A cystic echinococcosis spatio-temporal risk identification and aggregation detection method

Abstract

The application discloses a cystic echinococcosis spatiotemporal risk identification and aggregation detection method, and belongs to the field of public health and spatial epidemiology. The method first collects and pre-processes the incidence data, geographic data and multi-dimensional environmental factor data at the township scale; secondly, the overall aggregation characteristics are judged through global spatial autocorrelation, the high-incidence townships are located through local hotspot analysis, the spatiotemporal coupling aggregation area is identified through spatiotemporal scanning, the diffusion evolution trend is analyzed through standard deviation ellipse analysis, and a multi-dimensional aggregation detection system is constructed; finally, based on the maximum entropy niche model coupled with environmental factors, the global potential infection risk is predicted and classified. The application realizes the whole-process analysis of cystic echinococcosis from present aggregation identification to potential risk prediction, the method is standardized and reproducible, the output result is high in precision, and can provide scientific and technical support for the grass-roots disease control departments to formulate precise prevention and control strategies.

Description

Technical Field

[0001] This invention relates to the fields of public health and spatial epidemiology, and in particular to a method for spatiotemporal risk identification and cluster detection of cysticercosis. Background Technology

[0002] Cystic echinococcosis, also known as cystic echinococcosis, is a zoonotic parasitic disease caused by infection with the larvae of *Echinococcus granulosus*. It is mainly prevalent in the agricultural and pastoral areas of western my country, characterized by a long incubation period, high severity, and difficulty in prevention and control, placing a heavy burden on the health and socio-economic development of residents in endemic areas. The spread of this disease is closely related to various factors such as the natural environment, livestock farming, and socio-economic conditions, and its distribution exhibits significant spatial heterogeneity and spatiotemporal aggregation characteristics.

[0003] Current spatial analysis of cystic echinococcosis largely remains at the level of epidemiological description using single methods. It typically employs only a single spatial autocorrelation or spatiotemporal scanning method to analyze the distribution characteristics of existing cases, lacking a systematic and multi-dimensional cluster detection system. Simultaneously, existing risk prediction methods mostly focus on clinical diagnosis or large-scale prevalence estimation, lacking an integrated technical solution that combines spatiotemporal cluster detection with potential ecological risk prediction at the township level. This makes it difficult to accurately support grassroots disease control departments in carrying out tiered prevention and control and targeted measures. Therefore, there is an urgent need to construct a comprehensive methodology covering global cluster assessment, local hotspot identification, spatiotemporal cluster detection, diffusion trend analysis, and potential risk prediction. This methodology would address the problems of single-dimensional cluster detection and the disconnect between risk prediction and current status detection in existing technologies, enabling refined and quantitative identification of the spatiotemporal risk of cystic echinococcosis. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for spatiotemporal risk identification and aggregation detection of cystic echinococcosis, construct a multi-scale, all-dimensional spatiotemporal aggregation detection system, and combine it with an ecological niche model to achieve quantitative prediction and hierarchical zoning of potential risks, thus providing technical support for the precise prevention and control of cystic echinococcosis.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for spatiotemporal risk identification and clustering detection of cystic echinococcosis includes the following steps: Step 1: Data Collection and Preprocessing: Collect hydatid disease incidence monitoring data, basic geospatial data, and multi-dimensional environmental impact factor data within the study area, perform spatial matching and standardization preprocessing, and obtain township-scale analysis datasets; Step 2: Global Spatial Clustering Analysis: Based on the analysis dataset, the global Moran index is used to perform global spatial autocorrelation analysis on the incidence of cystic echinococcosis to determine the overall spatial distribution pattern of cystic echinococcosis within the study area. Step 3: Local hotspot detection: using local... The index statistics are used to conduct local spatial hotspot analysis, calculate the standardized statistics and significance level of each township unit, and identify high-incidence hotspot areas and low-incidence coldspot areas with statistical significance. Step 4: Spatiotemporal clustering detection: Using spatiotemporal scanning statistical methods, a scanning window is constructed through a Poisson distribution model to traverse all spatial locations and time intervals within the study area, identifying spatiotemporal clustering areas of cysticercosis with statistical significance, and obtaining the spatial range, clustering time period, and relative risk value of the clustering area; Step 5: Spatiotemporal diffusion direction analysis: Using the standard deviation ellipse method, the ellipse parameters of the incidence and distribution of cysticercosis at different time stages are calculated, including the ellipse center, major axis, minor axis, azimuth angle and flattening, to analyze the spatial diffusion direction, distribution range and evolution trend of the epidemic. Step Six: Potential Risk Prediction and Zoning: Based on the maximum entropy niche model, using the incidence points of cysticercosis as sample data and multi-dimensional environmental impact factors as predictive variables, a risk prediction model is trained and constructed. The potential infection risk probability of cysticercosis in the study area is obtained through the model output, and the study area is divided into different levels of risk zones by combining the natural breakpoint method.

