Melioidosis, an infectious disease caused by Burkholderia pseudomallei, poses significant public health challenges, particularly in regions where specific environmental factors play crucial roles in its spread. However, traditional risk assessment methods for melioidosis do not comprehensively incorporate the diverse environmental factors that influence the distribution of this bacteria. This paper presents a spatiotemporal analysis of melioidosis transmission in Ubon Ratchathani, Thailand, through a comparative evaluation of extreme gradient boosting (XGBoost) and TabNet models. To model the disease distribution over spatiotemporal scales, various environmental datasets were integrated, including land surface temperature, normalized difference vegetation index, normalized difference water index, and rainfall data. The models were trained and validated on data spanning from January 1, 2013, to December 31, 2022, which were obtained from 219 subdistricts. Our comparative analysis of the two models showed that TabNet outperformed XGBoost, particularly in capturing complex interactions between environmental variables and melioidosis cases, and achieved a higher accuracy score (0.950 for TabNet versus 0.892 for XGBoost). While both models performed similarly in terms of the area under the receiver operating characteristic curve, TabNet exhibited marginally more variability. These results underscore the importance of environmental data for refining predictive models that are used for melioidosis surveillance and management.
{"title":"Comparative analysis of extreme gradient boosting and TabNet models for spatiotemporal prediction of melioidosis using satellite-derived environmental data","authors":"Jaruwan Wongbutdee , Wacharapong Saengnill , Pongthep Thongsang","doi":"10.1016/j.sste.2025.100767","DOIUrl":"10.1016/j.sste.2025.100767","url":null,"abstract":"<div><div>Melioidosis, an infectious disease caused by <em>Burkholderia pseudomallei</em>, poses significant public health challenges, particularly in regions where specific environmental factors play crucial roles in its spread. However, traditional risk assessment methods for melioidosis do not comprehensively incorporate the diverse environmental factors that influence the distribution of this bacteria. This paper presents a spatiotemporal analysis of melioidosis transmission in Ubon Ratchathani, Thailand, through a comparative evaluation of extreme gradient boosting (XGBoost) and TabNet models. To model the disease distribution over spatiotemporal scales, various environmental datasets were integrated, including land surface temperature, normalized difference vegetation index, normalized difference water index, and rainfall data. The models were trained and validated on data spanning from January 1, 2013, to December 31, 2022, which were obtained from 219 subdistricts. Our comparative analysis of the two models showed that TabNet outperformed XGBoost, particularly in capturing complex interactions between environmental variables and melioidosis cases, and achieved a higher accuracy score (0.950 for TabNet versus 0.892 for XGBoost). While both models performed similarly in terms of the area under the receiver operating characteristic curve, TabNet exhibited marginally more variability. These results underscore the importance of environmental data for refining predictive models that are used for melioidosis surveillance and management.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100767"},"PeriodicalIF":1.7,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568717","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-01DOI: 10.1016/j.sste.2025.100765
Hélder Seixas Lima , Petrônio Cândido de Lima e Silva , Wagner Meira Jr. , Unaí Tupinambás , Marcelo Azevedo Costa , Frederico Gadelha Guimarães
The COVID-19 pandemic profoundly impacted Brazil, a country marked by significant socioeconomic and political disparities among its municipalities and regions. This study conducts an ecological analysis at the municipal level to examine sociodemographic and political variables correlating with COVID-19 mortality over the first three years of the pandemic (2020–2022). Employing the Gaussian Mixture algorithm, we clusterized the Brazilian municipalities into five sociodemographic clusters through a soft assignment. Negative binomial regression models were then applied to estimate the correlations between explanatory variables and age-sex standardized COVID-19 deaths in Brazilian municipalities. Our findings reveal that municipalities with lower human development indices experienced higher COVID-19 mortality rates in the early stages of the pandemic. As the pandemic progressed, the highest mortality rates shifted to municipalities with higher urbanization (Rate Ratios (RR) = 1.13, 95% Confidence Intervals (CI): 1.10–1.15), indicating this as the sociodemographic variable with the strongest correlations with COVID-19 mortality. This work also reveals that the variable investigated that reported the strongest correlation was the percentage of votes for Jair Bolsonaro in the 2022 Presidential Election (RR = 1.21, 95% CI: 1.19–1.23). This work highlights the importance of equipping health authorities and policymakers with methods to monitor future epidemics, emphasizing the need to address urbanization and poverty-related vulnerabilities, provide targeted support for specific populations, and combat misinformation to protect at-risk groups such as the aged.
