Spatiotemporal Analysis of Air Pollution Patterns and Their Environmental Impacts Using Advanced Statistical Models

INTRODUCTION

Air pollution remains a major environmental risk because pollutant concentrations vary sharply across neighborhoods, seasons, and meteorological regimes while producing health, agricultural, and ecological consequences. The global burden attributable to ambient air pollution includes premature mortality, respiratory and cardiovascular morbidity, and long-term exposure-related disease, making exposure assessment central to environmental statistics and spatial epidemiology (Cohen et al., 2017; Keller et al., 2017). Fine particulate matter, nitrogen dioxide, and ozone also affect crop productivity and ecological functioning, especially where emissions coincide with vulnerable populations or sensitive ecosystems (Ryalls et al., 2017; Tai & Martin, 2017). A rigorous spatiotemporal framework is therefore needed to estimate exposure surfaces and quantify environmental impacts with defensible uncertainty (Kang et al., 2019; Chakraborty et al., 2022).

Routine monitoring networks provide high-quality measurements but are often spatially sparse, unevenly distributed, and concentrated in urban or regulatory-priority areas. Simple interpolation can recover broad gradients but may fail where land use, traffic, industrial emissions, topography, and meteorology create non-stationary concentration fields (Just et al., 2020; Shao et al., 2020). Kriging models that ignore temporal dependence can miss seasonal inversions, wildfire-smoke events, and day-to-day transport, while regression models that ignore spatial autocorrelation can underestimate standard errors and overstate exposure-response precision (Fioravanti et al., 2021; Sáez Zafra & Barceló Rado, 2022). These limitations are particularly important when pollution surfaces are later used in health or crop-impact models (Harper et al., 2021; Sánchez-Balseca & Pérez-Foguet, 2022).

Recent advances in spatiotemporal statistics combine ground monitoring, satellite retrievals, land-use predictors, and meteorological reanalysis through hierarchical, geostatistical, and machine-learning frameworks. Bayesian hierarchical models, Gaussian random fields, and INLA-SPDE approximations provide probabilistic prediction and uncertainty quantification for large spatial datasets (Cameletti et al., 2019; Chen et al., 2023; Otto et al., 2024). Satellite-based land-use regression, random forest spatiotemporal prediction, and geographically weighted regression improve spatial coverage by incorporating aerosol optical depth, road density, population, emissions proxies, and atmospheric covariates (Huang et al., 2018; Lee, 2019; Schneider et al., 2020). Hybrid strategies that model nonlinear covariate effects and residual spatial dependence are especially useful where pollutant processes vary across climate zones or emission regimes (Wong et al., 2021; Shen et al., 2022).

This manuscript develops an STA framework for mapping air pollution patterns and linking exposure to environmental impacts using statistically coherent spatiotemporal models. The central thesis is that models accounting for space-time covariance, non-stationarity, measurement error, and exposure uncertainty produce more reliable pollution maps than isolated monitoring or single-source satellite estimates (Cameletti et al., 2019; Just et al., 2020; Pu & Yoo, 2021). The analysis emphasizes Bayesian hierarchical modeling with Matérn random fields, INLA-SPDE computation, geographically and temporally weighted regression, and kriging-based residual correction (Qin et al., 2017; Fioravanti et al., 2021; Otto et al., 2024). These methods are positioned not only as prediction tools but also as inferential instruments for hotspot detection, exceedance probability mapping, and uncertainty-aware burden estimation (Wei et al., 2019; Shen et al., 2024).

Figure 1 presents the hierarchical analytical workflow linking heterogeneous pollution data sources, preprocessing, exploratory dependence diagnostics, advanced spatiotemporal modeling, uncertainty propagation, environmental-impact estimation, and policy translation.

Background

Air pollution constituents and sources

PM2.5, PM10, NO2, SO2, O3, and CO arise from interacting anthropogenic and natural processes, including traffic, combustion, industrial activity, biomass burning, dust, and atmospheric chemical transformation. PM2.5 often reflects both primary combustion particles and secondary aerosol formation, while NO2 is strongly associated with traffic and fuel combustion, making it highly spatially heterogeneous near road networks and dense urban corridors (Lee, 2019; Harper et al., 2021). Ozone differs because it is a secondary pollutant shaped by photochemistry, temperature, precursor emissions, and regional transport, which can create spatial patterns that diverge from primary pollutants (Tai & Martin, 2017). Because emission sources and atmospheric reactions operate at different scales, a model must accommodate both local gradients and regional background structure (Qin et al., 2017; Huang et al., 2018).