[0006] Furthermore, the cystic echinococcosis incidence monitoring data mentioned in step one includes the annual number of cases, corresponding population, and incidence rate data for each township; the basic geospatial data includes township-level administrative boundary vector data for the study area; and the multi-dimensional environmental impact factor data includes one or more of the following: bioclimatic factors, land cover type factors, vegetation cover factors, socioeconomic factors, and livestock farming factors.

[0007] Furthermore, the preprocessing in step one includes: unifying all spatial data to the same coordinate system and spatial resolution; performing multicollinearity tests on environmental factors and eliminating factors with variance inflation factors greater than a set threshold; and performing outlier verification and spatial matching on incidence rate data to ensure that the incidence rate data corresponds one-to-one with administrative boundary units.

[0008] Furthermore, in the process of performing global spatial autocorrelation analysis on the incidence of cystic echinococcosis using the global Moran index described in step two, the global Moran index is calculated as follows: the total number of township units is divided by the sum of all spatial weights, then multiplied by the sum of the products of the spatial weights of all township pairs and their respective incidence rates from the mean, and then divided by the sum of the squares of the incidence rates from the mean of all townships. The statistical significance of the global Moran's index is determined by standardized tests. When the standardized statistic is greater than 1.96 and the significance level is less than 0.05, it is determined that there is a significant spatial positive correlation clustering in the incidence data.

[0009] Furthermore, the use of localization described in step three... In the process of using index statistics to conduct local spatial hotspot analysis, the calculation method of the local Gitz index is as follows: the weighted sum of the incidence rates of all related towns surrounding the target town minus the product of the mean incidence rate and the weighted sum, and then divided by the ratio of the standard deviation of the incidence rate to the standardized weighted term. Hotspots are classified according to the standardized statistics of the local Gitzo index: a standardized statistic greater than 2.58 and a significance level less than 0.01 is a first-level hotspot area, a standardized statistic greater than 1.96 and a significance level less than 0.05 is a second-level hotspot area, and a standardized statistic less than -1.96 and a significance level less than 0.05 is a coldspot area.

[0010] Furthermore, in step four, the spatiotemporal scanning statistical method is used to construct a scanning window through a Poisson distribution model. During the process of traversing all spatial locations and time intervals within the study area, the spatiotemporal scanning statistics adopt a cylindrical scanning window. The spatial radius and time length of the window are both variable parameters. The maximum spatial radius is set as a set proportion of the total population of the study area, and the maximum time length is set as a set proportion of the total study duration. The difference in incidence rates inside and outside the scanning window was measured by log-likelihood ratio, and the relative risk value was calculated. The statistical significance of the clustering area was tested using Monte Carlo simulation.

[0011] Furthermore, in step five, the standard deviation ellipse method is used to calculate the ellipse parameters of the incidence distribution of cystic echinococcosis at different time stages. In this process, the research period is divided into multiple sub-stages according to a set time interval, and the standard deviation ellipse parameters of each sub-stage are calculated separately. The major axis of the ellipse represents the dominant direction of the epidemic distribution, while the minor axis represents the degree of dispersion of the distribution. The greater the flattening, the stronger the directionality of the distribution. By comparing the offset trajectories of the ellipse center at different stages, the spatial migration trend of the epidemic can be obtained.

[0012] Furthermore, in step six, the risk prediction model based on the maximum entropy niche model, using the cysticercosis disease outbreak points as sample data and multi-dimensional environmental impact factors as predictive variables, is trained and constructed by randomly dividing the outbreak point data and environmental factor data into training set and test set according to a set ratio before the maximum entropy model is trained. The model uses the knife-cut test to examine the contribution of each environmental factor, and verifies the model's prediction accuracy by the area under the receiver operating characteristic curve. A value greater than 0.8 indicates that the model has a good prediction effect.

[0013] Furthermore, in step six, the model performance is optimized by adjusting the feature combination and regularization coefficient of the model; finally, a continuous risk probability layer is output, and the risk level is divided into five levels: low risk, lower risk, medium risk, higher risk and high risk using the natural breakpoint method.

[0014] Beneficial effects: Compared with the prior art, the present invention has the following advantages: First, a four-in-one clustering detection system was constructed, encompassing global to local and spatial to spatiotemporal dimensions. By using global autocorrelation to determine overall clustering characteristics, identifying local hotspots to accurately locate high-incidence townships, using spatiotemporal scanning to detect spatiotemporally coupled clustering areas, and using standard deviation ellipses to reveal diffusion and evolution trends, this system can comprehensively and systematically reveal the spatiotemporal distribution patterns of cystic echinococcosis, thus overcoming the limitations of single analysis methods.

[0015] Second, combining spatiotemporal clustering detection with maximum entropy niche risk prediction can not only identify existing clustering areas, but also quantitatively predict potential infection risks based on environmental factors, thus extending from "current status description" to "risk prediction" and providing a basis for moving the prevention and control checkpoint forward.

