Spatial characteristics and driving factors of human brucellosis and plague infections in the Meriones unguiculatus plague focus in China

Dongyue Lyu; Jindong Zuo; Shuai Qin; Liping Wang; Ziyi Bu; Hanyu Sha; Lingjie Bai; Ran Duan; Linxuan Yang; Zhengliang Chai; Meng Xiao; Zhaokai He; Deming Tang; Peng Zhang; Huaiqi Jing; Xin Wang

doi:10.1016/j.onehlt.2026.101355

. 2026 Feb 12;22:101355. doi: 10.1016/j.onehlt.2026.101355

Spatial characteristics and driving factors of human brucellosis and plague infections in the Meriones unguiculatus plague focus in China

Dongyue Lyu ^a,^1,², Jindong Zuo ^b,¹, Shuai Qin ^a,¹, Liping Wang ^b,¹, Ziyi Bu ^c,¹, Hanyu Sha ^a,¹, Lingjie Bai ^d, Ran Duan ^a, Linxuan Yang ^a, Zhengliang Chai ^a, Meng Xiao ^a, Zhaokai He ^e, Deming Tang ^f, Peng Zhang ^a, Huaiqi Jing ^a, Xin Wang ^a,^⁎

PMCID: PMC12955239 PMID: 41783617

Abstract

The Meriones unguiculatus plague focus in China constitutes a key zoonotic-disease hotspot where brucellosis and plague co-circulate, posing a syndemic public-health challenge. An integrated spatial risk assessment framework was constructed to systematically contrast the spatial patterns and drivers of the two pathogens and to pinpoint potential interfaces for integrated surveillance. During 2022–2025, 17,289 blood samples were collected from 13,703 individuals. The overall seroprevalence was 19.50% for brucellosis and 0.47% for plague. Spatial analysis revealed pronounced spatial clustering of brucellosis (global Moran's I = 0.295, P < 0.05), while plague seropositivity was sporadic or locally clustered; areas of dual high-risk for both diseases were specifically identified. By integrating serological results with multi-dimensional data, a spatially enhanced random forest model (MSE = 0.005, RMSE = 0.068; AUC = 0.878) was constructed for risk assessment. Significant predictors for brucellosis included male gender, livestock-related occupation, higher vegetation index (NDVI), elevated PM₁₀ levels, sheep density, and proximity to dairy retail outlets. NDVI and livestock operation density emerged as key indicators for joint surveillance. These findings suggest that the interplay between agricultural intensification and climate variability appears to jointly facilitate transmission, underscoring the importance of adopting integrated surveillance strategies to control these zoonotic diseases.

Keywords: Brucellosis, Plague, Spatial distribution, Driving factors, Random forest model, Eigenvector spatial filtering, Risk prediction

Graphical abstract

1. Introduction

Central-eastern Inner Mongolia represents one of the most critical and challenging areas for zoonosis prevention and control in China. The unique ecological-geographical features of this region—vast temperate grasslands interspersed with desert steppes—provide a suitable ecological niche for the survival and transmission of multiple zoonotic pathogens [1]. Increasingly frequent extreme climate events and persistent land-use changes, such as grassland reclamation, intensification of livestock farming, and increased trade and transportation, are reshaping host distributions, vector abundance, and the scope of human activities. Together, these factors form a complex ecological-social environment for zoonotic disease risk [[2], [3], [4]]. This area constitutes a natural plague focus for Meriones unguiculatus, identified in 1954 [5,6], where animal outbreaks have persisted and frequent zoonotic spillovers continue to occur. Between 2019 and 2024, human plague cases reported from this focus accounted for 60% (11/18) of the national total for the same period, matching the cumulative sum reported over the previous six decades [7,8], indicating a growing public health threat. Concurrently, brucellosis remains highly endemic in this region, with persistently high annual incidence in areas such as the Ordos Plateau [9,10]. In some localities, incidence rates can exceed 300 per 100,000 [11,12], and recent years have seen significant increases in the Yinshan Mountain area [[12], [13], [14]]. Cattle and sheep, as primary hosts, sustain transmission risks through direct contact or foodborne routes [[15], [16], [17]].

Although plague (primarily transmitted via rodent–flea cycles) and brucellosis (predominantly spread through direct livestock contact) differ in their main transmission mechanisms, they share significant overlaps in secondary/incidental hosts [18](e.g., sheep, camels, herding dogs [[19], [20], [21], [22], [23]]), environmental drivers, and high-risk human populations (e.g., herders) within this specific eco-social system of the M. unguiculatus plague focus. The intensified human–livestock–environment contact driven by agricultural intensification [[24], [25], [26], [27]] further reinforces the public health necessity for integrated surveillance and management of both diseases in this region. Despite evidence that these factors individually influence the transmission of plague and brucellosis [19], few studies have systematically integrated their spatiotemporal dynamics or quantitatively assessed their interactions. Moreover, co-infection risks and the compounding effects of multiple risk factors remain insufficiently studied. Consequently, it has been challenging to systematically reveal the convergence and divergence in the spatial risk patterns of the two diseases against a shared ecological and socio-economic background.

Guided by the “One Health” concept, this study establishes a systematic spatial comparative analytical framework within the M. unguiculatus plague focus. By integrating serological surveillance data from 2022 to 2025 with multi-dimensional environmental and socio-economic datasets, a spatially integrated risk assessment framework is constructed. The specific objectives are: (1) to compare and analyse the spatial distribution patterns of plague and brucellosis; (2) to quantitatively analyse and compare the key environmental and socio-economic drivers influencing the risk of both diseases; and (3) based on the risk assessment results, to identify high-risk areas and potential targets for coordinated surveillance. This study applies spatial regression and machine learning to evaluate and forecast the occurrence of both diseases, offering a framework for targeted, cross-sectoral prevention strategies in high-risk regions.

2. Materials and methods

2.1. Research area

The M. unguiculatus plague natural focus in the Inner Mongolia Plateau spans the Bayannur-Alxa Plateau, Ordos Plateau, Mu Us Desert, Hetao Plain, Yinshan Mountains, Ulanqab Plateau, and the Erlianhot–Xilingol Desert Steppe (Fig. 1). This study primarily focused on two high-incidence regions: the Yinshan Mountains (Region A: 107.261586°–112.995330°E, 40.244194°–43.378034°N) and the Ordos Plateau (Region B: 105.552565°–108.540257°E, 36.013499°–38.836706°N), encompassing 99 counties.

Fig. 1 — Geographical location, topographic distribution, and land cover types of the *Meriones unguiculatus* plague foci.

(a) Geographical location (b) Topographic distribution and altitude. (c) Land cover type map of Region A (d) Land cover type map of Region B.

2.2. Sample collection and data collection

2.2.1. Sample collection

From January 2022 to February 2025, active surveillance was carried out by sentinel hospitals and local Centers for Disease Control and Prevention to monitor individuals presenting with symptoms such as fever, fatigue, or joint pain in fever clinics within the study area, as well as those with a history of exposure to pastoral regions. In parallel, community-based searches were implemented, with regular adjustments to the coverage areas of sentinel sites to ensure broad geographical representation. This ensured the inclusion criteria effectively captured individuals under active monitoring who represented populations at elevated risk. Blood samples were collected and centrifuged at 4000 rpm for 8 min to separate the serum. Subsequently, serological testing for brucellosis and plague was performed to determine antibody titers.

Exclusion criteria included: (a) individuals with a laboratory-confirmed infection who had completed antimicrobial therapy within the previous 30 days, to avoid residual seroprevalence bias; (b) repeated visits of the same participant within a single monitoring round, to prevent duplicate records.

2.2.2. Background data collection

Demographic data collected included age, gender, occupation, and residential coordinates of participants (Table S1). In addition, the following datasets were gathered within the study area:

Livestock statistics: Distribution of cattle, sheep, goats, and pigs (source: Food and Agriculture Organization of the United Nations, https://data.apps.fao.org/catalog//iso/9d1e149b-d63f-4213-978b-317a8eb42d02).

Environmental data: Altitude, annual maximum Normalized Difference Vegetation Index (NDVI), soil moisture, and meteorological data including temperature, rainfall, PM_2.5, and PM₁₀ (sources: China High Air Pollutants (CHAP), https://data.tpdc.ac.cn/home; National Earth System Science Data Center, https://www.geodata.cn/oldindex.html).

Food and traffic data: Point of Interest (POI) data, including locations of livestock operations, dairy and meat retail outlets, and transportation infrastructure (comprising national highways, provincial roads, and rural roads), were obtained via the Baidu Maps open API. The density of livestock operations and dairy and meat retail outlets was subsequently calculated using the kernel density estimation method.

Map data: Administrative boundaries (approved by the National Administration of Surveying, Mapping and Geoinformation, map review number: GS (2024) No. 0650) and a 30-m resolution digital elevation model (DEM) (Geospatial Data Cloud Platform, Chinese Academy of Sciences, https://www.gscloud.cn).

2.3. Serological testing methods

2.3.1. Serological testing for Brucella

Brucella antibodies were screened using the Rose Bengal Plate Test (RBPT; Idexx Laboratories, Westbrook, Maine, USA; Lanzhou Institute of Biological Products Co., Ltd., Lanzhou, Gansu Province, China). Equal volumes (30 μL) of antigen and serum were mixed and observed after 4 min for agglutination, with positive results indicated by visible clumping.

RBPT-positive samples were further confirmed by the standard tube agglutination test (SAT). Serum dilutions of 1:50, 1:100, 1:200, 1:400, and 1:800 were prepared, with Brucella antigen added alongside positive and negative controls. After incubation at 37 °C for 20–22 h, an antibody titer ≥1:50 (++) was considered positive [19]. The RBPT and SAT were used sequentially for screening and confirmation. The calculated sensitivity based on the serial test is approximately 84.52%, and the specificity is approximately 99.91%.