Environmental and health impacts of air pollution

Air pollution affects human and environmental systems through chronic exposure, acute episodes, and cumulative stress across biological and ecological pathways. Spatiotemporal health studies link particulate and gaseous pollutants to respiratory admissions, cardiovascular mortality, and broader mortality burdens, with Bayesian disease-mapping and exposure-response models used to account for regional heterogeneity (Kang et al., 2019; Chakraborty et al., 2022; Sánchez-Balseca & Pérez-Foguet, 2022; Carpio-Vargas et al., 2023a, 2023b). Environmental effects include ozone-related crop-yield loss, impaired pollination, and harm to beneficial invertebrate communities, making air quality relevant to food systems and ecosystem services as well as public health (Tai & Martin, 2017; Ryalls et al., 2024; Ryalls et al., 2025). These impacts require exposure fields that match the spatial and temporal support of populations, farms, habitats, and administrative outcome data (Cohen et al., 2017; Keller et al., 2017).

Spatiotemporal data structures

Air-pollution data combine point-referenced monitoring observations, gridded satellite retrievals, chemical transport model outputs, meteorological reanalysis, and areal outcome records. Monitoring stations provide calibrated time series but are irregularly spaced, whereas satellite products such as MAIAC aerosol optical depth provide broad spatial coverage with missingness caused by clouds, snow, bright surfaces, and retrieval limitations (Lee, 2019; Schneider et al., 2020; Pu & Yoo, 2021). Land-use regression and machine-learning fusion models translate these heterogeneous sources into regular prediction grids, typically by joining spatial covariates, temporal indicators, satellite observations, and station measurements (Wu et al., 2017; Harper et al., 2021; Wong et al., 2021). Environmental outcomes such as hospital admissions or crop yield are often reported by administrative units, creating spatial-support mismatch that must be addressed in exposure assignment and uncertainty propagation (Cameletti et al., 2019; Sánchez-Balseca & Pérez-Foguet, 2022).

Classical geostatistical and time series methods

Classical geostatistical analysis uses variograms, covariance functions, and kriging to represent spatial dependence and predict concentrations at unsampled locations. Universal kriging and regression kriging improve ordinary kriging by incorporating covariates, but they can remain inadequate when temporal dynamics, non-stationarity, or satellite-missingness patterns dominate prediction error (Cameletti et al., 2019; Shao et al., 2020). Traditional time-series methods capture autocorrelation, seasonality, and long-term trends but usually treat locations independently or rely on simplified spatial structures (Kang et al., 2019; Wang et al., 2022). For air pollution, the main limitation of classical methods is that spatial transport, meteorological persistence, and emission heterogeneity jointly produce dependence that is neither purely spatial nor purely temporal (Luo et al., 2017; Sáez Zafra & Barceló Rado, 2022).

Advanced spatiotemporal statistical frameworks

Advanced STA frameworks represent pollutant concentration as the sum of fixed covariate effects, structured spatial random fields, temporal dependence, interaction terms, and measurement error. Bayesian hierarchical models allow the data model, process model, and parameter model to be specified separately, supporting coherent uncertainty quantification for prediction and downstream impact estimation (Cameletti et al., 2019; Fioravanti et al., 2021; Sáez Zafra & Barceló Rado, 2022). INLA-SPDE methods approximate Matérn Gaussian random fields efficiently by using a triangulated mesh and sparse precision matrices, enabling large-scale spatial inference without full dense covariance computation (Fioravanti et al., 2021; Otto et al., 2024). Machine-learning hybrids, geographically weighted regression, and satellite-enhanced land-use models complement Bayesian approaches by capturing nonlinear predictor-response relationships and spatially varying effects (Schneider et al., 2020; Shen et al., 2022; Shen et al., 2024).

Data sources and preprocessing

Ground monitoring and satellite data

The empirical design uses hourly and daily pollutant observations from at least 50 monitoring stations, aggregated to daily summaries for PM2.5, PM10, NO2, SO2, O3, and CO after quality-control screening. Station records are paired with satellite aerosol optical depth from MODIS MAIAC, trace-gas retrievals where available, and chemical transport or reanalysis products used for gap-filling and background concentration information (Huang et al., 2018; Lee, 2019; Pu & Yoo, 2021). Outlier detection removes negative values, instrument-failure spikes, and implausible temporal discontinuities, while missingness indicators are retained because satellite availability can be informative under cloudy, smoky, or high-humidity conditions (Schneider et al., 2020; Wong et al., 2021). The processed dataset is structured as a station-day panel and a prediction grid, allowing simultaneous calibration to ground observations and generation of continuous exposure surfaces (Huang et al., 2018; Shao et al., 2020).

Meteorological and land use covariates

Meteorological predictors include temperature, relative humidity, precipitation, wind speed, wind direction, boundary-layer height, pressure, and stagnation indicators obtained from reanalysis products and harmonized to the prediction grid. Land-use variables include road density, traffic intensity, industrial land cover, elevation, vegetation indices, impervious surface, population density, nighttime lights, distance to major roads, and distance to emission sources (Wu et al., 2017; Harper et al., 2021). These covariates are standardized to a 1-km spatial grid and aligned to daily time steps where meteorological variation is relevant, while static land-use covariates remain fixed or updated annually (Qin et al., 2017; Shen et al., 2022). Because pollutant processes differ by source and atmospheric behavior, covariate design allows local traffic effects for NO2, regional aerosol transport for PM2.5, and photochemical conditions for O3 (Qin et al., 2017; Huang et al., 2018).