{"title":"Sociodemographic and political factors associated with COVID-19 mortality in Brazilian municipalities across three years: An approach supported by Gaussian Mixture clustering","authors":"Hélder Seixas Lima , Petrônio Cândido de Lima e Silva , Wagner Meira Jr. , Unaí Tupinambás , Marcelo Azevedo Costa , Frederico Gadelha Guimarães","doi":"10.1016/j.sste.2025.100765","DOIUrl":"10.1016/j.sste.2025.100765","url":null,"abstract":"<div><div>The COVID-19 pandemic profoundly impacted Brazil, a country marked by significant socioeconomic and political disparities among its municipalities and regions. This study conducts an ecological analysis at the municipal level to examine sociodemographic and political variables correlating with COVID-19 mortality over the first three years of the pandemic (2020–2022). Employing the Gaussian Mixture algorithm, we clusterized the Brazilian municipalities into five sociodemographic clusters through a soft assignment. Negative binomial regression models were then applied to estimate the correlations between explanatory variables and age-sex standardized COVID-19 deaths in Brazilian municipalities. Our findings reveal that municipalities with lower human development indices experienced higher COVID-19 mortality rates in the early stages of the pandemic. As the pandemic progressed, the highest mortality rates shifted to municipalities with higher urbanization (Rate Ratios (RR) = 1.13, 95% Confidence Intervals (CI): 1.10–1.15), indicating this as the sociodemographic variable with the strongest correlations with COVID-19 mortality. This work also reveals that the variable investigated that reported the strongest correlation was the percentage of votes for Jair Bolsonaro in the 2022 Presidential Election (RR = 1.21, 95% CI: 1.19–1.23). This work highlights the importance of equipping health authorities and policymakers with methods to monitor future epidemics, emphasizing the need to address urbanization and poverty-related vulnerabilities, provide targeted support for specific populations, and combat misinformation to protect at-risk groups such as the aged.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100765"},"PeriodicalIF":1.7,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145519863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-01DOI: 10.1016/j.sste.2025.100764
Bibian N. Robert , Peter M. Macharia , M. Naser Lessani , Viola Chepkurui , Joseph Kamau , Robert W. Snow , Zhenlong Li , Emelda A. Okiro
Background
Everyone deserves legal recognition, yet millions of children remain unregistered, with the majority (87%) residing in sub-Saharan Africa and southern Asia. Despite global efforts to improve birth registration coverage, sub-national disparities persist. Across Kenya's 47 counties, birth registration completeness rates varies from nearly 100% to as low as 12.2%, suggesting local contextual factors are important. This study explores the influence of contextual factors on the spatially heterogeneous rates of birth registration in Kenya.
Methods
We utilized data from the 2022 Kenya Demographic and Health Survey. The association between registered births and its determinants (child factors, health care indicators, maternal, household and geographical factors) was assessed at the cluster level (villages) using four regression models: ordinary least square (OLS) and spatial local regression using Geographically Weighted Regression (GWR-single spatial scale for all predictors), Multiscale GWR (MGWR-each predictor operates at different spatial scale) and Similarity GWR (SGWR-single spatial scale for all predictors) models. Best-fit models were assessed using adjusted R2, AICc and Moran’s I (residual spatial autocorrelation). The key difference between GWR and SGWR lies in how spatial dependency is measured between locations.
Results
A total of 1673 survey clusters were analysed. MGWR was the best-fitting model (AICc = 14,870.57, adjusted R2 = 0.40, Moran’s I = -0.04 (p-value = 0.999)) and identified localised significant relationships for all variables examined. Evidence of spatially varying relationship (local influence) was observed between birth registration, bank account ownership, and unemployment. Regional influence was observed for female-headed households, while other associations maintained a uniform relationship across the study area (global influence).