2.3.2. Serological testing for Y. pestis

Yersinia pestis F1 antibodies were screened using colloidal gold immunochromatography (Beijing Jianaixi Biotechnology Co., Ltd., Beijing, China), and positive cases were confirmed by indirect hemagglutination assay (IHA; Qinghai Province Endemic Disease Prevention and Control Institute). The assay included F1 antigen inhibition controls and standard positive/negative controls. A titer of ≥1:16 was considered positive [28].

F1 antibody-positive sera were further analyzed via SDS-PAGE followed by Western blotting. The F1 antigen was separated and transferred to a membrane, then probed with primary antibodies and detected using anti-human IgG secondary antibodies [29]. Band intensity was quantified using ImageJ software to determine integrated optical density (IOD) values. Colloidal gold immunochromatography and the IHA were applied in series, with Western blot (F1 antigen) reserved for further verification of equivocal outcomes. Based on this approach, the overall diagnostic sensitivity derived from serial testing was approximately 74.48%, while specificity reached approximately 99.92%.

2.4. Statistical analysis

2.4.1. Data preprocessing

2.4.1.1. Demographic variable handling

Age was grouped into five 20-year intervals: 0–19, 20–39, 40–59, 60–79, and ≥ 80 years. Occupation was classified into high-risk (e.g., livestock farmers, veterinarians) and non-high-risk categories.

2.4.1.2. Management of repeated samples

Individuals providing multiple samples were considered re-examined participants. For population-level analysis, only the first seropositive result was used. For individual-level analysis, all serological results were retained.

2.4.1.3. Spatial data standardization

All geographic data, including sampling locations, were standardized using the WGS84 coordinate system and validated. A spatial weights matrix was constructed using the Queen contiguity rule.

2.4.1.4. Temporal dimension handling

To account for disease incubation periods, environmental variables such as NDVI and meteorological data were temporally lagged by one year [30].

2.4.1.5. Variable transformation and encoding

Continuous variables were standardized using Z-scores or log-transformed. Categorical variables were encoded using dummy variables (Table S2).

2.4.1.6. Quality control

Missing occupation data were imputed using the multiple imputation by chained equations (MICE) algorithm. Multicollinearity was assessed using variance inflation factors (VIF), with variables having VIF ≥5 excluded from analysis.

2.4.2. Analysis of population characteristics and individual serum antibody titers

2.4.2.1. Analysis of population characteristics

Seroprevalence of brucellosis and plague was calculated across the entire population. Chi-square tests assessed differences by gender, age, and occupation. Geometric mean titers (GMTs) were calculated and log-transformed for comparison. Normality of data was tested; normally distributed data were analyzed using t-tests, while non-normal data were assessed using the Mann–Whitney U test (for two groups) or Wilcoxon signed-rank test (for multiple groups). A P-value <0.05 was considered statistically significant.

2.4.2.2. Analysis of individual serum antibodies titers

Longitudinal trends in antibody titers for Brucella and Y. pestis were analyzed. Particular focus was given to participants initially negative for Brucella antibodies who later seroconverted, with their antibody titer trajectories recorded.

2.4.3. Detection of spatial clustering

Spatial patterns of brucellosis and plague seroprevalence were analyzed using Moran's I index. Based on significance and spatial clustering, regions were categorized as high-high, low-low, high-low, or low-high clusters. Bivariate Moran's I was used to assess spatial associations between brucellosis and plague seroprevalence.

2.4.4. Core analytical framework: a spatially enhanced chained random forest model

A Spatially Enhanced Chained Random Forest framework is adopted as the core analytical model to build a robust and interpretable predictive model, while effectively addressing spatial autocorrelation, complex nonlinear relationships among variables, and the severe class imbalance of plague-positive samples. The implementation consists of the following steps:

2.4.4.1. Spatial feature extraction and selection

To explicitly capture and control for spatial effects, the Rapid Eigenvector Spatial Filtering (RES-ESF) model [31] is first employed to extract spatial eigenvectors from the spatial distribution of disease seroprevalence exhibiting spatial clustering. From the complete set of generated eigenvectors, a subset is selected by ranking them in descending order based on their corresponding Moran's I values; the top K eigenvectors that collectively account for over 80% of the explained spatial variance are retained. This curated set of eigenvectors is incorporated as spatial covariates into the subsequent machine learning model, thereby quantifying and integrating the predominant spatial structure into the predictive framework.

2.4.4.2. Variable screening and Preprocessing

Independent variables included demographic data (gender, age, occupation), livestock-related indicators (number of sheep, goats, cattle, and pigs; livestock operation density), environmental variables (altitude, NDVI, annual mean temperature, annual mean precipitation, PM_2.5, PM₁₀), and food/transportation-related features (density of dairy and meat retail outlets, hospital accessibility, nighttime light intensity). The dependent variables were serological antibody results for Brucella and Yersinia pestis (Table S2). To address multicollinearity among the independent variables, key predictors were selected using Least Absolute Shrinkage and Selection Operator (LASSO) regression, with the optimal λ (lambda.min) used to determine the retained variables [32].

2.4.4.3. Chained random forest modeling and training

A two-stage chained random forest model is designed to address the modeling challenges posed by the low plague seroprevalence rate and to use plague as an indicator reflecting common ecological background risks that may influence the spatial distribution of zoonotic disease exposures:

Stage 1 (Plague Model): A random forest classifier is trained using plague serological results as the dependent variable and the screened environmental, livestock, and other covariates as independent variables. This model learns the risk patterns of plague infection.

Stage 2 (Brucellosis Model): Brucellosis serological results serve as the ultimate dependent variable. The independent variables for this stage include the selected covariates, the spatial eigenvectors, and the predicted probabilities from the Stage 1 Plague Model.

The iterative stratification method is employed to split the data into 80% training and 20% test sets, ensuring a consistent distribution of all class labels across both sets. On the training set, a 5-times repeated 5-fold cross-validation is performed, and a grid search is utilized to optimize hyperparameters, including: “ntree”, “mtry”, and “min.node.size”. The parameter combination demonstrating the highest mean AUC during cross-validation is selected for constructing the final model. To handle class imbalance, a weighted random forest strategy is implemented during the training of both forests, assigning higher weights to the minority class (positive samples) inversely proportional to their frequency in the training set via the “classwt” parameter.

2.4.4.4. Model performance and uncertainty assessment

Model performance was assessed using R², Akaike Information Criterion (AIC), sensitivity, precision, recall, F1-score, mean square error (MSE), and root MSE (RMSE) [33]. Ninety-five percent confidence intervals for all performance metrics are calculated using the Bootstrap method (1000 resamples). Furthermore, the contribution of each predictor is assessed using permutation importance.

The residual Moran’ s I of the RES-ESF model [34] was compared with that of traditional models (including Ordinary Least Squares (OLS) and logistic regression) to assess the model’ s ability to capture spatial autocorrelation and its robustness (Table S3). Additionally, the GeoDetector method [35] was used—integrating local livestock-related points of interest (POIs)—to examine spatial heterogeneity in brucellosis distribution (Table S4).

2.4.5. Risk mapping and hotspot prediction

The trained spatially enhanced chained random forest model is used to predict the brucellosis seroprevalence risk across the entire study area. To quantify prediction uncertainty, the aforementioned Bootstrap procedure is applied to generate the mean prediction as well as the 5th and 95th percentiles for each geographical location. Finally, ordinary Kriging interpolation is applied to spatialize the point-based predictions, producing maps of the mean predicted risk, the low-risk scenario (5th percentile), and the high-risk scenario (95th percentile), which enables the reliable identification of stable hotspot areas.

Specifically, in the spatially enhanced classifier chain random forest model, plague risk was introduced into the brucellosis model in the form of predicted probabilities rather than being used to generate a unified multi-disease risk surface. The resulting brucellosis risk maps were therefore intended to illustrate how the inclusion of another endemic zoonotic risk modifies spatial predictions, rather than to represent joint disease risk. Accordingly, these risk maps are presented as supplementary analyses in the Appendix and are not treated as primary results of the comparative risk assessment.

2.4.6. Analytical software

Data preprocessing and GeoDetector analyses were conducted using Excel 2024. Normality and chi-square tests were performed in SPSS 26.0.

Spatial analyses—including global, local, and bivariate Moran's I analyses—were performed using GeoDa software (http://geodacenter.github.io/index.html).

Model estimation and validation were performed in R program using the “stats”, “spdep”, “glmnet”, and “spmoran” functions [36]. Random forest regression was implemented with the “randomForest” package. Multicollinearity was assessed via the variance inflation factor (VIF), and AIC was used to guide model selection. R code, including data preprocessing, LASSO, RES-ESF, Random Forest settings, and visualization have been deposited as Supplementary file S1.

3. Results

3.1. Epidemiological status of brucellosis and plague in the Meriones unguiculatus plague focus

The M. unguiculatus plague focus in the Inner Mongolia Plateau spans diverse ecological regions including the Bayannur-Alxa Plateau, Ordos Plateau, and Yinshan Mountains (Fig. 1), where we conducted intensive surveillance from January 2022 to February 2025, collecting 17,289 serum samples from 13,703 individuals, of which 2299 individuals (5885 samples) underwent follow-up testing. Our study focused on two regions: (a) the Yinshan Mountains (Region A) characterized by 1500-2300 m elevation temperate grasslands, and (b) the Ordos Plateau (Region B) featuring 1000-1600 m desert steppes.