Environmental impact outcomes

Environmental-impact data are linked to predicted pollutant fields using spatial overlays and temporal aggregation matched to each outcome (Dorji & Wangchuk, 2024; Sivasli et al., 2024; Cavero & Ferraz, 2025; Hamaideh et al., 2025; Sagredo-Olivares & Bravo, 2025; Tutticci & Marian, 2025). For health analysis, daily respiratory or cardiovascular admissions by zip code are assigned population-weighted pollution exposure from daily concentration surfaces, with adjustment for temperature, seasonality, day of week, socioeconomic structure, and long-term trend (Keller et al., 2017; Kang et al., 2019; Chakraborty et al., 2022). For crop analysis, district-level wheat or rice yield is linked to growing-season PM2.5, ozone, temperature extremes, drought indicators, and phenological windows sensitive to pollution stress (Tai & Martin, 2017). For ecosystem analysis, predicted pollutants are paired with vegetation stress, pollinator exposure, forest vulnerability, or invertebrate-response indicators where ecological outcome data are available (Ryalls et al., 2024; Ryalls et al., 2025).

Exploratory spatiotemporal analysis

Spatial autocorrelation analysis

Exploratory spatial analysis estimates global Moran’s I for each daily or monthly pollutant surface to assess whether high and low concentrations cluster beyond random spatial arrangement. Local indicators of spatial association identify persistent hotspots near traffic corridors, industrial zones, basin-like topography, or regions affected by recurring stagnation and transport patterns (Luo et al., 2017; Wang et al., 2022). LISA maps are summarized across seasons to distinguish chronic hotspots from episodic clusters caused by dust, wildfire smoke, or inversion events (Shao et al., 2020; Pu & Yoo, 2021). These diagnostics guide model specification by indicating whether spatial random effects, local coefficients, anisotropic covariance, or non-stationary components are needed (Wei et al., 2019; Shen et al., 2024).

Temporal patterns and decomposition

Temporal exploratory analysis decomposes pollutant series into long-term trend, seasonal, weekly, diurnal, and residual components using station-level and region-level summaries. PM2.5 often shows wintertime enhancement under stable boundary layers or heating emissions, whereas NO2 reflects commuting patterns and O3 reflects photochemical production with stronger warm-season cycles (Qin et al., 2017; Tai & Martin, 2017; Wu et al., 2017). STL decomposition and temporal smoothing separate gradual emission or policy trends from short extreme episodes such as smoke intrusions, dust storms, or meteorological stagnation (Huang et al., 2018; Schneider et al., 2020). These temporal diagnostics inform autoregressive terms, seasonal basis functions, change-point indicators, and event covariates in the spatiotemporal model (Kang et al., 2019; Wang et al., 2022).

Space-time variography

Space-time variography quantifies how concentration similarity decays with both spatial distance and temporal lag. Empirical variograms estimate nugget, sill, spatial range, temporal range, and possible space-time interaction, providing evidence on whether a separable covariance structure is adequate (Shao et al., 2020; Fioravanti et al., 2021). Directional variograms evaluate anisotropy that may arise from prevailing winds, valley-channel transport, or regional emission corridors, while temporal variograms capture day-to-day persistence and seasonal dependence (Fioravanti et al., 2021; Otto et al., 2024). These diagnostics are used to choose between regression kriging, dynamic geostatistical models, Matérn random fields, and non-separable covariance formulations (Cameletti et al., 2019; Chen et al., 2023).

Advanced spatiotemporal model specification

Bayesian hierarchical model framework

The core model treats observed pollutant concentration as , where s denotes location and t denotes time. The process model is , where  represents fixed effects for meteorology, satellite retrievals, land use, and temporal indicators, ω(s,t) is a structured spatial or spatiotemporal random effect, and ξ(t) is an autoregressive temporal component (Cameletti et al., 2019; Fioravanti et al., 2021). A Matérn covariance is assigned to the spatial field, with penalized complexity priors on spatial range and marginal variance to reduce overfitting and improve interpretability (Chen et al., 2023; Otto et al., 2024). This hierarchy separates observation noise from latent pollution processes, enabling posterior prediction, uncertainty mapping, and exposure propagation into second-stage health or environmental models (Keller et al., 2017; Sánchez-Balseca & Pérez-Foguet, 2022).