[0016] Third, the method is adapted to the fine scale of townships, the process is standardized and reproducible, and the output of hotspot areas, spatiotemporal clusters and risk zoning results can directly support grassroots disease control departments in delineating key prevention and control areas and formulating differentiated prevention and control strategies, which has strong practical value. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process of the method of the present invention; Figure 2 This is a schematic diagram showing the local hotspot distribution of cystic echinococcosis. Figure 3 This is a schematic diagram showing the spatiotemporal aggregation area of ​​cystic echinococcosis. Figure 4 A schematic diagram illustrating the evolution of the standard deviation ellipse of the distribution of cystic echinococcosis at different stages; Figure 5 A schematic diagram showing the potential risk zoning results for cystic echinococcosis; Figure 6 Predicting the receiver operating characteristic curve of the maximum entropy model for the CE distribution; Figure 7 Output the simulation results for Group A's prevention and control strategy; Figure 8 Output the simulation results for Group B's prevention and control strategy; Figure 9 Output the simulation results for the prevention and control strategy of Group C; Figure 10 Output the simulation results for the prevention and control strategy of Group D; Figure 11 The simulation output results for the prevention and control strategy of Group E. Detailed Implementation

[0018] The present invention will be further described below with reference to specific embodiments.

[0019] This embodiment uses a county-level research area in a province in western my country where cystic echinococcosis is prevalent as an example to illustrate the method of the present invention in detail. The specific implementation steps are as follows: Step 1: Data Acquisition and Preprocessing; We collected surveillance data on cystic echinococcosis in the study area from 2015 to 2024, covering the annual number of confirmed cases and the year-end resident population of all townships within the jurisdiction, and calculated the annual incidence rate for each township. We also collected the latest township-level administrative boundary vector data for the study area as the basic unit for spatial analysis.

[0020] Collect multi-dimensional environmental impact factor data, including: bioclimatic factors, such as annual average temperature, annual precipitation, seasonality of temperature, seasonality of precipitation, etc.; land cover factors, such as grassland proportion, cultivated land proportion, desert proportion, etc.; vegetation cover factors, such as normalized difference vegetation index annual average; socio-economic factors, such as population density, nighttime light index, per capita GDP, etc.; and livestock breeding factors, such as cattle and sheep stock density.

[0021] All data were preprocessed: all spatial data were uniformly projected to the same geographic coordinate system and resampled to the same spatial resolution; multicollinearity tests were performed on environmental factors, and factors with variance inflation factors greater than 10 were removed to reduce linear interference between factors; the incidence data were spatially correlated and matched with township boundaries, outliers and missing values ​​were checked and corrected, and finally an analysis dataset containing 12 environmental factors matched to all township units was formed.

[0022] Step 2: Global Spatial Clustering Analysis Based on the annual township incidence rate data, an adjacency-based spatial weight matrix was constructed, and global Moran's index was used for global spatial autocorrelation analysis. The calculated global Moran's index values ​​for each year during the study period were all positive, with standardized statistics all greater than 1.96 and significance levels all less than 0.05. This indicates that the overall incidence rate of cystic echinococcosis in the study area exhibits a significant spatial positive correlation; that is, high-incidence townships are adjacent to high-incidence townships, and low-incidence townships are adjacent to low-incidence townships, demonstrating a clear spatial clustering of the disease distribution rather than a random distribution.

[0023] Step 3: Local hotspot detection; Local hotspot analysis was conducted using the Local Gitzo Index (LGI), calculating the statistical value and significance level for each township. Based on the calculation results, townships were categorized as follows: townships with a standardized statistic greater than 2.58 and a significance level less than 0.01 were designated as Level 1 hotspots, totaling 8 townships, mainly concentrated in the pastoral areas of the northwest of the study area; townships with a standardized statistic greater than 1.96 and a significance level less than 0.05 were designated as Level 2 hotspots, totaling 12 townships, distributed around the Level 1 hotspots; townships with a standardized statistic less than -1.96 and a significance level less than 0.05 were designated as cold spots, totaling 15 townships, concentrated in the agricultural areas of the southeast of the study area; the remaining townships showed no significant clustering characteristics. This local hotspot analysis accurately identified the core townships with high-incidence clusters, providing targeted objectives for key prevention and control measures.

[0024] Step 4: Spatiotemporal clustering scanning detection; Spatiotemporal scanning statistical tools were used, with the maximum spatial radius of the scanning window set to 15% of the total population of the study area and the maximum time length set to half of the total study duration. Spatiotemporal scanning was conducted based on a Poisson model. Significance tests were performed through 999 Monte Carlo simulations, ultimately identifying one primary spatiotemporal cluster and two secondary spatiotemporal clusters. The primary cluster is located in the northwest of the study area, covering six townships, with a clustering period from 2015 to 2020, and a relative risk value of 6.2, indicating that the incidence risk within this area is 6.2 times higher than outside the area. The two secondary clusters are located in the central and southwest regions, respectively, with different clustering periods and relative risk values ​​of 2.8 and 2.3, respectively. These results clearly demonstrate the coupled clustering characteristics of cysticercosis in both temporal and spatial dimensions.