Conclusion
Determinants of birth registration vary spatially at different geographical scales, necessitating context-specific targeted strategies to boost registration coverage across diverse areas and populations.
{"title":"Spatially varying relationships between birth registration and influencing factors in Kenya, using a suite of geographically weighted regressions","authors":"Bibian N. Robert , Peter M. Macharia , M. Naser Lessani , Viola Chepkurui , Joseph Kamau , Robert W. Snow , Zhenlong Li , Emelda A. Okiro","doi":"10.1016/j.sste.2025.100764","DOIUrl":"10.1016/j.sste.2025.100764","url":null,"abstract":"<div><h3>Background</h3><div>Everyone deserves legal recognition, yet millions of children remain unregistered, with the majority (87%) residing in sub-Saharan Africa and southern Asia. Despite global efforts to improve birth registration coverage, sub-national disparities persist. Across Kenya's 47 counties, birth registration completeness rates varies from nearly 100% to as low as 12.2%, suggesting local contextual factors are important. This study explores the influence of contextual factors on the spatially heterogeneous rates of birth registration in Kenya.</div></div><div><h3>Methods</h3><div>We utilized data from the 2022 Kenya Demographic and Health Survey. The association between registered births and its determinants (child factors, health care indicators, maternal, household and geographical factors) was assessed at the cluster level (villages) using four regression models: ordinary least square (OLS) and spatial local regression using Geographically Weighted Regression (GWR-single spatial scale for all predictors), Multiscale GWR (MGWR-each predictor operates at different spatial scale) and Similarity GWR (SGWR-single spatial scale for all predictors) models. Best-fit models were assessed using adjusted R<sup>2</sup>, AICc and Moran’s I (residual spatial autocorrelation). The key difference between GWR and SGWR lies in how spatial dependency is measured between locations.</div></div><div><h3>Results</h3><div>A total of 1673 survey clusters were analysed. MGWR was the best-fitting model (AICc = 14,870.57, adjusted R<sup>2</sup> = 0.40, Moran’s <em>I</em> = -0.04 (p-value = 0.999)) and identified localised significant relationships for all variables examined. Evidence of spatially varying relationship (local influence) was observed between birth registration, bank account ownership, and unemployment. Regional influence was observed for female-headed households, while other associations maintained a uniform relationship across the study area (global influence).</div></div><div><h3>Conclusion</h3><div>Determinants of birth registration vary spatially at different geographical scales, necessitating context-specific targeted strategies to boost registration coverage across diverse areas and populations.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100764"},"PeriodicalIF":1.7,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145519864","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Thailand has achieved leprosy elimination at the national level. However, sporadic cases still occur in rural areas, especially in the southern provinces. This study investigates the spatiotemporal distribution and burden of leprosy in Health Region 12 over 25 years, highlighting high-burden districts.
Methods
A retrospective, population-based cross-sectional study was conducted using leprosy surveillance data from 1999 to 2023 across 77 districts in seven southern provinces. A two-step analytical approach was used to analyse leprosy case counts classified by gender-age group, year, and district. The two-step analytic approach involves separately fitting a logistic model for leprosy occurrence and a log-linear regression model for leprosy incidence without zeros, and the results were combined.
Results
A total of 1233 new leprosy cases were reported, with a median age of 41 years and a predominance of multibacillary cases (73.2 %). Males accounted for 64 % of cases. Leprosy incidence increased with age, peaking among individuals aged 70 years and over. Leprosy occurrence and incidence rates are on a decreasing trend. Four districts in Pattani and seven districts in Narathiwat were identified as high-burden areas, characterised by above-average occurrence and incidence rates.
Conclusion
Although Thailand has achieved leprosy eradication on the national level, some districts in southernmost provinces have not to achieve leprosy elimination. These findings highlight the need for intensified surveillance and targeted interventions at the subnational level.