The overall brucellosis seroprevalence was 19.50% (2680/13,703), with a geometric mean antibody titer of 160.53. Positivity rates varied significantly by age, sex, and occupation. Males exhibited higher rates than females (34.9% vs. 21.2%, χ² = 264.78, P < 0.01), with the highest prevalence observed in the 40–59 age group. High-risk occupational groups such as farmers and veterinarians—those with frequent exposure to cattle and sheep—had significantly higher seroprevalence than those in non-exposed occupations (33.3% vs. 17.7%, χ² = 85.52, P < 0.01; Table S1). The positivity rate was 14.2% in Region A (1,016/7123) and 25.2% in Region B (1,664/6580; Figs. 2a i, 2a ii, 2b).

Fig. 2 — Seropositive rate distribution and antibody titer levels of brucellosis and plague in the study area.

(a) Seropositive rate distribution of brucellosis and plague, (b) antibody titer levels of brucellosis and plague, (c) Changes of serum antibodies titers against *Brucella*: (i) to (iv) are respectively the individual changes of typical serum antibodies titers against *Brucella* presented in the reexamined patients (rose, declined, rose first and then fell, first dropped to negative and then rose); (v) is the changes of serum antibodies titers against *Brucella* in reexamined patients with negative serum antibody against Brucella collected for the first time and subsequently turned positive. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

For plague, the overall F1 antibody positivity rate was 0.47% (65/13,703), with a geometric mean titer of 19.18. Age was significantly associated with F1 positivity (χ² = 5.35, P < 0.01), while gender (χ² = 1.25, P = 0.364) and occupation (χ² = 0.98, P = 0.523; Table S1) were not. Region A had a positivity rate of 0.63% (45/7123), compared to 0.30% (20/6580) in Region B (Figs. 2a iii, 2a iv, 2b).

3.2. Changes in antibody titers to brucellosis and plague

Individual trends in serum antibody responses to brucellosis were classified among 2299 subjects, revealing four distinct patterns (Fig. 2c i–iv): (i) consistently decreasing titers, (ii) consistently increasing titers, (iii) titers that initially increased and then declined, and (iv) titers that initially declined to undetectable levels before increasing. In contrast, plague antibody titers showed an overall declining trend (the results of Western blot are shown in Fig. S1).

Based on the criterion of an initial negative and subsequent positive brucellosis antibody result, 88 cases were selected for analysis of titer dynamics. These cases typically exhibited a rapid rise in antibody levels at the onset of infection, sustained for a period before gradually declining (Fig. 2c v).

3.3. Spatial analysis

Global Moran's I spatial autocorrelation analysis indicated that the distribution of brucellosis antibody positivity from 2022 to 2025 in the M. unguiculatus natural plague focus was not random but spatially clustered (Moran's I = 0.295; Z = 12.16; P < 0.05). Local Moran's I analysis further revealed that Region A predominantly exhibited low-low clusters—characterized by generally low positive rates with localized high “point-line” patterns. Region B showed high-high clustering, with overall elevated positivity and some localized low spots (Fig. S2). Brucellosis positivity in Region B was generally higher than in Region A.

Regional analyses yielded the following findings: ① In Region A, brucellosis cases were concentrated on the eastern and western flanks (Fig. 3a), while plague cases clustered in the central zone (Fig. 3b). Bivariate Moran's I analysis identified a central high-high cluster, suggesting that while brucellosis infections are highly prevalent in this area, plague infections are also relatively high. Other regions exhibited high-low and low-high patterns, suggesting spatial segregation in disease distribution; no low-low clusters were observed (Fig. 3c). ② In Region B, brucellosis cases were clustered in the southeast (Fig. 3d), and plague in the northeast (Fig. 3e). Bivariate Moran's I showed a dispersed spatial distribution of both infections (Fig. 3f). A spatial model was employed to account for the observed autocorrelation.

3.4. Multi-model analysis of infection risk factors

3.4.1. Inclusion variable screening

Seventeen potential risk factors were evaluated using Lasso regression, identifying 13 significant predictors: gender, occupation, densities of cattle, sheep, goats, and pigs, livestock operation density, densities of dairy and meat retail outlets, hospital accessibility, temperature, precipitation, vegetation index, and PM₁₀ concentration (Fig. 4a–b). Multicollinearity diagnostics showed all variance inflation factors <5, indicating no collinearity concerns.

Fig. 4 — Selection, ranking, and interaction analysis of influencing factors.

(a, b) Variable selection by LASSO regression, (c, d) Gini index and variable ranking of the random forest model, (e) Heatmap of the interaction analysis of the influencing factors of brucellosis by geodetector.

3.4.2. Spatial eigen-feature extraction

Brucellosis seroprevalence exhibited significant spatial clustering (Moran's I = 0.295). Eigenvector spatial filtering (RES-ESF) was therefore applied to the sampling-point coordinates to extract spatial eigenvectors. Ordinary least squares (OLS), logistic regression, and spatial autoregressive models were compared (Table S3). The spatial filtering model yielded a residual Moran's I of 0.036, effectively minimizing spatial autocorrelation.

3.4.3. Optimal model selection

Risk factors for brucellosis infection included male gender, agricultural occupation, high sheep and pig densities, elevated vegetation index, and exposure to inhalable particulates (P < 0.05, Table S3 and Table S5).

3.4.4. Combined ranking of important factors

Random forest Gini index rankings identified gender, cattle and sheep numbers, livestock operation density, vegetation index, precipitation, and PM₁₀ as key predictors of brucellosis (Fig. 4c). For plague, vegetation index, PM₁₀ concentration, and livestock operation density were most influential. The vegetation index emerged as the top factor for both diseases (Fig. 4d). The Gini index was applied as a core metric in the random forest model to measure the importance of a variable (feature) in distinguishing between different classes (e.g., brucellosis positive vs. negative). Variables ranked higher by the Gini index exhibit stronger predictive power for the disease risk. The importance of factors common to both diseases is illustrated using Gini scatter plots for the brucellosis and plague models.

3.4.5. Interaction of important factors

Geodetector analysis revealed a strong correlation between livestock-related industrial density and brucellosis positivity (q = 0.97, P < 0.05). Interactions between livestock operation density and retail outlet density showed nonlinear enhancement, as did the combination of average temperature and vegetation index (Fig. 4e).

3.5. Predictive analysis

After feature vector integration and model training, the optimized random forest model explained 84% of the training data. The model demonstrated robust performanceacross various metrics, with an average overall accuracy of 0.877 (Accuracy 95% CI: 0.870–0.883), macrosensitivity of 0.990, macro precision of 0.872, and an average AUC score of 0.878. On the prediction set, it achieved 78.9% accuracy (Fig. 5a–b). Kriging interpolation identified brucellosis antibody hotspots covering 10.4%–38.2% of the study area within the 5%–95% confidence interval (Fig. S3). The core high-infection-rate hotspots (red) identified in the mean map (Fig. S3(i)) are still stably present in the high-risk scenario map at the 95th percentile (Fig. S3(iii)), with an expanded range. This suggests that these areas are highly reliable priorities for control measures. In contrast, the boundaries of the moderate-risk areas (yellow) show some variation across different percentile maps, indicating uncertainty in the predictions for these areas and suggesting that their risk levels may fluctuate with local factors.

Fig. 5 — Prediction of random forest model based on spatial filtering method.

(a) Sensitivity and specificity of the training set and validation set, (b) ROC curve of the prediction model.

4. Discussion

This study investigated the spatial distribution patterns and underlying drivers of human brucellosis and plague within the M. unguiculatus plague focus from 2022 to 2025. Serological surveys revealed a brucellosis seroprevalence of 19.50% and a plague seroprevalence of 0.47%, confirming ongoing transmission of both diseases in the region. It should be emphasized that the findings of this serological survey must be interpreted within the framework of its active surveillance design. Therefore, the reported seroprevalence rates reflect the disease burden within high-risk populations and should not be directly extrapolated to the general community. However, this targeted approach proves effective in identifying genuine transmission hotspots. Spatial analysis showed marked heterogeneity in brucellosis distribution (global Moran's I = 0.295, P < 0.05). Region A exhibited a predominantly “low-low” cluster, while Region B displayed a “high-high” cluster (Fig. S2), with significantly higher infection rates in the latter. These findings align with historical surveillance data. From 2004 to 2019, Region B experienced clustered migration of brucellosis, with cases increasingly concentrated in this region [4]. Between 2010 and 2014, brucellosis seroprevalence in Region B remained high (23.8%–38.0%) [37], comparable to this study's observed rate of 25.2%, reaffirming its status as a long-standing endemic hotspot. In contrast, the brucellosis seroprevalence rate in Region A detected through active surveillance (14.2%) shows a several-fold increase compared to the 0.90–7.07% rate reported in the 2012–2016 general population passive surveillance [38]. This disparity can be attributed to the focused surveillance on traditional pastoral areas and high-risk occupational groups, as well as the emerging high-risk clusters gradually forming in this region, as indicated by 2022 data [13,14]. These findings indicate that although brucellosis remains more prevalent in Region B, Region A also carries a potential infection risk. Moreover, active surveillance offers a more comprehensive picture of disease distribution patterns. Furthermore, the serial serological testing protocol employed demonstrates high specificity, ensuring that the elevated seroprevalence rates are not due to false positives. Thus, the high seroprevalence rates provide evidence of successfully captured core transmission chains within the foci, highlighting the urgency and importance of implementing targeted active surveillance in such areas.