[0025] Step 5: Analysis of the spatiotemporal diffusion direction; The 10-year study period was divided into two phases: the first five years and the last five years. The standard deviation ellipses of the disease distribution in each phase were calculated. The results showed that in the first phase, the ellipse center was located in the central-northwest part of the study area, with a major axis oriented northwest-southeast and a flattening of 0.42. In the second phase, the ellipse center shifted southeast by approximately 12 kilometers, the azimuth of the major axis decreased slightly, and the flattening increased to 0.51. Analysis indicated that the distribution of cysticercosis during the study period generally showed a southeastward trend, and the directional characteristics of the distribution further strengthened. The dominant direction of the epidemic spread was basically consistent with the main traffic arteries and livestock movement directions within the county.

[0026] Step Six: Potential Risk Prediction and Zoning; Spatial coordinates of all cases were extracted from the incidence data as existing samples. Seventy percent of these were randomly selected as the training set, and thirty percent as the test set. The preprocessed 12 environmental factors were input into the maximum entropy model, with a regularization coefficient of 1. The model was trained using a combination of linear and quadratic features.

[0027] Model validation results show that the area under the curve (AUC) on the training set is 0.89 and the AUC on the test set is 0.85, indicating good model prediction accuracy. The knife-cut test reveals that annual average precipitation, grassland coverage, and cattle / sheep stock density are the top three dominant factors influencing the distribution of cysticercosis.

[0028] The trained model outputs a risk probability raster layer covering the entire study area. Using the natural breakpoint method, risk values ​​are categorized into five levels from low to high: low risk, relatively low risk, medium risk, relatively high risk, and high risk. Results show that high-risk areas are mainly distributed in the northwestern pastoral areas, closely coinciding with identified hotspots and spatiotemporal clusters. Additionally, three areas with high potential risk but no obvious high incidence were identified, providing early warning for proactive prevention and intervention. Example

[0029] This embodiment is based on the prevalence data of cystic echinococcosis in a county in Xinjiang from 2013 to 2024, and fully implements the spatiotemporal risk analysis and control effect evaluation method. The specific process is as follows: I. Data Sources and Preprocessing 1. Case data: All confirmed cases of cystic echinococcosis in the county's government-designated hospitals from 2013 to 2024 were collected. After cleaning and removing outliers and missing records, a total of 210 valid case data were obtained, including the onset time, latitude and longitude location and case number.

[0030] 2. Spatial purification process: Based on the environmental factor layer with a resolution of 500m×500m, redundant case points with spatial autocorrelation were removed, resulting in 107 spatially independent case distribution records, which were used as input data for subsequent modeling.

[0031] 3. Environmental variable data: Initially, 34 environmental variables were collected from three categories: bioclimate, geographical landscape, and socioeconomic. After correlation analysis and collinearity diagnosis of variance inflation factor, 8 variables without significant collinearity were finally selected for inclusion in the model. These variables are: average temperature in the driest season, maximum temperature in the warmest month, land cover type, normalized difference vegetation index, effective soil water content, soil organic matter content, nighttime light data, and population density. All variables were uniformly resampled to a resolution of 500 meters and masked to match the study area.

[0032] 4. Population and economic data: This includes population size parameters for the three main hosts (dogs, sheep, and humans) in the county, as well as economic parameters such as the supply, price, and supply-demand elasticity of mutton. All data are derived from local statistical bulletins and field surveys.

[0033] II. Spatiotemporal Epidemiological Characteristics of Cystic Hydatid Disease (1) Global spatial autocorrelation analysis Moran's index was used to measure the overall spatial distribution pattern of cystic echinococcosis incidence at the county-wide scale. The calculation logic was as follows: the difference between the number of cases in each township and the county average was used, combined with the spatial weight matrix between townships, to calculate the overall correlation. Statistical significance was then tested using Z-scores and p-values.

[0034] Simulation results show that the global Moran index for all years from 2013 to 2024 is negative and close to zero, the corresponding absolute Z score is less than 1.96, and the P value is greater than 0.05. This indicates that at the county-wide scale, the spatial distribution of cystic echinococcosis incidence does not show significant clustering or discrete characteristics, but rather exhibits a random distribution pattern.

[0035] (2) Local hotspot analysis Use local The statistical method identifies local high-incidence clusters. The calculation logic is as follows: for each township, the weighted sum of the number of cases in itself and its neighboring areas is calculated and compared with the expected value under random distribution to obtain the corresponding Z-score and confidence level, and hotspot areas with three confidence levels of 90%, 95%, and 99% are divided.