{"title":"Spatiotemporal Patterns of Leprosy in Southern Thailand: Identifying High-burden Districts Over 25 Years","authors":"Dasseema Muwannasing , Benjamin Atta Owusu , Phattrawan Tongkumchum , Nitinun Pongsiri , Lumpoo Ammatawiyanon , Penpicha Poolsawat","doi":"10.1016/j.sste.2025.100766","DOIUrl":"10.1016/j.sste.2025.100766","url":null,"abstract":"<div><h3>Background</h3><div>Thailand has achieved leprosy elimination at the national level. However, sporadic cases still occur in rural areas, especially in the southern provinces. This study investigates the spatiotemporal distribution and burden of leprosy in Health Region 12 over 25 years, highlighting high-burden districts.</div></div><div><h3>Methods</h3><div>A retrospective, population-based cross-sectional study was conducted using leprosy surveillance data from 1999 to 2023 across 77 districts in seven southern provinces. A two-step analytical approach was used to analyse leprosy case counts classified by gender-age group, year, and district. The two-step analytic approach involves separately fitting a logistic model for leprosy occurrence and a log-linear regression model for leprosy incidence without zeros, and the results were combined.</div></div><div><h3>Results</h3><div>A total of 1233 new leprosy cases were reported, with a median age of 41 years and a predominance of multibacillary cases (73.2 %). Males accounted for 64 % of cases. Leprosy incidence increased with age, peaking among individuals aged 70 years and over. Leprosy occurrence and incidence rates are on a decreasing trend. Four districts in Pattani and seven districts in Narathiwat were identified as high-burden areas, characterised by above-average occurrence and incidence rates.</div></div><div><h3>Conclusion</h3><div>Although Thailand has achieved leprosy eradication on the national level, some districts in southernmost provinces have not to achieve leprosy elimination. These findings highlight the need for intensified surveillance and targeted interventions at the subnational level.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100766"},"PeriodicalIF":1.7,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568716","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-01DOI: 10.1016/j.sste.2025.100761
Hyeji Kwon , Ansun Jeong , Jiwon Kim , Minseong Kang
Background
Busan, a densely populated metropolitan city in South Korea, has seen a continuous increase in 119 emergency calls.
Objectives
This study investigates the spatial distribution of 119 ambulance response times across Busan to identify regional disparities and provide policy recommendations for enhancing emergency medical services (EMS).
Methods
Data from the Busan Fire and Disaster Headquarters for the year 2023 were analyzed, including dispatch times, on-site arrival times, patient contact times, travel distances, and administrative region information. The analysis was conducted at the level of 192 administrative divisions (eup, myeon, and dong) in Busan. Spatial autocorrelation was assessed using Moran’s I to identify regions with longer or shorter response times and travel distances (hot spots and cold spots). T-tests were used to compare demographic and geographic characteristics between hot spot and cold spot areas.
Results
A total of 21 regions (10.9 %) were identified as hot spots for both average response time and distance to incident sites. Disparities tended to widen moving from central Busan toward the outskirts. The average response time in hot spot areas was 12.0 min, compared to 9.8 min in cold spots. The average travel distance to incident sites was 3.5 km in hot spots and 1.3 km in cold spots. Hot spot areas were characterized by larger elderly male population, higher frequencies of emergency patients, and notably more cases of cardiac arrest.
Conclusions
This study confirms spatial disparities in EMS provision across Busan and underscores the need for region-specific policies to improve equitable emergency care access.
