation to generate a comprehensive state matrix of the pond environment. The comprehensive state matrix is ​​used to characterize the coordinated change patterns of environmental factors in different water layers. By calling a pre-constructed knowledge graph of spore-borne diseases, pathogen suitability assessment is performed on the comprehensive state matrix to identify potential risk periods and spatial coordinates of risk water layers that meet the conditions for spore proliferation. Based on the potential risk period and the spatial coordinates of the risk water layer, and combined with the physiological behavior monitoring data of the pond aquaculture objects, a multidimensional risk transmission path model is constructed. The multidimensional risk transmission path model is used to simulate the spread of pathogens in the pond environment. Based on the output of the multidimensional risk transmission path model, a risk warning level for spore-borne diseases and a corresponding environmental control scheme are generated. The environmental control scheme includes oxygenation strategy, water flow disturbance strategy and feed feeding adjustment scheme. The dynamic environmental factor dataset is subjected to spatiotemporal correlation processing to generate a comprehensive state matrix of the pond environment, including: The dynamic environmental factor dataset includes water temperature stratification data, dissolved oxygen concentration data, pH data, organic suspended matter concentration data, and time-series data on phytoplankton abundance. The water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data from the same water depth are time-stamped and aligned to form a time series set of environmental parameters for each of the multiple water layers. For each water layer, the cross-correlation function between water temperature and dissolved oxygen concentration, and the covariance matrix between pH and organic suspended matter concentration are calculated. Extract the population change curves of specific indicative plankton species from the time-series plankton abundance data; The cross-correlation function, covariance matrix and quantity change curve of each water layer are spliced ​​together as tensors to form a feature tensor describing the coupling relationship between the corresponding water layer environment and biological factors. The feature tensors of all water layers are stacked along the depth dimension to form a comprehensive state matrix of the pond environment; Based on the potential risk period and spatial coordinates of the risk water layer, combined with physiological behavior monitoring data of pond aquaculture organisms, a multidimensional risk propagation path model is constructed, including: Acquire physiological and behavioral monitoring data of pond-cultured organisms, including feeding activity heat map, cluster swimming trajectory map, and respiratory rate change curve; The potential risk period and the spatial coordinates of the risk water layer are marked in the three-dimensional pond model and used as the initial distribution source of the pathogen. Based on the feeding activity hotspot distribution map, determine the main water layer and horizontal area of ​​the cultured species during the potential risk period; Computational fluid dynamics was used to simulate the water flow in the pond during the potential risk period, and to calculate the passive diffusion trajectory and concentration field of the pathogen from the initial pathogen distribution source to the main active water layer and horizontal area. By combining the swimming trajectory map of the cluster with the respiratory rate change curve, the risk probability of farmed animals actively coming into contact with pathogens through breathing and feeding behavior is assessed. The risk of passive spread and active contact are coupled and calculated to complete the construction of the multidimensional risk transmission path model.

2. The multi-environment factor-based risk early warning method for Sphaerularia chuzhouensis according to claim 1, characterized in that, The process of calling a pre-constructed knowledge graph of spore-borne diseases to assess the pathogen suitability of the comprehensive state matrix, identifying potential risk periods and spatial coordinates of risk water layers that meet the conditions for spore proliferation, includes: Slice the comprehensive state matrix by time dimension and extract all water layer feature tensors corresponding to each time slice; The feature tensor of each time slice is input into the matching engine of the spore disease risk knowledge graph and compared item by item with the environmental condition threshold range required for different life stages of spores stored in the knowledge graph. When multiple environmental factors in a water layer feature tensor simultaneously fall within the corresponding spore proliferation threshold range, the water layer is determined to be in a risky state during the time slice. Record the depth information and time slice start time of all water layers in a risky state to form an initial set of risky water layer time pairs; Spatiotemporal clustering analysis is performed on the initial risk water layer time pair set to aggregate risk states that are temporally continuous and spatially adjacent into risk events. The core spatiotemporal coordinates of each risk event are the potential risk period and the spatial coordinates of the risk water layer.

3. The multi-environment factor-based risk early warning method for Sphaerotheca chizhouensis according to claim 2, characterized in that, Based on the output of the multidimensional risk transmission path model, a risk warning level for spore-borne diseases and corresponding environmental control measures are generated, including: The output of the multidimensional risk transmission path model is analyzed to extract the spatial range, time duration, and proportion of farmed animals affected by the predicted pathogen exposure concentration exceeding the safety threshold; Based on the spatial range, time duration, and proportion of affected aquaculture species, a preset risk level mapping table is consulted to determine the corresponding spore disease risk warning level; For areas where the predicted exposure concentration of pathogens exceeds the standard, the dominant limiting environmental factors in the dynamic environmental factor dataset for the area are analyzed. Based on the type of the dominant limiting environmental factor, basic control measures are matched from a preset control strategy library; Based on the specific time period of the potential risk period, the activation time, intensity, and duration of the basic control measures are parameterized to generate a customized environmental control scheme.