To further assess the spatial relationship between brucellosis and plague seroprevalence in Region A, bivariate Moran's I analysis was performed. The results indicated the presence of predominantly “high-low” and “low-high” clusters, with only a few areas exhibiting “high-high” clustering (Fig. 3c, f). These limited “high-high” clusters represent critical risk interfaces identified through integrated analysis within a complex eco-geographical context and should be prioritized as key targets for joint prevention and control strategies. These areas suggest that in certain traditional pastoral regions, human activity zones, livestock production systems, and elements of the plague natural focus may be spatially adjacent or locally overlapping, thereby carrying particular public health relevance. In contrast, the lack of overlap between most brucellosis high-risk areas and plague high-risk areas can be attributed to two primary factors. First, the distribution of M. unguiculatus, the principal host of plague in this region, is influenced by their ecological preferences. Vegetation variability alters rodent density—areas affected by severe grassland desertification support higher M. unguiculatus populations, whereas regions with denser vegetation support fewer M. unguiculatus rodents [39]. Second, brucellosis-prone areas within the plague-endemic zones of M. unguiculatus are typically pastoral landscapes with denser human populations. These areas prioritize rodent control due to zoonotic disease prevention efforts and grazing management needs, thereby reducing rodent densities. Therefore, in the joint ranking of key factors for both brucellosis and plague, the prominence of NDVI provides evidence that geographical environment contributes to the spatial dissociation of high-risk areas for the two diseases, while also explaining why NDVI ranks highly among various influencing factors.

Changes in individual serum antibody titers offer insights into the immunological dynamics of brucellosis and plague in this region. Fluctuations in anti-Brucella antibody titers reflect the long-term endemicity of brucellosis in the population. Prior studies have shown that repeated exposure to infected animals leads to antibody patterns characterized by fluctuation and gradual decline [40]. Similarly, seropositive individuals in this study displayed antibody dynamics distinct from the classical three-phase model—comprising a rapid rise, rapid decline, and subsequent slow decline [41]. Specifically, some individuals showed recurring fluctuations, with antibody titers declining and later rising again, suggesting repeated exposure. Others maintained prolonged seroprevalence exceeding 200 days, possibly indicating incomplete treatment and a risk of chronic infection (Fig. 2c v). In contrast, anti-Yersinia pestis F1 antibody titers in plague-seropositive individuals showed a consistent downward trend, indicative of convalescence rather than ongoing infection. Western blot analysis further demonstrated that, alongside the plague-specific 17 kDa and 35 kDa bands, human sera exhibited additional reactive bands not present in canine sera [42]. This suggests interspecies differences in immune responses to Y. pestis F1 antigen, potentially due to species-specific immune recognition mechanisms [43].

In recent years, intensification of livestock production has been shaped by expanding regional herd sizes, increased inter-provincial animal trade, and climate change. These factors collectively influence the spatial and temporal patterns of brucellosis and plague. This study integrates regional variables to evaluate their combined effect on brucellosis prevalence (Fig. 4e). Demographically, although plague shows no statistically significant gender bias, male agro-pastoral workers face elevated infection risks due to high-exposure behaviors such as animal slaughter and midwifery. Regarding livestock distribution, sheep remain a major risk factor for brucellosis, consistent with previous findings [44]. Local surveillance reports indicate that seroprevalence for plague in key animal hosts ranges between 5% and 10%, reaching even higher levels during epizootic periods [14,18]. Concurrently, documented brucellosis seroprevalence in livestock such as dairy cattle and sheep in these areas is also considerable—for instance, 10–20% in dairy cattle and 1.8–3.5% in sheep [1,12]. The presence of these two zoonotic pathogens within the same animal populations and geographical landscapes suggests a plausible reservoir for potential co-circulation, even though direct evidence of co-infection in individual animals from our specific study area remains limited. Moreover, cases of plague-infected sheep leading to human transmission have been documented [23]. Other animals, such as camels (6.2% seroprevalence) and sheepdogs (1.8%), have also tested positive for brucellosis [19]. Notably, camels have been identified as Y. pestis hosts in Africa [45], and sheepdogs serve as sentinel animals for plague surveillance, offering early outbreak warnings [46]. These findings highlight the risk of zoonotic spillover from secondary or incidental host species, warranting continued vigilance toward dogs, camels, and other non-primary hosts of brucellosis and plague [47]. Environmental factors further shape disease patterns. This study confirmed that the Normalized Difference Vegetation Index (NDVI) positively correlates with brucellosis expansion, aligning with findings from Inner Mongolia. Lush vegetation supports grazing activities and thereby increases opportunities for Brucella transmission. Other research has also suggested that vegetation type may influence brucellosis transmission dynamics [48], a topic meriting further investigation. NDVI ranked prominently among the joint predictors of brucellosis and plague, reinforcing the role of environmental geography in driving spatial segregation between high-incidence zones. PM₁₀ emerged as a shared and significant predictor for both diseases in the model; however, it is important to note that this statistical association does not necessarily imply a direct or mechanistic role in pathogen transmission. PM₁₀ levels in this semi-arid region arise from complex sources, including natural dust, regional transport, and localized anthropogenic activities such as livestock movement and land use. Within the analytical framework of this study, PM₁₀ is therefore better regarded as a non-specific composite indicator of general environmental conditions, which may co-vary with the actual factors influencing disease exposure. Its predictive value likely stems from its correlation with broader ecological and anthropogenic landscape features. Future research aimed at disentangling the specific components of this environmental mixture would be of considerable scientific value. Socioeconomic factors were also examined. Although urbanization indicators—such as nighttime light intensity and healthcare accessibility—showed no statistical association with disease prevalence in this study, previous evidence suggests urbanization fosters intensive livestock production [49]. Despite recent declines in unregulated livestock trade [50], changes in land-use patterns and the rise of intensive farming systems continue to heighten zoonotic transmission risks [27].

The study also identified synergistic interactions among key influencing factors. (1) Industry–consumption network effects: The interaction between livestock operation density and the spatial distribution of dairy and meat product retail (q-value increased from 0.609 to 0.625) indicates that contact-based and foodborne exposures mutually reinforce brucellosis transmission (Fig. 4e). Region B, a livestock hub bordering four provinces [51,52], is characterized by intensive farming and large-scale inter-regional livestock movement. Concurrently, regional product circulation centers act as secondary transmission nodes via contaminated food consumption [50]. This dual pathway helps explain infections in individuals without direct animal contact, underscoring the presence of a production-to-consumption transmission chain. Previous studies have linked large-scale farms to increased brucellosis seroprevalence, while small- and medium-scale farms exhibit neutral or negative associations [53], likely due to differences in technological adoption and agricultural knowledge. (2) Climate–ecology synergistic effects: Rising annual temperatures indirectly increase transmission risk by promoting vegetation growth (higher NDVI). This occurs through: (i) enhanced pasture productivity and altered grazing patterns due to moderate warming, and (ii) temperature ranges that accelerate Brucella propagation [54]. This climate–vegetation–pathogen cascade, when coupled with agricultural intensification, forms a multidimensional driver network for brucellosis spread. Public health significance: This integrated spatial risk modeling framework provides an operational tool for precision surveillance of brucellosis and plague. The model outputs can guide local CDCs to prioritize high-risk zones and occupational groups for intervention, allocate limited diagnostic and prophylactic resources, and enhance communication strategies in endemic areas. Because all input variables were derived from publicly available or routine surveillance datasets, the model is replicable and can be extended to other zoonotic diseases in China and beyond.

A multi-stage modeling strategy was employed to systematically assess brucellosis and plague transmission risks. Initial linear models revealed significant nonlinear associations, prompting the adoption of machine learning approaches. These models can capture complex interactions—both positive and negative—and improve performance on unseen data [55]. Among the evaluated models, the random forest algorithm outperformed others (MSE = 0.005, RMSE = 0.068), likely due to its Bagging-based architecture, which constructs independent decision trees well-suited for data with multiple local optima. To address spatial autocorrelation, eigenvector spatial filtering was applied to extract spatial eigenvectors from geographic coordinates and integrate them into the random forest along with selected non-spatial covariates. This approach improved model fit (R²) and reduced residual spatial autocorrelation (Moran's I), thereby significantly enhancing predictive accuracy. The spatially enhanced classifier chain random forest model employed effectively controlled spatial autocorrelation by spatial eigenvector (RES-ESF). Through its chain-based design, plague risk was introduced into the brucellosis prediction framework as background ecological information, thereby improving model robustness and providing a scalable framework for multi-disease risk modeling.

By integrating spatial analytical approaches with machine learning methods, this study achieved a comparative risk assessment of brucellosis and plague within the M. unguiculatus plague focus. The results elucidate the distinct spatial logics of the two diseases operating under the same ecological context and provide empirical evidence and methodological tools to support the development of coordinated, precise, and sustainable “One Health” surveillance and response systems. Future research may further extend this framework by incorporating animal infection data, environmental pathogen indicators, and dynamic meteorological information, thereby advancing toward real-time early warning and dynamic risk assessment.

CRediT authorship contribution statement

Dongyue Lyu: Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Data curation. Jindong Zuo: Writing – original draft, Resources, Project administration, Methodology, Investigation, Data curation. Shuai Qin: Writing – original draft, Validation, Software, Methodology, Investigation, Data curation. Liping Wang: Writing – original draft, Resources, Project administration, Investigation, Data curation. Ziyi Bu: Writing – original draft, Validation, Software, Investigation, Data curation. Hanyu Sha: Writing – original draft, Visualization, Software, Investigation, Data curation. Lingjie Bai: Writing – review & editing, Resources, Project administration, Methodology. Ran Duan: Writing – review & editing, Supervision, Investigation. Linxuan Yang: Writing – review & editing, Supervision, Investigation. Zhengliang Chai: Writing – review & editing, Supervision, Investigation. Meng Xiao: Writing – review & editing, Supervision, Investigation. Zhaokai He: Writing – review & editing, Supervision, Investigation. Deming Tang: Writing – review & editing, Supervision, Investigation. Peng Zhang: Writing – review & editing, Supervision, Investigation. Huaiqi Jing: Writing – review & editing, Project administration, Methodology, Formal analysis, Conceptualization. Xin Wang: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis, Conceptualization.