SPDE/INLA approximation for large datasets

For large datasets, the Matérn Gaussian field is represented through a stochastic partial differential equation and approximated on a triangulated spatial mesh, yielding sparse precision matrices suitable for integrated nested Laplace approximation. This SPDE/INLA strategy avoids direct inversion of dense covariance matrices and is therefore appropriate for daily multi-year monitoring, satellite, and gridded covariate datasets with more than 10,000 observations (Fioravanti et al., 2021; Chen et al., 2023). Mesh design balances computational feasibility and spatial fidelity by using finer triangles in dense monitoring areas and coarser triangles in peripheral regions (Otto et al., 2024). The resulting posterior surfaces provide daily predicted concentrations, posterior standard deviations, exceedance probabilities, and spatial random-effect maps that distinguish observed covariate effects from residual spatial dependence (Cameletti et al., 2019; Fioravanti et al., 2021).

Alternative: geographically and temporally weighted regression

Geographically and temporally weighted regression provides a local modeling alternative in which regression coefficients vary across space and time according to kernel weights and bandwidth parameters. Bandwidth selection by cross-validation allows the model to adapt to heterogeneous relationships between pollutants and predictors such as traffic intensity, land cover, meteorology, and satellite retrievals (Shen et al., 2022; Shen et al., 2024). GTWR is particularly useful for diagnosing non-stationarity and comparing local coefficient surfaces with global Bayesian or kriging-based estimates (Qin et al., 2017; Zhao et al., 2020). However, because local regression does not automatically provide the same probabilistic dependence structure as hierarchical geostatistical models, it is best used as a complementary tool for interpretation, sensitivity analysis, and model comparison (Luo et al., 2017; Wei et al., 2019).

Table 1 consolidates the analytical contribution of each model class by distinguishing its dependence structure, uncertainty capacity, interpretability, and suitability for environmental-impact inference.

Table 1. Analytical role of major spatiotemporal model classes in air-pollution exposure estimation

Spatial and temporal dependence structures

Separable vs. non-separable covariance

A separable covariance assumes that space-time dependence can be written as the product of a spatial correlation function and a temporal correlation function, but air-pollution processes often violate this simplification. For example, pollutant transport depends on wind, boundary-layer evolution, emissions timing, and atmospheric chemistry, so spatial correlation may be stronger on some days than others and temporal persistence may differ across regions (Shao et al., 2020; Fioravanti et al., 2021). Non-separable covariance structures allow the spatial range, temporal decay, and space-time interaction to vary jointly, improving representation of smoke episodes, urban rush-hour plumes, and regional stagnation events (Cameletti et al., 2019; Fioravanti et al., 2021). Model comparison can evaluate separability using likelihood-based criteria, predictive performance under spatiotemporal holdouts, and posterior diagnostics of residual dependence (Chen et al., 2023; Otto et al., 2024).

Anisotropy and non-stationarity

Anisotropy occurs when spatial correlation depends on direction, which is common for air pollutants transported by prevailing winds, constrained by topography, or aligned with road and industrial corridors. Directional variograms can reveal whether concentration similarity decays more slowly along dominant transport pathways than across them, motivating anisotropic Matérn fields, deformation methods, or spatially varying covariance parameters (Fioravanti et al., 2021; Otto et al., 2024). Non-stationarity also arises when source intensity, land use, meteorology, and chemical regimes differ across urban, rural, coastal, and mountainous environments (Shen et al., 2022, 2024). Local regression models and geographically weighted frameworks help diagnose these variations, while hierarchical models can incorporate spatially varying coefficients or regional random effects to preserve probabilistic uncertainty quantification (Qin et al., 2017; Wei et al., 2019; Zhao et al., 2020).

Incorporating expert knowledge via priors

Bayesian spatiotemporal modeling allows expert knowledge to enter through priors on spatial range, marginal variance, temporal autocorrelation, regression coefficients, and measurement-error variance. Physical dispersion knowledge can inform plausible spatial ranges for traffic-related NO2, regional PM2.5, and secondary pollutants, while penalized complexity priors help prevent excessively flexible random fields that absorb meaningful covariate effects (Cameletti et al., 2019; Chen et al., 2023). Prior sensitivity analysis is essential because strong assumptions about spatial range or temporal persistence can affect hotspot probability, exposure estimates, and downstream health or crop-response inference (Keller et al., 2017; Sánchez-Balseca & Pérez-Foguet, 2022). Transparent reporting of posterior changes under alternative priors strengthens the credibility of environmental-impact estimates and distinguishes statistical uncertainty from subjective modeling choices (Kang et al., 2019; Chakraborty et al., 2022).

Environmental impact modeling

Second-stage exposure-response models

Predicted pollutant fields can be used as exposure inputs in second-stage models for health, crop, and ecosystem outcomes, with uncertainty propagated from the first-stage spatiotemporal surface (Feng et al., 2024; Hernandez et al., 2024; Kranjc et al., 2024; Alves et al., 2025; Eriksson et al., 2025; Kunie et al., 2025; Kwatra et al., 2025; Pantiș et al., 2025; Scott et al., 2025; Yildiz & Karaca, 2025). For health outcomes, daily respiratory or cardiovascular admissions can be modeled with generalized additive or Bayesian regression models that adjust for temperature, seasonality, population structure, socioeconomic factors, and temporal trends (Kang et al., 2019; Chakraborty et al., 2022). For crop outcomes, growing-season exposure to PM2.5 and O3 can be linked to district-level yield using spatial panel models or nonlinear exposure-response functions that account for heat, drought, and phenological sensitivity (Tai & Martin, 2017). For ecosystem outcomes, predicted exposure surfaces can be matched to pollinator, invertebrate, vegetation, or forest-stress indicators to assess ecological vulnerability under chronic and episodic pollution regimes (Ryalls et al., 2024; Ryalls et al., 2025).