[0036] The results showed that 12 hotspot townships were identified during the study period, and no statistically significant cold spots were found. Among them, Xidi Town, Northwest Bay Township, and Banjiegou Town were repeatedly identified as high-confidence hotspots in multiple years, belonging to stable core transmission areas; the hotspot areas showed a dynamic evolution characteristic of "initial concentration - mid-term dispersion - later intensity increase - recent re-spread" over time.

[0037] A directional analysis was performed on the collected data from 2013 to 2024, and the results are as follows: Figure 2 As shown, the detailed attribute values ​​of the standard deviation ellipse are shown in Table 1.

[0038] Table 1. Relevant parameters for directional distribution analysis

[0039] Overall, the elliptic flattening of CE (Cerebrovascular Encephalopathy) outbreak sites in the study area from 2013 to 2024 was 0.155, with an azimuth of 95.14° (nearly east-west), indicating a slight east-west extension trend in the spatial spread of the epidemic throughout the study period. Figure 2 As shown.

[0040] (3) Directional distribution analysis The standard deviation ellipse method was used to analyze the directional evolution of the epidemic spread. The study period was divided into four stages, and the ellipse parameters for each stage were calculated: Phase 1 (2013-2014): Elliptical azimuth angle 37.4 degrees, flattening 0.288, distributed in a northeast-southwest direction, the epidemic was in its initial spread stage.

[0041] The second phase, from 2015 to 2017, saw the ellipse's azimuth angle at 127.24 degrees and its flattening at 0.181. It shifted to a northwest-southeast orientation, with the ellipse's center moving south, corresponding to the formation of the southern agglomeration area.

[0042] The third stage, 2018-2020: the ellipse azimuth angle was 92.25 degrees, the flattening was 0.351, it was close to east-west, and the semi-major axis reached its maximum value, corresponding to the stage of the epidemic spreading throughout the region.

[0043] Phase 4 (2021-2024): The elliptic flattening decreases to 0.052, the shape becomes close to a circle, the epidemic loses its dominant direction of spread and turns into a multi-point sporadic state.

[0044] Based on the epidemiological trend of CE (cerebrovascular accident), the collected CE case data from 2013 to 2024 were divided into four consecutive phases according to year, and directional distribution analysis was performed on each phase. The results are shown below. Figure 4 The attribute values ​​of the standard deviation ellipse for each time period are shown in Table 2.

[0045] Table 2. Relevant parameters of the staged standard deviation ellipse.

[0046] Analysis shows that the spatial distribution of the CE epidemic underwent a significant directional evolution during the study period.

[0047] (4) Spatiotemporal clustering analysis A retrospective spatiotemporal scan statistical Poisson model was used, with a maximum spatial scan radius of 20 kilometers and a maximum time window of 50% of the study duration. The significance was tested using 999 Monte Carlo simulations.

[0048] The results identified four statistically significant spatiotemporal clusters, such as... Figure 3 As shown, the detailed values ​​of each spatiotemporal cluster region obtained from the scan are shown in Table 3: Level 1 cluster: The center is located in Banjiegou Town, with a radius of 19.78 kilometers. The cluster period is from 2015 to 2019. It contains 39 cases, with a relative risk of 5.7, and covers 5 townships.

[0049] Secondary cluster: The center is located in Xidi Town, with a radius of 9.45 kilometers. The cluster period is from 2017 to 2021. It contains 14 cases, with a relative risk of 4.97, and covers 2 townships.

[0050] Level 3 cluster: The center is located in Tatar Township, with a radius of 15.22 kilometers. The clustering period is from 2018 to 2019. It contains 8 cases, with a relative risk of 9.54, and covers 2 townships.

[0051] Level 4 cluster area: The center is located in Northwest Bay Township, the cluster period is from 2016 to 2019, it contains 17 cases, and the relative risk is 3.61.

[0052] All clusters are located in the agro-pastoral ecotone in the southern part of the county.

[0053] Table 3. Detailed information on each spatiotemporal cluster region.

[0054] III. Disease Occurrence Risk Prediction Based on Maximum Entropy Model A maximum entropy model was used to construct an environmental suitability prediction model for cysticercosis. 25% of the case data was set as the test set, and 10 cross-validations were performed. The accuracy of the model was evaluated by the area under the receiver operating characteristic curve, and the importance of variables was tested by the knife cut method.

[0055] Model accuracy: The average area under the curve after 10 cross-validations is 0.963, indicating that the model's prediction accuracy is excellent and its generalization ability is good.

[0056] Key driving factors: Ranked from highest to lowest contribution, land cover type contributed 31.3%, normalized difference vegetation index contributed 30.5%, and nighttime light data contributed 22.7%, with the three contributing a total of 84.5%, making them the core driving factors influencing disease distribution; the contributions of the remaining variables were all less than 10%.