{"title":"Regional disparities in 119 Emergency medical services response times in South Korea: A focus on Busan","authors":"Hyeji Kwon , Ansun Jeong , Jiwon Kim , Minseong Kang","doi":"10.1016/j.sste.2025.100761","DOIUrl":"10.1016/j.sste.2025.100761","url":null,"abstract":"<div><h3>Background</h3><div>Busan, a densely populated metropolitan city in South Korea, has seen a continuous increase in 119 emergency calls.</div></div><div><h3>Objectives</h3><div>This study investigates the spatial distribution of 119 ambulance response times across Busan to identify regional disparities and provide policy recommendations for enhancing emergency medical services (EMS).</div></div><div><h3>Methods</h3><div>Data from the Busan Fire and Disaster Headquarters for the year 2023 were analyzed, including dispatch times, on-site arrival times, patient contact times, travel distances, and administrative region information. The analysis was conducted at the level of 192 administrative divisions (eup, myeon, and dong) in Busan. Spatial autocorrelation was assessed using Moran’s I to identify regions with longer or shorter response times and travel distances (hot spots and cold spots). T-tests were used to compare demographic and geographic characteristics between hot spot and cold spot areas.</div></div><div><h3>Results</h3><div>A total of 21 regions (10.9 %) were identified as hot spots for both average response time and distance to incident sites. Disparities tended to widen moving from central Busan toward the outskirts. The average response time in hot spot areas was 12.0 min, compared to 9.8 min in cold spots. The average travel distance to incident sites was 3.5 km in hot spots and 1.3 km in cold spots. Hot spot areas were characterized by larger elderly male population, higher frequencies of emergency patients, and notably more cases of cardiac arrest.</div></div><div><h3>Conclusions</h3><div>This study confirms spatial disparities in EMS provision across Busan and underscores the need for region-specific policies to improve equitable emergency care access.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100761"},"PeriodicalIF":1.7,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145417625","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-13DOI: 10.1016/j.sste.2025.100760
Garyfallos Konstantinoudis , Anthony Hauser , Julien Riou
Extreme heat has been linked to increased mortality. Monitoring the mortality burden attributable to extreme heat is crucial to inform policies, such as heat warnings, and prevent heat-related deaths. In this study, we evaluate excess mortality during summer 2022 in Switzerland, identify vulnerable populations and estimate temperature thresholds relevant for heat adaptation policies. We use nationwide mortality and population data during 2011–2022 by age, sex, day and canton. We develop a Bayesian ensemble modelling approach with dynamic population to predict expected mortality in summer 2022 and calculate excess by comparing expected with observed mortality. We account for covariates associated with mortality such as national holidays, and spatiotemporal random effects to improve predictions. After accounting for the effect of the COVID-19 pandemic, we observed a total of 487 (95% Credible Interval: 10–935) excess deaths during summer 2022 in people older than 80 years. We illustrate that for periods of extreme heat longer than four days, the minimum excess mortality temperature threshold in the oldest age group is the 70th percentile of the temperature. Our approach highlights the importance of lower temperature thresholds during prolonged periods of extreme heat and advocates for integrating this insight in heat adaptation policies.
{"title":"Ensemble Bayesian modelling with dynamic population to estimate excess deaths due to extreme temperatures","authors":"Garyfallos Konstantinoudis , Anthony Hauser , Julien Riou","doi":"10.1016/j.sste.2025.100760","DOIUrl":"10.1016/j.sste.2025.100760","url":null,"abstract":"<div><div>Extreme heat has been linked to increased mortality. Monitoring the mortality burden attributable to extreme heat is crucial to inform policies, such as heat warnings, and prevent heat-related deaths. In this study, we evaluate excess mortality during summer 2022 in Switzerland, identify vulnerable populations and estimate temperature thresholds relevant for heat adaptation policies. We use nationwide mortality and population data during 2011–2022 by age, sex, day and canton. We develop a Bayesian ensemble modelling approach with dynamic population to predict expected mortality in summer 2022 and calculate excess by comparing expected with observed mortality. We account for covariates associated with mortality such as national holidays, and spatiotemporal random effects to improve predictions. After accounting for the effect of the COVID-19 pandemic, we observed a total of 487 (95% Credible Interval: 10–935) excess deaths during summer 2022 in people older than 80 years. We illustrate that for periods of extreme heat longer than four days, the minimum excess mortality temperature threshold in the oldest age group is the 70th percentile of the temperature. Our approach highlights the importance of lower temperature thresholds during prolonged periods of extreme heat and advocates for integrating this insight in heat adaptation policies.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100760"},"PeriodicalIF":1.7,"publicationDate":"2025-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145363633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-08DOI: 10.1016/j.sste.2025.100757
Alex Ziyu Jiang , Jon Wakefield
Prediction is a classic challenge in spatial statistics and the inclusion of spatial covariates can greatly improve predictive performance when incorporated into a model with latent spatial effects. It is desirable to develop flexible regression models that allow for nonlinearities and interactions in the covariate specification. Existing machine learning approaches that allow for spatial dependence in the residuals fail to provide reliable uncertainty estimates. In this paper, we investigate the combination of a Gaussian process spatial model with a Bayesian Additive Regression Tree (BART) model. The computational burden of the approach is reduced by combining Markov chain Monte Carlo (MCMC) with the Integrated Nested Laplace Approximation (INLA) technique. We study the performance of the method first via simulation. We then use the model to predict anthropometric responses in Kenya, with the data collected via a complex sampling design. In particular, household survey data are collected via stratified two-stage unequal probability cluster sampling, which requires special care when modeled.