4. The multi-environment factor-based risk early warning method for Chuzhou Carassius auratus cypriiniasporeosis according to claim 3, characterized in that, For areas where the predicted pathogen exposure concentration exceeds the standard, the dominant limiting environmental factors in the dynamic environmental factor dataset for these areas are analyzed, including: Obtain fragments of historical dynamic environmental factor data for areas where pathogen predicted exposure concentrations exceed standards during periods of concentration exceedance; Calculate the coefficient of variation for each of the following data in the historical dynamic environmental factor data set fragment: water temperature stratification data, dissolved oxygen concentration data, pH data, and organic suspended matter concentration data; The comprehensive impact index of each environmental factor is obtained by multiplying the coefficient of variation of each environmental factor data with its weight coefficient for its influence on the proliferation of sporozoites. The environmental factor with the highest comprehensive impact index is selected and identified as the dominant limiting environmental factor.

5. The method for early warning of crucian carp sporozoan disease based on multiple environmental factors according to claim 4, characterized in that, It also includes steps for early warning verification and knowledge graph updating: After issuing a risk warning level for spore disease and implementing an environmental control plan, the actual monitoring procedure for pathogens in ponds is initiated, and water samples are collected for quantitative detection of spore cysts or live organisms. The spatiotemporal consistency of the actual density data of spores obtained by quantitative detection was compared with the pathogen exposure concentration data predicted by the multidimensional risk transmission path model. The accuracy of this warning will be assessed based on the deviation results of the comparative analysis. If the warning is accurate, the comprehensive state matrix features that triggered the warning, the corresponding potential risk spatiotemporal coordinates, and the final actual density data will be stored as a positive sample in the training sample library of the spore disease risk knowledge graph. The threshold ranges and matching rules in the knowledge graph of sporidiosis risk are retrained and optimized regularly using an updated training sample library.

6. The method for early warning of crucian carp spore disease based on multiple environmental factors according to claim 5, characterized in that, The step of comparing the actual density data of sporozoites obtained by quantitative detection with the pathogen exposure concentration data predicted by the multidimensional risk transmission path model in terms of spatiotemporal consistency includes: The actual density data of sporozoites obtained from the quantitative detection is mapped to the corresponding spatiotemporal grid of the three-dimensional pond model according to the location of the sampling point and the sampling time. From the output of the multidimensional risk transmission path model, extract the pathogen prediction exposure concentration data on the same spatiotemporal grid; Calculate the absolute and relative errors between the actual density data and the predicted exposure concentration data for each spatiotemporal grid. The average absolute error, average relative error, and proportion of grids whose predicted and measured concentration trends are consistent across all spatiotemporal grids were statistically analyzed. Based on the mean absolute error, mean relative error, and the proportion of grid cells with consistent trends, a quantitative evaluation report on the model's predictive performance is generated.

7. The method for early warning of crucian carp spore disease based on multiple environmental factors according to claim 6, characterized in that, It also includes steps to incorporate individual susceptibility differences among farmed animals into the risk calculation of a multidimensional risk propagation path model: Biosensor tags were attached to representative cultured individuals in the pond to collect real-time physiological and behavioral parameters of the representative cultured individuals; Establish an individual health status assessment model for aquaculture subjects, and calculate the real-time health score for each representative aquaculture subject based on the real-time physiological and behavioral parameters. By combining the real-time health scores of individual aquaculture subjects with their real-time location information in the pond, a spatial distribution map of the susceptibility of individual aquaculture subjects is constructed. When calculating risk using the multidimensional risk transmission path model, the spatial distribution map of individual susceptibility of the farmed objects is introduced as a weight field, and the pathogen exposure concentration is adjusted by susceptibility weighting to generate a differentiated risk distribution map for farmed objects with different immunity levels. The establishment of an individual health status assessment model for farmed animals, based on the real-time physiological and behavioral parameters, calculates a real-time health score for each representative farmed animal individual, specifically including: The baseline physiological parameters and standard behavioral patterns of individual farmed animals are obtained as a reference benchmark. The collected real-time physiological parameters are compared with the baseline physiological parameter range in the reference benchmark, and the deviation of each physiological parameter is calculated. The collected real-time behavioral parameters are matched with the standard behavioral patterns in the reference benchmark to calculate the behavioral anomaly index. The deviations of various physiological parameters and behavioral abnormality indices were standardized and assigned different weights for their health impacts. The weighted summation of all standardized deviations and anomaly indices yields a preliminary health score. Then, a preset mapping function is used to convert the preliminary health score into a real-time health score that falls within a standard range.

8. A risk early warning system for crucian carp spore disease based on multiple environmental factors in Chuzhou, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for risk warning of crucian carp spore disease based on multiple environmental factors as described in any one of claims 1 to 7.

Images