[0057] Risk zone classification: Suitability probabilities were divided into four levels based on thresholds: a suitability probability less than 0.2 was considered a safe zone, 0.2 to 0.4 a low-risk zone, 0.4 to 0.6 a medium-risk zone, and greater than 0.6 a high-risk zone. The results showed that the total area of ​​high-risk zones in the county was approximately 3021 square kilometers, accounting for 16% of the county's total area, concentrated in the agro-pastoral ecotone in the southern part of the county.

[0058] IV. Construction of Multi-Host Transmission Dynamics Model and Prevention and Control Assessment (1) Model compartment structure Construct a propagation dynamics model for four groups of compartments: dog-sheep-human-environment. The canine population is divided into two compartments: susceptible dogs and infected dogs. Susceptible dogs become infected dogs after ingesting the internal organs of infected sheep containing parasite cysts. Infected dogs continuously release parasite eggs into the environment.

[0059] The sheep population is divided into two enclosures: one for susceptible sheep and one for infected sheep. Susceptible sheep ingest parasite eggs from the environment and then become infected. The internal organs of infected sheep can further infect dogs.

[0060] The population is divided into three compartments: susceptible individuals, latent individuals, and infected individuals. Susceptible individuals enter the latent period after being exposed to parasite eggs in the environment. After the latent period, they become clinically infected individuals. Infected individuals can recover through treatment.

[0061] The environment is set up with a separate egg chamber, where the eggs come from the excrement of infected dogs and also die naturally.

[0062] The dynamic logic of the model is described by a set of ordinary differential equations. The rate of change of the number of each compartment is determined by parameters such as the birth rate, death rate, infection conversion rate, and intervention effectiveness.

[0063] (2) Basic parameter settings All parameters are calibrated based on local survey data and literature. The core parameters are as follows: Canine parameters: The average number of newly added dogs per unit time is 2,400, the natural mortality rate of dogs per year is 0.08, the infection coefficient of infected sheep to dogs is 2.9 per ten million, and the annual egg release intensity of each infected dog is 9,799.

[0064] Sheep flock parameters: average number of new sheep per unit time 294,900, annual comprehensive mortality rate of sheep 0.62, and infection coefficient of environmental parasite eggs in sheep 7.4 per billion.

[0065] Population parameters: average annual increase of 1395 people, annual natural mortality rate of 0.00429, annual mortality rate of echinococcosis of 0.00194, incubation period of about 14 years, annual cure rate of 0.75, and infection coefficient of environmental eggs to humans of 4.2 parts per trillion.

[0066] Environmental parameters: The annual mortality rate of insect eggs in the environment is 10.42%.

[0067] (3) Basic flow steady-state simulation The model was run long-term to a dynamic equilibrium state, yielding the following cell sizes under endemic disease steady-state conditions: 172,078 susceptible sheep, 303,567 infected sheep; 14,282 susceptible dogs, 15,717 infected dogs; 319,297 susceptible individuals, 2,616 latent individuals, and 247 infected individuals; and a total of 14,780,524 parasite eggs in the environment. This steady-state state served as a unified baseline for simulating all intervention measures.

[0068] (4) Simulation of the effects of a single intervention measure All intervention simulations started from a steady state and lasted for 9 years. Each measure was set with three intensity levels: low, medium, and high. Regular deworming of dogs: Under low intensity, the overall effective deworming rate was 0.3 and the coverage rate was 0.5; under medium intensity, the overall effective deworming rate was 0.6 and the coverage rate was 0.7; under high intensity, the overall effective deworming rate was 0.9 and the coverage rate was 0.9. Simulation results show that canine deworming is the most effective single measure to break the transmission chain. After 9 years of high-intensity intervention, the number of infected dogs decreased to 1.6, the number of infected people decreased to 141, and the number of parasite eggs in the environment decreased to 1588. Moreover, the higher the intervention intensity, the faster the effect.

[0069] Sheep immunization: At low intensity, the vaccination rate was 0.3% and the immunization failure rate was 0.1%; at medium intensity, the vaccination rate was 0.6% and the immunization failure rate was 0.15%; at high intensity, the vaccination rate was 0.9% and the immunization failure rate was 0.2%. Simulation results showed that sheep immunization had a clear protective effect on the sheep themselves, but its indirect blocking effect on canine infection and environmental pollution was weak. Nine years after high-intensity intervention, more than 11,500 dogs were still infected, and the number of infected individuals remained above 244.

[0070] Human health education: The impact coefficient of health education was 0.2 at low intensity, 0.4 at medium intensity, and 0.6 at high intensity. Simulation results show that health education only affects the human infection process. After 9 years of high-intensity intervention, the number of infected individuals in the population decreased to 182, but it had no impact on animal hosts and environmental transmission.