Quantifying attributable burden

Attributable burden estimation translates exposure-response coefficients and predicted pollutant surfaces into policy-relevant quantities such as excess admissions, premature deaths, crop-yield losses, and exposed ecosystem area. For health analysis, the population-attributable fraction can be estimated by comparing predicted outcomes under observed pollution with a counterfactual surface that meets a regulatory or health-protective threshold (Cohen et al., 2017; Keller et al., 2017). For agriculture, yield-loss functions can compare observed growing-season exposure with a low-pollution counterfactual, producing district-specific estimates of production loss and uncertainty intervals (Tai & Martin, 2017). Because exposure surfaces are estimated rather than directly observed, uncertainty propagation from the spatiotemporal model is necessary to avoid overconfident burden estimates (Cameletti et al., 2019; Sánchez-Balseca & Pérez-Foguet, 2022).

Practical applications for policy and planning

Identifying hotspots and high-risk periods

Spatiotemporal models support hotspot detection by producing daily and seasonal maps of predicted concentrations, posterior uncertainty, and exceedance probabilities. Exceedance probability maps for PM2.5, NO2, or O3 can identify locations where concentrations are likely to surpass health-based or regulatory thresholds, even when monitoring stations are absent (Schneider et al., 2020; Pu & Yoo, 2021). Persistent hotspot classification can combine predicted means, local spatial association, and recurrence frequency to distinguish chronic exposure zones from short-lived event-driven clusters (Luo et al., 2017; Wang et al., 2022). These outputs are useful for early warning, monitoring-network redesign, environmental justice screening, and prioritization of interventions during wildfire smoke, dust, inversion, or photochemical episodes (Huang et al., 2018; Shao et al., 2020).

Cost-benefit analysis of mitigation strategies

Policy analysis can use spatiotemporal predictions to identify where emission reductions are likely to produce the largest health, agricultural, or ecological gains. Seasonal traffic restrictions, industrial-emission controls, agricultural burning limits, and fertilizer-related precursor reductions can be evaluated by simulating counterfactual concentration surfaces and comparing predicted impact reductions across space and time (Qin et al., 2017; Shen et al., 2022). Because local responses may differ across urban cores, transport corridors, agricultural districts, and background regions, geographically weighted and hierarchical models help avoid assuming a uniform policy effect (Zhao et al., 2020; Shen et al., 2024). The resulting benefit estimates can combine avoided admissions, reduced mortality burden, prevented crop loss, and ecosystem protection while preserving uncertainty from exposure prediction and exposure-response modeling (Cohen et al., 2017; Keller et al., 2017; Tai & Martin, 2017; Ryalls et al., 2024).

Model evaluation and validation strategy

Spatiotemporal cross-validation

Model validation should use spatial holdout, temporal holdout, and spatiotemporal block cross-validation rather than relying only on random folds. Leave-one-station-out validation evaluates prediction at unmonitored locations, leave-one-season-out validation evaluates temporal transferability, and event-based validation tests performance during smoke, dust, inversion, or high-ozone episodes (Just et al., 2020; Schneider et al., 2020). Core performance metrics include root mean square prediction error, mean absolute error, mean bias, R², continuous ranked probability score, prediction-interval coverage, and interval width (Shao et al., 2020; Pu & Yoo, 2021). Because uncertainty calibration is as important as point accuracy, probabilistic models should be judged by whether nominal credible or prediction intervals achieve empirical coverage under spatial and temporal holdouts (Cameletti et al., 2019; Otto et al., 2024).

Comparison to baseline models

Advanced models should be compared with ordinary kriging, universal kriging, land-use regression, global linear mixed models, random forest without residual spatial correction, and satellite-only prediction. Baseline comparisons clarify whether improvements come from satellite data fusion, nonlinear machine learning, local coefficient estimation, explicit covariance modeling, or Bayesian uncertainty propagation (Wu et al., 2017; Lee, 2019; Wong et al., 2021). A model that improves RMSPE but undercovers prediction intervals may be less useful for environmental-impact assessment than a slightly less accurate model with better uncertainty calibration (Keller et al., 2017; Sáez Zafra & Barceló Rado, 2022). Comparisons should therefore report both predictive accuracy and inferential reliability, especially when predicted fields are used in downstream health, crop, or ecosystem models (Tai & Martin, 2017; Sánchez-Balseca & Pérez-Foguet, 2022).