Ethics approval statement

The study was approved by the ethics committee of the National Institute for Communicable Disease Control and Prevention of the Chinese Center for Disease Control and Prevention (NO. ICDC-LPJ-2024005) and the ethics committee of People's Hospital of Yanchi County (January 2023–December 2024).

Code availability

The data analysis was coded by R 4.4.2, whose code is available on Supplementary File 1. R code.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:

I have nothing to declare If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

This work was supported by the National Key Research and Development Program of China (NO. 2024YFC2311500 and No. 2022YFC2602203).

Footnotes

^{Appendix A}

Supplementary data to this article can be found online at https://doi.org/10.1016/j.onehlt.2026.101355.

Appendix A. Supplementary data

R code

mmc1.docx^{(41.6KB, docx)}

Study population characteristics, variable definitions, and additional spatial modeling results

mmc2.docx^{(384.5KB, docx)}

Data availability

All data generated or analyzed during this study are included in this published article, and the supplementary information files will be freely available to any scientist for non-commercial purposes upon request to the corresponding author via email.

References

1.Du H.-W., Wang Y., Zhuang D.-F., Jiang X.-S. Temporal and spatial distribution characteristics in the natural plague foci of Chinese Mongolian gerbils based on spatial autocorrelation. Infect. Dis. Poverty. 2017;6:124. doi: 10.1186/s40249-017-0338-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Zheng L., Zhang Q., Zhang A., Hussain H.A., Liu X., Yang Z. Spatiotemporal characteristics of the bearing capacity of cropland based on manure nitrogen and phosphorus load in mainland China. J. Clean. Prod. 2019;233:601–610. doi: 10.1016/j.jclepro.2019.06.049. [DOI] [Google Scholar]
3.Cheng M., Quan J., Yin J., Liu X., Yuan Z., Ma L. High-resolution maps of intensive and extensive livestock production in China. Resour. Environ. Sustain. 2023;12 doi: 10.1016/j.resenv.2022.100104. [DOI] [Google Scholar]
4.Yunlong M.A., Ying L.I.U., Xiaolong L.I., Lina S., Tong Y., Xiaoxia M.A., Changsheng Z., Yu Z. Spatial-temporal distribution and influencing factors of Brucellosis in China. zhjbkzzz. 2023;27:1241–1246. doi: 10.16462/j.cnki.zhjbkz.2023.11.001. [DOI] [Google Scholar]
5.Ben-Ari T., Neerinckx S., Agier L., Cazelles B., Xu L., Zhang Z., Fang X., Wang S., Liu Q., Stenseth N.C. Identification of Chinese plague foci from long-term epidemiological data. Proc. Natl. Acad. Sci. USA. 2012;109:8196–8201. doi: 10.1073/pnas.1110585109. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Williams W. The Anatomy of the Mongolian Gerbil (Meriones unguiculatus) 1974. https://www.semanticscholar.org/paper/The-anatomy-of-the-Mongolian-gerbil-(Meriones-Williams/ce0bf69a70dcc0c08472ffddefec8407a257f9f8 (accessed May 30, 2025)
7.Liu B., Zhang D., Chen Y., He Z., Liu J., Lyu D., Wu W., Duan R., Qin S., Liang J., Jing H., Wang X. Epidemiological characteristics of plague in the Meriones unguiculatus plague focus - Inner Mongolia Autonomous Region, China, 1950-2019. China CDC Wkly. 2020;2:935–945. doi: 10.46234/ccdcw2020.256. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.National Disease Control and Prevention Administration Government Information Disclosure. 2025. https://www.ndcpa.gov.cn/jbkzzx/c100028/common/list.html Accessed May 30, 2025. (accessed May 30, 2025)
9.Zhang W.-Y., Guo W.-D., Sun S.-H., Jiang J.-F., Sun H.-L., Li S.-L., Fang L.-Q., Cao W.-C. Human Brucellosis, Inner Mongolia, China. Emerg. Infect. Dis. 2010;16(12) doi: 10.3201/eid1612.091081. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Tao Z., Chen Q., Chen Y., Li Y., Mu D., Yang H., Yin W. Epidemiological characteristics of human brucellosis - China, 2016-2019. China CDC Wkly. 2021;3:114–119. doi: 10.46234/ccdcw2021.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Wang X. Analysis of the epidemiological characteristics and trends of human brucellosis in Ordos City from 2005 to 2016. Chin. J. Ctrl. Endem. Dis. 2019;34(01):66–69. (In Chinese) [Google Scholar]
12.Gao M. Analysis of epidemiological characteristics of human brucellosis in Ordos City from 2019 to 2023. Chin. J. Ctrl. Endem. Dis. 2024;39(06):479–480. (In Chinese) [Google Scholar]
13.Peng C., Li Y.-J., Huang D.-S., Guan P. Spatial-temporal distribution of human brucellosis in mainland China from 2004 to 2017 and an analysis of social and environmental factors. Environ. Health Prev. Med. 2020;25:1. doi: 10.1186/s12199-019-0839-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Song J., Han S., Wang H.H. Spatio-temporal aggregation analysis of human brucellosis in Inner Mongolia Autonomous Region from 2018 to 2022. Chin. J. Zoonoses. 2023;39(10):993–997. (In Chinese) [Google Scholar]
15.Negrón M.E., Kharod G.A., Bower W.A., Walke H. Notes from the field: human Brucella abortus RB51 infections caused by consumption of unpasteurized domestic dairy products - United States, 2017-2019. MMWR Morb. Mortal Wkly. Rep. 2019;68:185. doi: 10.15585/mmwr.mm6807a6. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Samadi A., Amiri M., Hailat N. The reasons behind long-term endemicity of brucellosis in low and middle-income countries: challenges and future perspectives. Curr. Microbiol. 2024;81:82. doi: 10.1007/s00284-023-03605-5. [DOI] [PubMed] [Google Scholar]
17.Khoshnood S., Pakzad R., Koupaei M., Shirani M., Araghi A., Irani G.M., Moradi M., Pakzad I., Sadeghifard N., Heidary M. Prevalence, diagnosis, and manifestations of brucellosis: a systematic review and meta-analysis. Front. Vet. Sci. 2022;9 doi: 10.3389/fvets.2022.976215. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.He Z., Wei B., Zhang Y., Liu J., Xi J., Ciren D., Qi T., Liang J., Duan R., Qin S., Lv D., Chen Y., Xiao M., Fan R., Song Z., Jing H., Wang X. Distribution and characteristics of human plague cases and Yersinia pestis isolates from 4 Marmota plague foci, China, 1950-2019. Emerg. Infect. Dis. 2021;27:2544–2553. doi: 10.3201/eid2710.202239. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Qin S., Liang J., Tang D., Chen Y., Duan R., Lu X., Bukai A., Zheng X., Lv D., He Z., Wu W., Han H., Jing H., Wang X. Serological investigation of plague and brucellosis infection in Marmota himalayana plague foci in the Altun Mountains on the Qinghai-Tibet plateau. Front. Public Health. 2022;10 doi: 10.3389/fpubh.2022.990218. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Bin Saeed A.A., Al-Hamdan N.A., Fontaine R.E. Plague from eating raw camel liver. Emerg. Infect. Dis. 2005;11:1456–1457. doi: 10.3201/eid1109.050081. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Christie A.B., Chen T.H., Elberg S.S. Plague in camels and goats: their role in human epidemics. J. Infect. Dis. 1980;141:724–726. doi: 10.1093/infdis/141.6.724. [DOI] [PubMed] [Google Scholar]
22.Fedorov V.N. Plague in camels and its prevention in the USSR. Bull. World Health Organ. 1960;23:275–281. [PMC free article] [PubMed] [Google Scholar]
23.Dai R., Qi M., Xiong H., Yang X., He J., Zhang Z., Yang H., Jin J., Li X., Xin Y., Yang Y., Li C., Li Z., Xu J., Wang Z., Li W., Wei B. Serological epidemiological investigation of Tibetan sheep (Ovis aries) plague in Qinghai, China. Vector Borne Zoonotic Dis. 2019;19:3–7. doi: 10.1089/vbz.2017.2257. [DOI] [PubMed] [Google Scholar]
24.Gibb R., Redding D.W., Chin K.Q., Donnelly C.A., Blackburn T.M., Newbold T., Jones K.E. Zoonotic host diversity increases in human-dominated ecosystems. Nature. 2020;584:398–402. doi: 10.1038/s41586-020-2562-8. [DOI] [PubMed] [Google Scholar]
25.Jones B.A., Grace D., Kock R., Alonso S., Rushton J., Said M.Y., McKeever D., Mutua F., Young J., McDermott J., Pfeiffer D.U. Zoonosis emergence linked to agricultural intensification and environmental change. Proc. Natl. Acad. Sci. USA. 2013;110:8399–8404. doi: 10.1073/pnas.1208059110. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Murray G.G.R., Hossain A.S.M.M., Miller E.L., Bruchmann S., Balmer A.J., Matuszewska M., Herbert J., Hadjirin N.F., Mugabi R., Li G., Ferrando M.L., de Oliveira I.M. Fernandes, Nguyen T., Yen P.L.K., Phuc H.D., Moe A. Zaw, Wai T. Su, Gottschalk M., Aragon V., Valentin-Weigand P., Heegaard P.M.H., Vrieling M., Maw M. Thein, Myint H. Thidar, Win Y. Tun, Hoa N. Thi, Bentley S.D., Clavijo M.J., Wells J.M., Tucker A.W., Weinert L.A. The emergence and diversification of a zoonotic pathogen from within the microbiota of intensively farmed pigs. Proc. Natl. Acad. Sci. USA. 2023;120 doi: 10.1073/pnas.2307773120. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.White R.J., Razgour O. Emerging zoonotic diseases originating in mammals: a systematic review of effects of anthropogenic land-use change. Mammal Rev. 2020;50:336–352. doi: 10.1111/mam.12201. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Ministry of Health of China WS 279–2008 Diagnostic Criteria for Plague. 2025. https://icdc.chinacdc.cn/zcfgybz/bz/202112/t20211202_253341.html Accessed May 31, 2025.