[0071] (5) Simulation of combined prevention and control strategies Five representative integrated prevention and control strategies were designed, and a simulation period of 9 years was conducted to compare the prevention and control effects and cost-effectiveness. Strategy A, a conventional equilibrium strategy: all three measures are taken at a moderate intensity, with a comprehensive resource input index of 0.47. At the end of the simulation, 2.7 dogs, 1421 sheep, 140.9 people in the human population became infected, and 2731.8 environmental eggs were found. The output results are shown below. Figure 7 .

[0072] Strategy B, a strengthened canine control strategy: high-intensity canine deworming combined with low-intensity sheep immunization and health education, with a comprehensive resource input index of 0.44. At the end of the simulation, 1.3 dogs, 1340.2 sheep, 140.4 people in the human population became infected, and 1266.9 parasite eggs were found in the environment. The cost-effectiveness score was 2.10, the highest among all strategies. See the output results below. Figure 8 .

[0073] Strategy C, Strengthening Sheep Immunization: High-intensity sheep immunization combined with low-intensity canine deworming and health education, with a comprehensive resource input index of 0.42. At the end of the simulation, this strategy resulted in the highest infection indicators among the five groups, with a cost-effectiveness score of 2.15, but it was slow to take effect and offered weak protection to the general population. See the output results below. Figure 9 .

[0074] Strategy D, a strengthened population education strategy: Moderate-intensity canine deworming and sheep immunization combined with high-intensity health education, resulting in a comprehensive resource input index of 0.54. Except for a slightly lower incidence of disease in the population compared to Strategy A, all other indicators were identical to Strategy A, indicating that the systemic contribution of solely strengthening education was limited. See the output results below. Figure 10 .

[0075] Strategy E, Comprehensive Enhancement Strategy: All three measures are upgraded to high intensity, with a comprehensive resource input index of 0.77. At the end of the simulation, 1.2 dogs, 1267 sheep, 139.6 people in the human population became infected, and 1198.7 environmental eggs were detected. The overall effect was optimal, but the cost was the highest, resulting in the lowest cost-effectiveness score. See the output results below. Figure 11 .

[0076] Based on a comprehensive evaluation across multiple dimensions, Strategy B, which strengthens dog control, demonstrates the best overall performance under limited resource conditions, and is characterized by its rapid effectiveness and high cost-effectiveness.

[0077] V. Coupled Economic Impact Assessment Based on the theory of partial equilibrium, an economic transmission model of "sheep flock infection - supply loss - price fluctuation" is constructed. The core parameters are set as follows: the proportion of fattened sheep that can be marketed locally is 0.76, the commercialization rate of sheep is 0.5, the live weight of each sheep at slaughter is 43 kg, the benchmark price of mutton is 58 yuan / kg, the market supply in a normal year is 6,164,949 kg, the price elasticity of supply is 0.0559, and the price elasticity of demand is -0.81.

[0078] Baseline scenario without intervention: During the simulated 2-year period, the cumulative supply loss of mutton in the market is 56.021 million kg, the peak price increase is 267.62%, and the peak price is 213.48 yuan / kg; the price at the end of the simulated period is still 27.46% higher than the baseline level, with an ending price of 73.93 yuan / kg.

[0079] Economic Effects of Intervention Scenario: All intervention strategies significantly reduced supply losses and price increases. Among them, the comprehensive reinforcement strategy E recovered the most direct economic losses, accumulating 2.277 billion yuan. The enhanced dog control strategy B achieved a similar economic recovery amount to the conventional equilibrium strategy with lower input, approximately 2.18 billion yuan, demonstrating outstanding cost-effectiveness. Incorporating the economic recovery dimension into the comprehensive evaluation further verifies the balance between health and economic benefits of strategy B under resource constraints.

[0080] The embodiments disclosed herein are preferred embodiments, but are not limited thereto. Those skilled in the art can readily grasp the spirit of the present invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of the present invention, they are all within the protection scope of the present invention.

Claims

1. A method for spatiotemporal risk identification and clustering detection of cystic echinococcosis, characterized in that, Includes the following steps: Step 1: Data Collection and Preprocessing: Collect hydatid disease incidence monitoring data, basic geospatial data, and multi-dimensional environmental impact factor data within the study area, perform spatial matching and standardization preprocessing, and obtain township-scale analysis datasets; Step 2: Global Spatial Clustering Analysis: Based on the analysis dataset, the global Moran index is used to perform global spatial autocorrelation analysis on the incidence of cystic echinococcosis to determine the overall spatial distribution pattern of cystic echinococcosis within the study area. Step 3: Local hotspot detection: using local... The index statistics are used to conduct local spatial hotspot analysis, calculate the standardized statistics and significance level of each township unit, and identify high-incidence hotspot areas and low-incidence coldspot areas with statistical significance. Step 4: Spatiotemporal clustering detection: Using spatiotemporal scanning statistical methods, a scanning window is constructed through a Poisson distribution model to traverse all spatial locations and time intervals within the study area, identifying spatiotemporal clustering areas of cysticercosis with statistical significance, and obtaining the spatial range, clustering time period, and relative risk value of the clustering area; Step 5: Spatiotemporal diffusion direction analysis: Using the standard deviation ellipse method, the ellipse parameters of the incidence and distribution of cysticercosis at different time stages are calculated, including the ellipse center, major axis, minor axis, azimuth angle and flattening, to analyze the spatial diffusion direction, distribution range and evolution trend of the epidemic. Step Six: Potential Risk Prediction and Zoning: Based on the maximum entropy niche model, using the incidence points of cysticercosis as sample data and multi-dimensional environmental impact factors as predictive variables, a risk prediction model is trained and constructed. The potential infection risk probability of cysticercosis in the study area is obtained through the model output, and the study area is divided into different levels of risk zones by combining the natural breakpoint method.

2. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, The cystic echinococcosis incidence monitoring data mentioned in Step 1 includes the annual number of cases, corresponding population, and incidence rate data for each township; the basic geospatial data includes township-level administrative boundary vector data for the study area; and the multi-dimensional environmental impact factor data includes one or more of the following: bioclimatic factors, land cover type factors, vegetation cover factors, socioeconomic factors, and livestock farming factors.

3. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 2, characterized in that, The preprocessing in step one includes: unifying all spatial data to the same coordinate system and spatial resolution; performing multicollinearity tests on environmental factors and removing factors with variance inflation factors greater than a set threshold; and performing outlier verification and spatial matching on incidence rate data to ensure that the incidence rate data corresponds one-to-one with administrative boundary units.

4. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, In the process of performing global spatial autocorrelation analysis on the incidence of cystic echinococcosis using the global Moran index described in step two, the global Moran index is calculated as follows: the total number of township units is divided by the sum of all spatial weights, then multiplied by the sum of the products of the spatial weights of all township pairs and their respective incidence rates from the mean, and then divided by the sum of the squares of the incidence rates from the mean of all townships. The statistical significance of the global Moran's index is determined by standardized tests. When the standardized statistic is greater than 1.96 and the significance level is less than 0.05, it is determined that there is a significant spatial positive correlation clustering in the incidence data.

5. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, In the process of using the local Getis-OrdGi index statistic for local spatial hotspot analysis described in step three, the local Getis-OrdGi index is calculated as follows: the weighted sum of the incidence rates of all related towns surrounding the target town minus the product of the mean incidence rate and the weighted sum, and then divided by the ratio of the standard deviation of the incidence rate to the standardized weighted term. Hotspots are classified according to the standardized statistics of the local Gitzo index: a standardized statistic greater than 2.58 and a significance level less than 0.01 is a first-level hotspot area, a standardized statistic greater than 1.96 and a significance level less than 0.05 is a second-level hotspot area, and a standardized statistic less than -1.96 and a significance level less than 0.05 is a coldspot area.

6. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, Step four describes the use of a spatiotemporal scanning statistical method. By constructing a scanning window using a Poisson distribution model, the method traverses all spatial locations and time intervals within the study area. The spatiotemporal scanning statistics employ a cylindrical scanning window, with both the spatial radius and time length of the window being variable parameters. The maximum spatial radius is set as a predetermined proportion of the total population of the study area, and the maximum time length is set as a predetermined proportion of the total study duration. The difference in incidence rates inside and outside the scanning window was measured by log-likelihood ratio, and the relative risk value was calculated. The statistical significance of the clustering area was tested using Monte Carlo simulation.

7. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, In step five, the standard deviation ellipse method is used to calculate the ellipse parameters of the incidence distribution of cystic echinococcosis at different time stages. This involves dividing the research period into multiple sub-stages according to a set time interval and calculating the standard deviation ellipse parameters for each sub-stage separately. The major axis of the ellipse represents the dominant direction of the epidemic distribution, while the minor axis represents the degree of dispersion of the distribution. The greater the flattening, the stronger the directionality of the distribution. By comparing the offset trajectories of the ellipse center at different stages, the spatial migration trend of the epidemic can be obtained.

8. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 1, characterized in that, In step six, the maximum entropy niche model is used as a sample data source for cysticercosis disease outbreaks and multi-dimensional environmental factors as predictive variables. During the process of training and constructing the risk prediction model, before training the maximum entropy model, the outbreak data and environmental factor data are randomly divided into training set and test set according to a set ratio. The model uses the knife-cut test to examine the contribution of each environmental factor, and verifies the model's prediction accuracy by the area under the receiver operating characteristic curve. A value greater than 0.8 indicates that the model has a good prediction effect.

9. The method for spatiotemporal risk identification and clustering detection of cysticercosis according to claim 8, characterized in that, In step six, the model performance is further optimized by adjusting the feature combination and regularization coefficient of the model; finally, a continuous risk probability layer is output, and the risk level is divided into five levels: low risk, lower risk, medium risk, higher risk and high risk using the natural breakpoint method.

Images