Sensitivity and robustness analyses

Sensitivity analysis should perturb covariance parameters, prior distributions, spatial mesh resolution, satellite gap-filling choices, pollutant averaging windows, and the definition of extreme-event indicators. Excluding high-leverage monitoring stations tests whether hotspots are driven by single sites, while alternative spatial grids evaluate whether conclusions depend on the selected prediction support (Fioravanti et al., 2021; Chen et al., 2023). Robustness checks should also compare Bayesian hierarchical estimates with GTWR, land-use regression, and machine-learning hybrids to assess whether spatial patterns remain stable across modeling assumptions (Huang et al., 2018; Wei et al., 2019; Shen et al., 2022). For impact models, additional sensitivity analyses should vary exposure-response lag structures, confounder adjustment, counterfactual thresholds, and uncertainty propagation methods (Cohen et al., 2017; Kang et al., 2019; Chakraborty et al., 2022).

Table 2 extends the manuscript’s analytical depth by mapping pollutant-specific process behavior to diagnostic evidence, model-design decisions, validation strategies, and environmental-impact endpoints.

Table 2. Decision matrix linking pollutant processes, diagnostics, model choices, and environmental-impact endpoints

Alves, R., Mendes, T., Pinto, J., & Correia, N. (2025). Warfarin pharmacogenomics in the era of precision medicine: Persistent underrepresentation of non-European ancestries and implications for global health equity. Specialty Journal of Pharmacognosy, Phytochemistry and Biotechnology, 5, 1–26. doi:10.51847/3jxE52PnLi

Cameletti, M., Gómez-Rubio, V., & Blangiardo, M. (2019). Bayesian modelling for spatially misaligned health and air pollution data through the INLA-SPDE approach. Spatial Statistics, 31, 100353.

Carpio-Vargas, E. E., Ibarra-Cabrera, E. M., Ibarra, M. J., Choquejahua-Acero, R., Calderon-Vilca, H. D., & Torres-Cruz, F. (2023a). Categorical stress predictors in higher education students amidst remote learning in COVID-19 pandemic. Journal of Advanced Pharmacy Education and Research, 13(2), 131–139. doi:10.51847/ImofrnDDZg

Carpio-Vargas, E. E., Torres-Cruz, F., Bernedo, E. G. M., Yanqui, F. J. M., Chaiña, H. C. P., Zapana, W. H. M., & Villasante-Saravia, F. H. (2023b). Triadic mental quotient and lifestyles in university students during pandemic-induced confinement. Journal of Advanced Pharmacy Education and Research, 13(3), 88–95. doi:10.51847/bDdfeMvcJ1

Cavero, M., & Ferraz, J. P. (2025). Sedentary behavior determinants in pregnant women during the second and third trimesters: Prospective findings from the large-scale Japan Environment and Children’s Study. Bulletin of Pioneer Research in Medical and Clinical Sciences, 5(1), 1–12. doi:10.51847/bdDIEuY7cl

Chakraborty, S., Dey, T., Jun, Y., Lim, C. Y., Mukherjee, A., & Dominici, F. (2022). A spatiotemporal analytical outlook of the exposure to air pollution and COVID-19 mortality in the USA. Journal of Agricultural, Biological and Environmental Statistics, 27(3), 419–439.

Chen, J., Miao, C., Yang, D., Liu, Y., Zhang, H., & Dong, G. (2023). Estimation of fine-resolution PM2.5 concentrations using the INLA-SPDE method. Atmospheric Pollution Research, 14(7), 101781.

Cohen, A. J., Brauer, M., Burnett, R., Anderson, H. R., Frostad, J., Estep, K., Balakrishnan, K., Brunekreef, B., Dandona, L., Dandona, R., et al. (2017). Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: An analysis of data from the Global Burden of Diseases Study 2015. The Lancet, 389(10082), 1907–1918.

Dorji, T., & Wangchuk, P. (2024). Ethical considerations in health monitoring within the armed forces: A Dutch case study during the COVID-19 pandemic. Asian Journal of Ethics in Health and Medicine, 4, 204–214. doi:10.51847/XV2L2pctU9

Eriksson, J., & Lindgren, M. (2025). Semaglutide’s impact on kidney health in obese mice: A metabolomics study. Pharmaceutical Sciences and Drug Design, 5, 114–134. doi:10.51847/5wHbT71L9K

Feng, L., Wei, G., & Lei, Z. (2024). Pharmacists’ contributions to the management of mental health conditions: A comprehensive review. Annals of Pharmacy Practice and Pharmacotherapy, 4, 125–139. doi:10.51847/ReKLpACV6c

Fioravanti, G., Martino, S., Cameletti, M., & Cattani, G. (2021). Spatio-temporal modelling of PM10 daily concentrations in Italy using the SPDE approach. Atmospheric Environment, 248, 118192.