29.Miller J., Williamson E.D., Lakey J.H., Pearce M.J., Jones S.M., Titball R.W. Macromolecular organisation of recombinant Yersinia pestis F1 antigen and the effect of structure on immunogenicity. FEMS Immunol. Med. Microbiol. 1998;21:213–221. doi: 10.1016/S0928-8244(98)00063-7. [DOI] [PubMed] [Google Scholar]
30.Ma X., Sun G.-Q. Global dynamics of a periodic brucellosis model with time delay and environmental factors. Appl. Math. Model. 2024;130:288–309. doi: 10.1016/j.apm.2024.03.002. [DOI] [Google Scholar]
31.Helbich M., Arsanjani J.J. Spatial eigenvector filtering for spatiotemporal crime mapping and spatial crime analysis. Cartogr. Geogr. Inf. Sci. 2015 doi: 10.1080/15230406.2014.893839. (accessed May 6, 2025) [DOI] [Google Scholar]
32.Engebretsen S., Bohlin J. Statistical predictions with glmnet. Clin. Epigenetics. 2019;11:123. doi: 10.1186/s13148-019-0730-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.LeSage J., Pace R.K. Chapman and Hall/CRC; New York: 2009. Introduction to Spatial Econometrics. [DOI] [Google Scholar]
34.Lin X. Identifying peer effects in student academic achievement by spatial autoregressive models with group unobservables. J. Labor Econ. 2010 doi: 10.1086/653506. [DOI] [Google Scholar]
35.Wang J.-F., Li X.-H., Christakos G., Liao Y.-L., Zhang T., Gu X., Zheng X.-Y. Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China. Int. J. Geogr. Inf. Sci. 2010 doi: 10.1080/13658810802443457. [DOI] [Google Scholar]
36.Murakami D. 2017. Spmoran: An R Package for Moran's Eigenvector-Based Spatial Regression Analysis. [DOI] [Google Scholar]
37.Ning C., Shuyi G., Tao Y., Hao Z., Zhang X. Epidemiological survey of human brucellosis in Inner Mongolia, China, 2010-2014: a high risk groups-based survey. J. Infect. Public Health. 2018;11:24–29. doi: 10.1016/j.jiph.2017.02.013. [DOI] [PubMed] [Google Scholar]
38.Liu Z.-G., Wang M., Ta N., Fang M.-G., Mi J.-C., Yu R.-P., Luo Y., Cao X., Li Z.-J. Seroprevalence of human brucellosis and molecular characteristics of Brucella strains in Inner Mongolia Autonomous region of China, from 2012 to 2016. Emerg. Microbes Infect. 2020;9:263–274. doi: 10.1080/22221751.2020.1720528. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Li K., Wang Z., Wei X., Ji H., Shang M., Chang N., Wang Z., Guo C., Xu L., Zhao N., Liu Q. Relationships between meteorological factors and Mongolian gerbils and its flea burdens - Xilingol league, Inner Mongolia Autonomous Region, China, 2012-2021. China CDC Wkly. 2024;6:547–552. doi: 10.46234/ccdcw2024.107. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Coghlan J.D., Weir D.M. Antibodies in human brucellosis. Br Med J. 1967;2 doi: 10.1136/bmj.2.5547.269. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Sha H., Duan Q., Lyu D., Qian F., Zheng X., Guo J., He Z., Lu X., Bukai A., Qin S., Duan R., Guli S., Zhang P., Xiao M., Jing H., Wang X. Follow-up of antibody changes in brucellosis patients in Gansu, China. Microbiol. Spectr. 2025 doi: 10.1128/spectrum.02862-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Tang D., Bukai A., Qin S., Duan R., Lyu D., He Z., Zheng X., Wu W., Liang J., Qu H., Bieke A., Zhang P., Zhang D., Han H., Duan Q., Jing H., Wang X. Maternal F1 antibodies and cytokines in mother-neonate dog pairs in the Marmota himalayana plague focus. Heliyon. 2025;11 doi: 10.1016/j.heliyon.2025.e42336. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Mizel S.B., Graff A.H., Sriranganathan N., Ervin S., Lees C.J., Lively M.O., Hantgan R.R., Thomas M.J., Wood J., Bell B. Flagellin-F1-V fusion protein is an effective plague vaccine in Mice and two species of nonhuman Primates. Clin. Vaccine Immunol. 2009;16:21–28. doi: 10.1128/CVI.00333-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Shen L., Jiang C., Weng F., Sun M., Zhao C., Fu T., An C., Shao Z., Liu K. Spatiotemporal risk of human brucellosis under intensification of livestock keeping based on machine learning techniques in Shaanxi, China. Epidemiol. Infect. 2024;152 doi: 10.1017/S0950268824001018. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Khalafalla A.I. Zoonotic diseases transmitted from the camels. Front. Vet. Sci. 2023;10 doi: 10.3389/fvets.2023.1244833. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Rajerison M., Andrianaivoarimanana V., Ratsitorahina M., Rahelinirina S., Chanteau S., Telfer S., Rahalison L. Field assessment of dog as sentinel animal for plague in endemic foci of Madagascar. Integr. Zool. 2021;16:886–892. doi: 10.1111/1749-4877.12541. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Djokic V., Freddi L., de Massis F., Lahti E., van den Esker M.H., Whatmore A., Haughey A., Ferreira A.C., Garofolo G., Melzer F., Sacchini F., Koets A., Wyllie S., Fontbonne A., Girault G., Vicente A.F., McGiven J., Ponsart C. The emergence of Brucella canis as a public health threat in Europe: what we know and what we need to learn. Emerg. Microbes Infect. 2023;12 doi: 10.1080/22221751.2023.2249126. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Deqiu S., Donglou X., Jiming Y. Epidemiology and control of brucellosis in China. Vet. Microbiol. 2002;90:165–182. doi: 10.1016/s0378-1135(02)00252-3. [DOI] [PubMed] [Google Scholar]
49.Alirol E., Getaz L., Stoll B., Chappuis F., Loutan L. Urbanisation and infectious diseases in a globalised world. Lancet Infect. Dis. 2011;11:131–141. doi: 10.1016/S1473-3099(10)70223-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.An C., Shen L., Sun M., Sun Y., Fan S., Zhao C., Nie S., Luo B., Fu T., Liu K., Shao Z., Chang W. Exploring risk transfer of human brucellosis in the context of livestock agriculture transition: a case study in Shaanxi, China. Front. Public Health. 2022;10 doi: 10.3389/fpubh.2022.1009854. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Guo Y., Zhang T. Monitoring and analysis of brucellosis epidemic in Yanchi County, Ningxia from 2010 to 2014. J. Ningxia Med. Univ. 2015;37(09):1078–1080. 10.16050/j.cnki.issn1674-6309.2015.09.024. (In Chinese) [Google Scholar]
52.Zhang Q., Wang L., Liu J. Research on ecological security evaluation of typical agricultural and animal husbandry interlaced areas – a case study of Yanchi County of Ningxia Hui Autonomous Region, China. Pol. J. Environ. Stud. 2025 doi: 10.15244/pjoes/152881. [DOI] [Google Scholar]
53.Hu Y., Li B., Zhang Z., Wang J. Farm size and agricultural technology progress: evidence from China. J. Rural. Stud. 2022;93:417–429. doi: 10.1016/j.jrurstud.2019.01.009. [DOI] [Google Scholar]
54.Lou P., Wang L., Zhang X., Xu J., Wang K. Modelling seasonal brucellosis epidemics in Bayingolin Mongol autonomous prefecture of Xinjiang, China, 2010-2014. Biomed. Res. Int. 2016;2016 doi: 10.1155/2016/5103718. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Lu L., Yang T., Chen Z., Ge Q., Yang J., Sen G. Prediction analysis of human brucellosis cases in Ili Kazakh autonomous prefecture Xinjiang China based on time series. Sci. Rep. 2025;15:1232. doi: 10.1038/s41598-024-80513-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

R code

mmc1.docx^{(41.6KB, docx)}

Study population characteristics, variable definitions, and additional spatial modeling results

mmc2.docx^{(384.5KB, docx)}

Data Availability Statement

[bb0005] 1.Du H.-W., Wang Y., Zhuang D.-F., Jiang X.-S. Temporal and spatial distribution characteristics in the natural plague foci of Chinese Mongolian gerbils based on spatial autocorrelation. Infect. Dis. Poverty. 2017;6:124. doi: 10.1186/s40249-017-0338-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0010] 2.Zheng L., Zhang Q., Zhang A., Hussain H.A., Liu X., Yang Z. Spatiotemporal characteristics of the bearing capacity of cropland based on manure nitrogen and phosphorus load in mainland China. J. Clean. Prod. 2019;233:601–610. doi: 10.1016/j.jclepro.2019.06.049. [DOI] [Google Scholar]

[bb0015] 3.Cheng M., Quan J., Yin J., Liu X., Yuan Z., Ma L. High-resolution maps of intensive and extensive livestock production in China. Resour. Environ. Sustain. 2023;12 doi: 10.1016/j.resenv.2022.100104. [DOI] [Google Scholar]