{"title":"BARTSIMP: Flexible spatial covariate modeling and prediction using Bayesian Additive Regression Trees","authors":"Alex Ziyu Jiang , Jon Wakefield","doi":"10.1016/j.sste.2025.100757","DOIUrl":"10.1016/j.sste.2025.100757","url":null,"abstract":"<div><div>Prediction is a classic challenge in spatial statistics and the inclusion of spatial covariates can greatly improve predictive performance when incorporated into a model with latent spatial effects. It is desirable to develop flexible regression models that allow for nonlinearities and interactions in the covariate specification. Existing machine learning approaches that allow for spatial dependence in the residuals fail to provide reliable uncertainty estimates. In this paper, we investigate the combination of a Gaussian process spatial model with a Bayesian Additive Regression Tree (BART) model. The computational burden of the approach is reduced by combining Markov chain Monte Carlo (MCMC) with the Integrated Nested Laplace Approximation (INLA) technique. We study the performance of the method first via simulation. We then use the model to predict anthropometric responses in Kenya, with the data collected via a complex sampling design. In particular, household survey data are collected via stratified two-stage unequal probability cluster sampling, which requires special care when modeled.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100757"},"PeriodicalIF":1.7,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145325733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-03DOI: 10.1016/j.sste.2025.100758
José Mário Nunes da Silva , Fabrício Dos Santos Menezes , Diego Rodrigues Mendonça e Silva , Tarsila Guimarães Vieira da Silva , Luiz Paulo Kowalski
This study analyzed the temporal trends, spatial and spatio-temporal patterns of OC and OPC mortality in Brazil between 1980 and 2023, and explored their association with socioeconomic inequality. We conducted an ecological study using age- and sex-standardized mortality rates, smoothed via a local empirical Bayesian method. We assessed temporal trends through joinpoint regression. We evaluated global and local spatial autocorrelation and detected spatio-temporal clusters using a retrospective space–time scan statistic based on a Poisson model. We observed a decrease in OC mortality, particularly among men aged 40–59 years in the Southeast and South regions. In contrast, OPC mortality increased throughout the study period in both sexes, especially among individuals aged 60–79 years, with the largest increases occurring in the North, Northeast, and Central-West regions. Moran’s I revealed significant spatial dependence for both cancers. Spatial analyses identified persistent high-risk clusters in the Southeast and South, which expanded toward the Northeast and Central-West. Spatio-temporal analysis showed a recent shift of major OC clusters from the Southeast and South towards the Northeast, whereas OPC clusters continued to expand into the Central-West. Municipalities within clusters characterized by a low Human Development Index exhibited comparatively stronger increases in mortality trends for both cancers. These results underscore the need for more equitable and regionally tailored public policies to strengthen cancer control efforts in Brazil.
{"title":"Spatio-temporal clustering analysis, temporal trends, and inequality in oral and oropharyngeal cancer mortality in Brazil over 44 years (1980–2023)","authors":"José Mário Nunes da Silva , Fabrício Dos Santos Menezes , Diego Rodrigues Mendonça e Silva , Tarsila Guimarães Vieira da Silva , Luiz Paulo Kowalski","doi":"10.1016/j.sste.2025.100758","DOIUrl":"10.1016/j.sste.2025.100758","url":null,"abstract":"<div><div>This study analyzed the temporal trends, spatial and spatio-temporal patterns of OC and OPC mortality in Brazil between 1980 and 2023, and explored their association with socioeconomic inequality. We conducted an ecological study using age- and sex-standardized mortality rates, smoothed via a local empirical Bayesian method. We assessed temporal trends through joinpoint regression. We evaluated global and local spatial autocorrelation and detected spatio-temporal clusters using a retrospective space–time scan statistic based on a Poisson model. We observed a decrease in OC mortality, particularly among men aged 40–59 years in the Southeast and South regions. In contrast, OPC mortality increased throughout the study period in both sexes, especially among individuals aged 60–79 years, with the largest increases occurring in the North, Northeast, and Central-West regions. Moran’s I revealed significant spatial dependence for both cancers. Spatial analyses identified persistent high-risk clusters in the Southeast and South, which expanded toward the Northeast and Central-West. Spatio-temporal analysis showed a recent shift of major OC clusters from the Southeast and South towards the Northeast, whereas OPC clusters continued to expand into the Central-West. Municipalities within clusters characterized by a low Human Development Index exhibited comparatively stronger increases in mortality trends for both cancers. These results underscore the need for more equitable and regionally tailored public policies to strengthen cancer control efforts in Brazil.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100758"},"PeriodicalIF":1.7,"publicationDate":"2025-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145269594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-13DOI: 10.1016/j.sste.2025.100756
Komi Mensah Agboka , Allan Muohi Ngángá , Bonoukpoè Mawuko Sokame , Steve Soh Bernard Baleba , Tobias Landmann , Elfatih M. Abdel-Rahman , Chrysantus M. Tanga , Souleymane Diallo
Tularemia, a neglected zoonosis, remains underreported in Africa despite growing concern over its climate-driven expansion. This study aims to quantify the specific contribution of climate to tularemia risk using a climate attribution framework. We trained a Least Squares Dummy Variable (LSDV) fixed-effects panel model on United States (U.S.) county-level tularemia incidence data from 2011–2020 (n = 500, R² = 0.90), incorporating only climatic predictors: cumulative temperature, cumulative precipitation, and their respective variabilities. The climate-only model explained 86% of variance in the training data, demonstrating strong climate influence on tularemia disease dynamics. We then applied the model to East Africa, using environmental similarity analysis to assess transferability. Results show moderate-to-high climatic analogues in northern Kenya, eastern Uganda, and South Sudan. Between 2017 and 2020, predicted tularemia suitability increased by a median of +0.18 compared to the 2012–2015 baseline, particularly in arid and semi-arid zones. Low interannual variability suggests persistent climatic suitability. A thermal plausibility test showed strong agreement (r = 0.82) between predicted risk and the Gaussian thermal profile of Francisella tularensis. Our findings suggest that climate alone can spatially explain tularemia risk across Africa’s drylands. This method provides a transferable framework for early warning in data-poor regions and supports anticipatory surveillance in the context of climate change.