Hamaideh, S., Khait, A. A., Al-Modallal, H., Masa’deh, R., Hamdan-Mansour, A., & AlBashtawy, M. (2025). Investigating the role of different factors in the quality of professional life of nurses: A review study. Journal of Integrated Nursing and Palliative Care, 6, 12–17. doi:10.51847/k8xYY6EKKp

Harper, A., Baker, P. N., Xia, Y., Kuang, T., Zhang, H., Chen, Y., Han, T. L., & Gulliver, J. (2021). Development of spatiotemporal land use regression models for PM2.5 and NO2 in Chongqing, China, and exposure assessment for the CLIMB study. Atmospheric Pollution Research, 12(7), 101096.

Hernandez, V., Romero, L., Sanchez, P., & Morales, D. (2024). Real-world implementation of pharmacogenetics in a public tertiary hospital: A decade of progress and 35% annual growth. Specialty Journal of Pharmacognosy, Phytochemistry and Biotechnology, 4, 118–128. doi:10.51847/N7jyuZHwDc

Huang, K., Xiao, Q., Meng, X., Geng, G., Wang, Y., Lyapustin, A., Gu, D., & Liu, Y. (2018). Predicting monthly high-resolution PM2.5 concentrations with random forest model in the North China Plain. Environmental Pollution, 242, 675–683.

Just, A. C., Arfer, K. B., Rush, J., Dorman, M., Shtein, A., Lyapustin, A., & Kloog, I. (2020). Advancing methodologies for applying machine learning and evaluating spatiotemporal models of fine particulate matter (PM2.5) using satellite data over large regions. Atmospheric Environment, 239, 117649.

Kang, D., Jang, Y., Choi, H., Hwang, S. S., Koo, Y., & Choi, J. (2019). Space–time relationship between short-term exposure to fine and coarse particles and mortality in a nationwide analysis of Korea: A Bayesian hierarchical spatio-temporal model. International Journal of Environmental Research and Public Health, 16(12), 2111.

Keller, J. P., Chang, H. H., Strickland, M. J., & Szpiro, A. A. (2017). Measurement error correction for predicted spatiotemporal air pollution exposures. Epidemiology, 28(3), 338–345.

Kranjc, M., Cankar, U., & Novak, T. (2024). Prognostic value of homologous recombination deficiency identified by comprehensive genomic profiling in incurable pancreatic cancer. Asian Journal of Current Research in Clinical Cancer, 4(1), 101–110. doi:10.51847/mFl0yDhABz

Kunie, K., Kawakami, N., Shimazu, A., Yonekura, Y., & Miyamoto, Y. (2025). Examining the impact of managerial communication on the link between nurses' job performance and psychological empowerment. Annals of Organizational Culture, Leadership and External Engagement Journal, 6, 1–7. doi:10.51847/SF5ZX3J4OT

Kwatra, D., Venugopal, A., & Anant, S. (2025). Exploring the radiosensitizing effects of tolmetin in radiotherapy for human clonal cancer cells. Asian Journal of Current Research in Clinical Cancer, 5(1), 1–7. doi:10.51847/SIEnZ1q0j8

Lee, H. J. (2019). Benefits of high resolution PM2.5 prediction using satellite MAIAC AOD and land use regression for exposure assessment: California examples. Environmental Science & Technology, 53(21), 12774–12783.

Luo, J., Du, P., Samat, A., Xia, J., Che, M., & Xue, Z. (2017). Spatiotemporal pattern of PM2.5 concentrations in mainland China and analysis of its influencing factors using geographically weighted regression. Scientific Reports, 7(1), 40607.

Otto, P., Fusta Moro, A., Rodeschini, J., Shaboviq, Q., Ignaccolo, R., Golini, N., Cameletti, M., Maranzano, P., Finazzi, F., & Fassò, A. (2024). Spatiotemporal modelling of PM2.5 concentrations in Lombardy (Italy): A comparative study. Environmental and Ecological Statistics, 31(2), 245–272.

Pantiș, C., Cheregi, C. D., Căiță, G. A., & Szilagyi, G. (2025). Evaluating the distribution and adequacy of human resources in hospital settings. Annals of Organizational Culture, Leadership and External Engagement Journal, 6, 51–55. doi:10.51847/hamJD5d7E3

Pu, Q., & Yoo, E. H. (2021). Ground PM2.5 prediction using imputed MAIAC AOD with uncertainty quantification. Environmental Pollution, 274, 116574.

Qin, K., Rao, L., Xu, J., Bai, Y., Zou, J., Hao, N., Li, S., & Yu, C. (2017). Estimating ground level NO2 concentrations over Central-Eastern China using a satellite-based geographically and temporally weighted regression model. Remote Sensing, 9(9), 950.

Ryalls, J. M., Bishop, J., Mofikoya, A. O., Bromfield, L. M., Nakagawa, S., & Girling, R. D. (2024). Air pollution disproportionately impairs beneficial invertebrates: A meta-analysis. Nature Communications, 15(1), 5447.

Ryalls, J. M., Bromfield, L. M., Mullinger, N. J., Langford, B., Mofikoya, A. O., Pfrang, C., Nemitz, E., Blande, J. D., & Girling, R. D. (2025). Diesel exhaust and ozone adversely affect pollinators and parasitoids within flying insect communities. Science of the Total Environment, 958, 177802.

Sáez Zafra, M., & Barceló Rado, M. A. (2022). Spatial prediction of air pollution levels using a hierarchical Bayesian spatiotemporal model in Catalonia, Spain. Environmental Modelling & Software, 151, 105369.

Sagredo-Olivares, K., & Bravo, P. R. (2025). FKBP5 gene polymorphisms and insomnia symptoms during depressive episodes in stress-related bipolar disorder. Bulletin of Pioneer Research in Medical and Clinical Sciences, 5(1), 137–151. doi:10.51847/jKltbXlwMc

Sánchez-Balseca, J., & Pérez-Foguet, A. (2022). Spatially-structured human mortality modelling using air pollutants with a compositional approach. Science of the Total Environment, 813, 152486.

Schneider, R., Vicedo-Cabrera, A. M., Sera, F., Masselot, P., Stafoggia, M., de Hoogh, K., Kloog, I., Reis, S., Vieno, M., & Gasparrini, A. (2020). A satellite-based spatio-temporal machine learning model to reconstruct daily PM2.5 concentrations across Great Britain. Remote Sensing, 12(22), 3803.

Scott, O., Miller, W., & White, C. (2025). Potential molecular targets for hypertension in Allium schoenoprasum identified through network pharmacology and molecular docking approaches. Annals of Pharmacy Practice and Pharmacotherapy, 5, 164–173. doi:10.51847/nKkK01MaCz

Shao, Y., Ma, Z., Wang, J., & Bi, J. (2020). Estimating daily ground-level PM2.5 in China with random-forest-based spatiotemporal kriging. Science of the Total Environment, 740, 139761.

Shen, Y., de Hoogh, K., Schmitz, O., Clinton, N., Tuxen-Bettman, K., Brandt, J., Christensen, J. H., Frohn, L. M., Geels, C., Karssenberg, D., & Vermeulen, R. (2022). Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression. Environment International, 168, 107485.

Shen, Y., de Hoogh, K., Schmitz, O., Clinton, N., Tuxen-Bettman, K., Brandt, J., Christensen, J. H., Frohn, L. M., Geels, C., Karssenberg, D., et al. (2024). Monthly average air pollution models using geographically weighted regression in Europe from 2000 to 2019. Science of the Total Environment, 918, 170550.

Sivasli, A., Cinar, F., & Sanlier, N. B. (2024). Severe earthquake effects on the nurses working in the neonatal units in the disaster region. Journal of Integrated Nursing and Palliative Care, 5, 68–75. doi:10.51847/3wpDhqr9Y4

Tai, A. P., & Martin, M. V. (2017). Impacts of ozone air pollution and temperature extremes on crop yields: Spatial variability, adaptation and implications for future food security. Atmospheric Environment, 169, 11–21.

Tutticci, S., & Marian, M. (2025). Integrating environmental sustainability into clinical decision-making: A systematic review of rationale. Asian Journal of Ethics in Health and Medicine, 5, 79–94. doi:10.51847/oGhDOKCuki

Wang, H., Chen, Z., & Zhang, P. (2022). Spatial autocorrelation and temporal convergence of PM2.5 concentrations in Chinese cities. International Journal of Environmental Research and Public Health, 19(21), 13942.

Wei, Q., Zhang, L., Duan, W., & Zhen, Z. (2019). Global and geographically and temporally weighted regression models for modeling PM2.5 in Heilongjiang, China from 2015 to 2018. International Journal of Environmental Research and Public Health, 16(24), 5107.

Wong, P. Y., Lee, H. Y., Chen, Y. C., Zeng, Y. T., Chern, Y. R., Chen, N. T., Lung, S. C., Su, H. J., & Wu, C. D. (2021). Using a land use regression model with machine learning to estimate ground level PM2.5. Environmental Pollution, 277, 116846.

Wu, C. D., Chen, Y. C., Pan, W. C., Zeng, Y. T., Chen, M. J., Guo, Y. L., & Lung, S. C. (2017). Land-use regression with long-term satellite-based greenness index and culture-specific sources to model PM2.5 spatial-temporal variability. Environmental Pollution, 224, 148–157.

Yildiz, A., & Karaca, S. (2025). Pharmacokinetics and safety of imrecoxib, a novel selective cyclooxygenase-2 inhibitor, in elderly healthy subjects: A comparative study. Pharmaceutical Sciences and Drug Design, 5, 85–95. doi:10.51847/AUoiF17rVt

Zhao, R., Zhan, L., Yao, M., & Yang, L. (2020). A geographically weighted regression model augmented by Geodetector analysis and principal component analysis for the spatial distribution of PM2.5. Sustainable Cities and Society, 56, 102106.