[bb0020] 4.Yunlong M.A., Ying L.I.U., Xiaolong L.I., Lina S., Tong Y., Xiaoxia M.A., Changsheng Z., Yu Z. Spatial-temporal distribution and influencing factors of Brucellosis in China. zhjbkzzz. 2023;27:1241–1246. doi: 10.16462/j.cnki.zhjbkz.2023.11.001. [DOI] [Google Scholar]

[bb0025] 5.Ben-Ari T., Neerinckx S., Agier L., Cazelles B., Xu L., Zhang Z., Fang X., Wang S., Liu Q., Stenseth N.C. Identification of Chinese plague foci from long-term epidemiological data. Proc. Natl. Acad. Sci. USA. 2012;109:8196–8201. doi: 10.1073/pnas.1110585109. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0030] 6.Williams W. The Anatomy of the Mongolian Gerbil (Meriones unguiculatus) 1974. https://www.semanticscholar.org/paper/The-anatomy-of-the-Mongolian-gerbil-(Meriones-Williams/ce0bf69a70dcc0c08472ffddefec8407a257f9f8 (accessed May 30, 2025)

[bb0035] 7.Liu B., Zhang D., Chen Y., He Z., Liu J., Lyu D., Wu W., Duan R., Qin S., Liang J., Jing H., Wang X. Epidemiological characteristics of plague in the Meriones unguiculatus plague focus - Inner Mongolia Autonomous Region, China, 1950-2019. China CDC Wkly. 2020;2:935–945. doi: 10.46234/ccdcw2020.256. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0040] 8.National Disease Control and Prevention Administration Government Information Disclosure. 2025. https://www.ndcpa.gov.cn/jbkzzx/c100028/common/list.html Accessed May 30, 2025. (accessed May 30, 2025)

[bb0045] 9.Zhang W.-Y., Guo W.-D., Sun S.-H., Jiang J.-F., Sun H.-L., Li S.-L., Fang L.-Q., Cao W.-C. Human Brucellosis, Inner Mongolia, China. Emerg. Infect. Dis. 2010;16(12) doi: 10.3201/eid1612.091081. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0050] 10.Tao Z., Chen Q., Chen Y., Li Y., Mu D., Yang H., Yin W. Epidemiological characteristics of human brucellosis - China, 2016-2019. China CDC Wkly. 2021;3:114–119. doi: 10.46234/ccdcw2021.030. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0055] 11.Wang X. Analysis of the epidemiological characteristics and trends of human brucellosis in Ordos City from 2005 to 2016. Chin. J. Ctrl. Endem. Dis. 2019;34(01):66–69. (In Chinese) [Google Scholar]

[bb0060] 12.Gao M. Analysis of epidemiological characteristics of human brucellosis in Ordos City from 2019 to 2023. Chin. J. Ctrl. Endem. Dis. 2024;39(06):479–480. (In Chinese) [Google Scholar]

[bb0065] 13.Peng C., Li Y.-J., Huang D.-S., Guan P. Spatial-temporal distribution of human brucellosis in mainland China from 2004 to 2017 and an analysis of social and environmental factors. Environ. Health Prev. Med. 2020;25:1. doi: 10.1186/s12199-019-0839-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0070] 14.Song J., Han S., Wang H.H. Spatio-temporal aggregation analysis of human brucellosis in Inner Mongolia Autonomous Region from 2018 to 2022. Chin. J. Zoonoses. 2023;39(10):993–997. (In Chinese) [Google Scholar]

[bb0075] 15.Negrón M.E., Kharod G.A., Bower W.A., Walke H. Notes from the field: human Brucella abortus RB51 infections caused by consumption of unpasteurized domestic dairy products - United States, 2017-2019. MMWR Morb. Mortal Wkly. Rep. 2019;68:185. doi: 10.15585/mmwr.mm6807a6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0080] 16.Samadi A., Amiri M., Hailat N. The reasons behind long-term endemicity of brucellosis in low and middle-income countries: challenges and future perspectives. Curr. Microbiol. 2024;81:82. doi: 10.1007/s00284-023-03605-5. [DOI] [PubMed] [Google Scholar]

[bb0085] 17.Khoshnood S., Pakzad R., Koupaei M., Shirani M., Araghi A., Irani G.M., Moradi M., Pakzad I., Sadeghifard N., Heidary M. Prevalence, diagnosis, and manifestations of brucellosis: a systematic review and meta-analysis. Front. Vet. Sci. 2022;9 doi: 10.3389/fvets.2022.976215. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0090] 18.He Z., Wei B., Zhang Y., Liu J., Xi J., Ciren D., Qi T., Liang J., Duan R., Qin S., Lv D., Chen Y., Xiao M., Fan R., Song Z., Jing H., Wang X. Distribution and characteristics of human plague cases and Yersinia pestis isolates from 4 Marmota plague foci, China, 1950-2019. Emerg. Infect. Dis. 2021;27:2544–2553. doi: 10.3201/eid2710.202239. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0095] 19.Qin S., Liang J., Tang D., Chen Y., Duan R., Lu X., Bukai A., Zheng X., Lv D., He Z., Wu W., Han H., Jing H., Wang X. Serological investigation of plague and brucellosis infection in Marmota himalayana plague foci in the Altun Mountains on the Qinghai-Tibet plateau. Front. Public Health. 2022;10 doi: 10.3389/fpubh.2022.990218. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0100] 20.Bin Saeed A.A., Al-Hamdan N.A., Fontaine R.E. Plague from eating raw camel liver. Emerg. Infect. Dis. 2005;11:1456–1457. doi: 10.3201/eid1109.050081. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0105] 21.Christie A.B., Chen T.H., Elberg S.S. Plague in camels and goats: their role in human epidemics. J. Infect. Dis. 1980;141:724–726. doi: 10.1093/infdis/141.6.724. [DOI] [PubMed] [Google Scholar]

[bb0110] 22.Fedorov V.N. Plague in camels and its prevention in the USSR. Bull. World Health Organ. 1960;23:275–281. [PMC free article] [PubMed] [Google Scholar]

[bb0115] 23.Dai R., Qi M., Xiong H., Yang X., He J., Zhang Z., Yang H., Jin J., Li X., Xin Y., Yang Y., Li C., Li Z., Xu J., Wang Z., Li W., Wei B. Serological epidemiological investigation of Tibetan sheep (Ovis aries) plague in Qinghai, China. Vector Borne Zoonotic Dis. 2019;19:3–7. doi: 10.1089/vbz.2017.2257. [DOI] [PubMed] [Google Scholar]

[bb0120] 24.Gibb R., Redding D.W., Chin K.Q., Donnelly C.A., Blackburn T.M., Newbold T., Jones K.E. Zoonotic host diversity increases in human-dominated ecosystems. Nature. 2020;584:398–402. doi: 10.1038/s41586-020-2562-8. [DOI] [PubMed] [Google Scholar]

[bb0125] 25.Jones B.A., Grace D., Kock R., Alonso S., Rushton J., Said M.Y., McKeever D., Mutua F., Young J., McDermott J., Pfeiffer D.U. Zoonosis emergence linked to agricultural intensification and environmental change. Proc. Natl. Acad. Sci. USA. 2013;110:8399–8404. doi: 10.1073/pnas.1208059110. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0130] 26.Murray G.G.R., Hossain A.S.M.M., Miller E.L., Bruchmann S., Balmer A.J., Matuszewska M., Herbert J., Hadjirin N.F., Mugabi R., Li G., Ferrando M.L., de Oliveira I.M. Fernandes, Nguyen T., Yen P.L.K., Phuc H.D., Moe A. Zaw, Wai T. Su, Gottschalk M., Aragon V., Valentin-Weigand P., Heegaard P.M.H., Vrieling M., Maw M. Thein, Myint H. Thidar, Win Y. Tun, Hoa N. Thi, Bentley S.D., Clavijo M.J., Wells J.M., Tucker A.W., Weinert L.A. The emergence and diversification of a zoonotic pathogen from within the microbiota of intensively farmed pigs. Proc. Natl. Acad. Sci. USA. 2023;120 doi: 10.1073/pnas.2307773120. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0135] 27.White R.J., Razgour O. Emerging zoonotic diseases originating in mammals: a systematic review of effects of anthropogenic land-use change. Mammal Rev. 2020;50:336–352. doi: 10.1111/mam.12201. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0140] 28.Ministry of Health of China WS 279–2008 Diagnostic Criteria for Plague. 2025. https://icdc.chinacdc.cn/zcfgybz/bz/202112/t20211202_253341.html Accessed May 31, 2025.