{"title":"Climate-driven potential for tularemia in East Africa: skill testing and ecological consistency of a transferred risk model","authors":"Komi Mensah Agboka , Allan Muohi Ngángá , Bonoukpoè Mawuko Sokame , Steve Soh Bernard Baleba , Tobias Landmann , Elfatih M. Abdel-Rahman , Chrysantus M. Tanga , Souleymane Diallo","doi":"10.1016/j.sste.2025.100756","DOIUrl":"10.1016/j.sste.2025.100756","url":null,"abstract":"<div><div>Tularemia, a neglected zoonosis, remains underreported in Africa despite growing concern over its climate-driven expansion. This study aims to quantify the specific contribution of climate to tularemia risk using a climate attribution framework. We trained a Least Squares Dummy Variable (LSDV) fixed-effects panel model on United States (U.S.) county-level tularemia incidence data from 2011–2020 (n = 500, R² = 0.90), incorporating only climatic predictors: cumulative temperature, cumulative precipitation, and their respective variabilities. The climate-only model explained 86% of variance in the training data, demonstrating strong climate influence on tularemia disease dynamics. We then applied the model to East Africa, using environmental similarity analysis to assess transferability. Results show moderate-to-high climatic analogues in northern Kenya, eastern Uganda, and South Sudan. Between 2017 and 2020, predicted tularemia suitability increased by a median of +0.18 compared to the 2012–2015 baseline, particularly in arid and semi-arid zones. Low interannual variability suggests persistent climatic suitability. A thermal plausibility test showed strong agreement (r = 0.82) between predicted risk and the Gaussian thermal profile of <em>Francisella tularensis</em>. Our findings suggest that climate alone can spatially explain tularemia risk across Africa’s drylands. This method provides a transferable framework for early warning in data-poor regions and supports anticipatory surveillance in the context of climate change.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100756"},"PeriodicalIF":1.7,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145120052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-09-06DOI: 10.1016/j.sste.2025.100753
Andrew Curtis , Jayakrishnan Ajayakumar , Rigan Louis , Vanessa Rouzier , J Glenn Morris Jr
In this paper we describe the spatial data challenges faced in terms of providing accurate and timely analysis for a clinic during a cholera epidemic that spread through Port au Prince, Haiti in late 2022. This “triage” spatial epidemiology involved developing a bespoke geocoder that allowed for weekly maps of spread to be created in near real time. Resulting case data were also analyzed using a novel grid heatmapping approach which considers the epidemiological curve for each neighborhood. Adding further complexity during this period to both the data generation, and explaining cholera amplification and spread patterns, was a rising gang presence in the Port au Prince neighborhoods. Results identify a coastal pattern of amplification, which is expected given the informal settlement style living environments found in many of these neighborhoods. A second pattern then emerges of spread along a western and southern axis, which is far better captured in the grid heat mapping approach because of the lower numbers of patients seeking care at the clinic. The combination of traditional cartography and grid heat mapping help reveal the overall pattern of the epidemic, while also identifying key neighborhoods that require additional epidemiological investigation. Knowing why these neighborhoods played such an important role, possibly due to specific gang activity, is important in terms of understanding future disease spread in and around Port au Prince. Indeed, results presented can help contextualize official cholera reporting in 2025 where data availability is still hampered by ongoing gang rule.
{"title":"Providing spatial support during a major cholera outbreak in Port-au-Prince, Haiti: Creative mapping solutions in a challenging data poor environment","authors":"Andrew Curtis , Jayakrishnan Ajayakumar , Rigan Louis , Vanessa Rouzier , J Glenn Morris Jr","doi":"10.1016/j.sste.2025.100753","DOIUrl":"10.1016/j.sste.2025.100753","url":null,"abstract":"<div><div>In this paper we describe the spatial data challenges faced in terms of providing accurate and timely analysis for a clinic during a cholera epidemic that spread through Port au Prince, Haiti in late 2022. This “triage” spatial epidemiology involved developing a bespoke geocoder that allowed for weekly maps of spread to be created in near real time. Resulting case data were also analyzed using a novel grid heatmapping approach which considers the epidemiological curve for each neighborhood. Adding further complexity during this period to both the data generation, and explaining cholera amplification and spread patterns, was a rising gang presence in the Port au Prince neighborhoods. Results identify a coastal pattern of amplification, which is expected given the informal settlement style living environments found in many of these neighborhoods. A second pattern then emerges of spread along a western and southern axis, which is far better captured in the grid heat mapping approach because of the lower numbers of patients seeking care at the clinic. The combination of traditional cartography and grid heat mapping help reveal the overall pattern of the epidemic, while also identifying key neighborhoods that require additional epidemiological investigation. Knowing why these neighborhoods played such an important role, possibly due to specific gang activity, is important in terms of understanding future disease spread in and around Port au Prince. Indeed, results presented can help contextualize official cholera reporting in 2025 where data availability is still hampered by ongoing gang rule.</div></div>","PeriodicalId":46645,"journal":{"name":"Spatial and Spatio-Temporal Epidemiology","volume":"55 ","pages":"Article 100753"},"PeriodicalIF":1.7,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145050553","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号