[bb0145] 29.Miller J., Williamson E.D., Lakey J.H., Pearce M.J., Jones S.M., Titball R.W. Macromolecular organisation of recombinant Yersinia pestis F1 antigen and the effect of structure on immunogenicity. FEMS Immunol. Med. Microbiol. 1998;21:213–221. doi: 10.1016/S0928-8244(98)00063-7. [DOI] [PubMed] [Google Scholar]

[bb0150] 30.Ma X., Sun G.-Q. Global dynamics of a periodic brucellosis model with time delay and environmental factors. Appl. Math. Model. 2024;130:288–309. doi: 10.1016/j.apm.2024.03.002. [DOI] [Google Scholar]

[bb0155] 31.Helbich M., Arsanjani J.J. Spatial eigenvector filtering for spatiotemporal crime mapping and spatial crime analysis. Cartogr. Geogr. Inf. Sci. 2015 doi: 10.1080/15230406.2014.893839. (accessed May 6, 2025) [DOI] [Google Scholar]

[bb0160] 32.Engebretsen S., Bohlin J. Statistical predictions with glmnet. Clin. Epigenetics. 2019;11:123. doi: 10.1186/s13148-019-0730-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0165] 33.LeSage J., Pace R.K. Chapman and Hall/CRC; New York: 2009. Introduction to Spatial Econometrics. [DOI] [Google Scholar]

[bb0170] 34.Lin X. Identifying peer effects in student academic achievement by spatial autoregressive models with group unobservables. J. Labor Econ. 2010 doi: 10.1086/653506. [DOI] [Google Scholar]

[bb0175] 35.Wang J.-F., Li X.-H., Christakos G., Liao Y.-L., Zhang T., Gu X., Zheng X.-Y. Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China. Int. J. Geogr. Inf. Sci. 2010 doi: 10.1080/13658810802443457. [DOI] [Google Scholar]

[bb0180] 36.Murakami D. 2017. Spmoran: An R Package for Moran's Eigenvector-Based Spatial Regression Analysis. [DOI] [Google Scholar]

[bb0185] 37.Ning C., Shuyi G., Tao Y., Hao Z., Zhang X. Epidemiological survey of human brucellosis in Inner Mongolia, China, 2010-2014: a high risk groups-based survey. J. Infect. Public Health. 2018;11:24–29. doi: 10.1016/j.jiph.2017.02.013. [DOI] [PubMed] [Google Scholar]

[bb0190] 38.Liu Z.-G., Wang M., Ta N., Fang M.-G., Mi J.-C., Yu R.-P., Luo Y., Cao X., Li Z.-J. Seroprevalence of human brucellosis and molecular characteristics of Brucella strains in Inner Mongolia Autonomous region of China, from 2012 to 2016. Emerg. Microbes Infect. 2020;9:263–274. doi: 10.1080/22221751.2020.1720528. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0195] 39.Li K., Wang Z., Wei X., Ji H., Shang M., Chang N., Wang Z., Guo C., Xu L., Zhao N., Liu Q. Relationships between meteorological factors and Mongolian gerbils and its flea burdens - Xilingol league, Inner Mongolia Autonomous Region, China, 2012-2021. China CDC Wkly. 2024;6:547–552. doi: 10.46234/ccdcw2024.107. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0200] 40.Coghlan J.D., Weir D.M. Antibodies in human brucellosis. Br Med J. 1967;2 doi: 10.1136/bmj.2.5547.269. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0205] 41.Sha H., Duan Q., Lyu D., Qian F., Zheng X., Guo J., He Z., Lu X., Bukai A., Qin S., Duan R., Guli S., Zhang P., Xiao M., Jing H., Wang X. Follow-up of antibody changes in brucellosis patients in Gansu, China. Microbiol. Spectr. 2025 doi: 10.1128/spectrum.02862-24. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0210] 42.Tang D., Bukai A., Qin S., Duan R., Lyu D., He Z., Zheng X., Wu W., Liang J., Qu H., Bieke A., Zhang P., Zhang D., Han H., Duan Q., Jing H., Wang X. Maternal F1 antibodies and cytokines in mother-neonate dog pairs in the Marmota himalayana plague focus. Heliyon. 2025;11 doi: 10.1016/j.heliyon.2025.e42336. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0215] 43.Mizel S.B., Graff A.H., Sriranganathan N., Ervin S., Lees C.J., Lively M.O., Hantgan R.R., Thomas M.J., Wood J., Bell B. Flagellin-F1-V fusion protein is an effective plague vaccine in Mice and two species of nonhuman Primates. Clin. Vaccine Immunol. 2009;16:21–28. doi: 10.1128/CVI.00333-08. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0220] 44.Shen L., Jiang C., Weng F., Sun M., Zhao C., Fu T., An C., Shao Z., Liu K. Spatiotemporal risk of human brucellosis under intensification of livestock keeping based on machine learning techniques in Shaanxi, China. Epidemiol. Infect. 2024;152 doi: 10.1017/S0950268824001018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0225] 45.Khalafalla A.I. Zoonotic diseases transmitted from the camels. Front. Vet. Sci. 2023;10 doi: 10.3389/fvets.2023.1244833. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0230] 46.Rajerison M., Andrianaivoarimanana V., Ratsitorahina M., Rahelinirina S., Chanteau S., Telfer S., Rahalison L. Field assessment of dog as sentinel animal for plague in endemic foci of Madagascar. Integr. Zool. 2021;16:886–892. doi: 10.1111/1749-4877.12541. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0235] 47.Djokic V., Freddi L., de Massis F., Lahti E., van den Esker M.H., Whatmore A., Haughey A., Ferreira A.C., Garofolo G., Melzer F., Sacchini F., Koets A., Wyllie S., Fontbonne A., Girault G., Vicente A.F., McGiven J., Ponsart C. The emergence of Brucella canis as a public health threat in Europe: what we know and what we need to learn. Emerg. Microbes Infect. 2023;12 doi: 10.1080/22221751.2023.2249126. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0240] 48.Deqiu S., Donglou X., Jiming Y. Epidemiology and control of brucellosis in China. Vet. Microbiol. 2002;90:165–182. doi: 10.1016/s0378-1135(02)00252-3. [DOI] [PubMed] [Google Scholar]

[bb0245] 49.Alirol E., Getaz L., Stoll B., Chappuis F., Loutan L. Urbanisation and infectious diseases in a globalised world. Lancet Infect. Dis. 2011;11:131–141. doi: 10.1016/S1473-3099(10)70223-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0250] 50.An C., Shen L., Sun M., Sun Y., Fan S., Zhao C., Nie S., Luo B., Fu T., Liu K., Shao Z., Chang W. Exploring risk transfer of human brucellosis in the context of livestock agriculture transition: a case study in Shaanxi, China. Front. Public Health. 2022;10 doi: 10.3389/fpubh.2022.1009854. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0255] 51.Guo Y., Zhang T. Monitoring and analysis of brucellosis epidemic in Yanchi County, Ningxia from 2010 to 2014. J. Ningxia Med. Univ. 2015;37(09):1078–1080. 10.16050/j.cnki.issn1674-6309.2015.09.024. (In Chinese) [Google Scholar]

[bb0260] 52.Zhang Q., Wang L., Liu J. Research on ecological security evaluation of typical agricultural and animal husbandry interlaced areas – a case study of Yanchi County of Ningxia Hui Autonomous Region, China. Pol. J. Environ. Stud. 2025 doi: 10.15244/pjoes/152881. [DOI] [Google Scholar]

[bb0265] 53.Hu Y., Li B., Zhang Z., Wang J. Farm size and agricultural technology progress: evidence from China. J. Rural. Stud. 2022;93:417–429. doi: 10.1016/j.jrurstud.2019.01.009. [DOI] [Google Scholar]

[bb0270] 54.Lou P., Wang L., Zhang X., Xu J., Wang K. Modelling seasonal brucellosis epidemics in Bayingolin Mongol autonomous prefecture of Xinjiang, China, 2010-2014. Biomed. Res. Int. 2016;2016 doi: 10.1155/2016/5103718. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bb0275] 55.Lu L., Yang T., Chen Z., Ge Q., Yang J., Sen G. Prediction analysis of human brucellosis cases in Ili Kazakh autonomous prefecture Xinjiang China based on time series. Sci. Rep. 2025;15:1232. doi: 10.1038/s41598-024-80513-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Spatial characteristics and driving factors of human brucellosis and plague infections in the Meriones unguiculatus plague focus in China

Dongyue Lyu

Jindong Zuo

Shuai Qin

Liping Wang

Ziyi Bu

Hanyu Sha

Lingjie Bai

Ran Duan

Linxuan Yang

Zhengliang Chai

Meng Xiao

Zhaokai He

Deming Tang

Peng Zhang

Huaiqi Jing

Xin Wang

Abstract

Graphical abstract

1. Introduction

2. Materials and methods

2.1. Research area

Fig. 1.

2.2. Sample collection and data collection

2.2.1. Sample collection

2.2.2. Background data collection

2.3. Serological testing methods

2.3.1. Serological testing for Brucella

2.3.2. Serological testing for Y. pestis

2.4. Statistical analysis

2.4.1. Data preprocessing

2.4.1.1. Demographic variable handling

2.4.1.2. Management of repeated samples

2.4.1.3. Spatial data standardization

2.4.1.4. Temporal dimension handling

2.4.1.5. Variable transformation and encoding

2.4.1.6. Quality control

2.4.2. Analysis of population characteristics and individual serum antibody titers

2.4.2.1. Analysis of population characteristics

2.4.2.2. Analysis of individual serum antibodies titers

2.4.3. Detection of spatial clustering

2.4.4. Core analytical framework: a spatially enhanced chained random forest model

2.4.4.1. Spatial feature extraction and selection

2.4.4.2. Variable screening and Preprocessing

2.4.4.3. Chained random forest modeling and training

2.4.4.4. Model performance and uncertainty assessment

2.4.5. Risk mapping and hotspot prediction

2.4.6. Analytical software

3. Results

3.1. Epidemiological status of brucellosis and plague in the Meriones unguiculatus plague focus

Fig. 2.

3.2. Changes in antibody titers to brucellosis and plague

3.3. Spatial analysis

Fig. 3.

3.4. Multi-model analysis of infection risk factors

3.4.1. Inclusion variable screening

Fig. 4.

3.4.2. Spatial eigen-feature extraction

3.4.3. Optimal model selection

3.4.4. Combined ranking of important factors

3.4.5. Interaction of important factors

3.5. Predictive analysis

Fig. 5.

4. Discussion

CRediT authorship contribution statement

Ethics approval statement

Code availability

Declaration of competing interest

Acknowledgment

Footnotes

Appendix A. Supplementary data

Data availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES