Constrained Mixed-Effect Models with Ensemble Learning for Prediction of Nitrogen Oxides Concentrations at High Spatiotemporal Resolution

High-Resolution Geospatial Database: National Criteria-Air-Pollutant Concentrations in the Contiguous U.S., 2016-2020

Concentration estimates for ambient air pollution are used widely in fields such as environmental epidemiology, health impact assessment, urban planning, environmental equity and sustainability. This study builds on previous efforts by developing an updated high-resolution geospatial database of population-weighted annual-average concentrations for six criteria air pollutants (PM2.5, PM10, CO, NO2, SO2, O3) across the contiguous U.S. during a five-year period (2016-2020). We developed Land Use Regression (LUR) models within a partial-least-squares-universal kriging framework by incorporating several land use, geospatial and satellite-based predictor variables. The LUR models were validated using conventional and clustered cross-validation, with the former consistently showing superior performance in capturing the variability of air quality. Most models demonstrated reliable performance (e.g., mean squared error-based R 2 > 0.8, standardised root mean squared error < 0.1). We used the best modelling approach to develop estimates by Census Block, which were then population-weighted averaged at Census Block Group, Census Tract and County geographies. Our database provides valuable insights into the dynamics of air pollution, with utility for environmental risk assessment, public health, policy and urban planning.

Independent and joint effects of neighborhood-level environmental and socioeconomic exposures on body mass index in early childhood: The environmental influences on child health outcomes (ECHO) cohort

Past studies support the hypothesis that the prenatal period influences childhood growth. However, few studies explore the joint effects of exposures that occur simultaneously during pregnancy. To explore the feasibility of using mixtures methods with neighborhood-level environmental exposures, we assessed the effects of multiple prenatal exposures on body mass index (BMI) from birth to age 24 months. We used data from two cohorts: Healthy Start (n = 977) and Maternal and Developmental Risks from Environmental and Social Stressors (MADRES; n = 303). BMI was measured at delivery and 6, 12, and 24 months and standardized as z-scores. We included variables for air pollutants, built and natural environments, food access, and neighborhood socioeconomic status (SES). We used two complementary statistical approaches: single-exposure linear regression and quantile-based g-computation. Models were fit separately for each cohort and time point and were adjusted for relevant covariates. Single-exposure models identified negative associations between NO2 and distance to parks and positive associations between low neighborhood SES and BMI z-scores for Healthy Start participants; for MADRES participants, we observed negative associations between O3 and distance to parks and BMI z-scores. G-computations models produced comparable results for each cohort: higher exposures were generally associated with lower BMI, although results were not significant. Results from the g-computation models, which do not require a priori knowledge of the direction of associations, indicated that the direction of associations between mixture components and BMI varied by cohort and time point. Our study highlights challenges in assessing mixtures effects at the neighborhood level and in harmonizing exposure data across cohorts. For example, geospatial data of neighborhood-level exposures may not fully capture the qualities that might influence health behavior. Studies aiming to harmonize geospatial data from different geographical regions should consider contextual factors when operationalizing exposure variables.

A PM10 chemically characterized nation-wide dataset for Italy. Geographical influence on urban air pollution and source apportionment

Urban textures of the Italian cities are peculiarly shaped by the local geography generating similarities among cities placed in different regions but comparable topographical districts. This suggested the following scientific question: can different topographies generate significant differences on the PM10 chemical composition at Italian urban sites that share similar geography despite being in different regions? To investigate whether such communalities can be found and are applicable at Country-scale, we propose here a novel methodological approach. A dataset comprising season-averages of PM10 mass concentration and chemical composition data was built, covering the decade 2005-2016 and referring to urban sites only (21 cities). Statistical analyses, estimation of missing data, identification of latent clusters and source apportionment modeling by Positive Matrix Factorization (PMF) were performed on this unique dataset. The first original result is the demonstration that a dataset with atypical time resolution can be successfully exploited as an input matrix for PMF obtaining Country-scale representative chemical profiles, whose physical consistency has been assessed by different tests of modeling performance. Secondly, this dataset can be considered a reference repository of season averages of chemical species over the Italian territory and the chemical profiles obtained by PMF for urban Italian agglomerations could contribute to emission repositories. These findings indicate that our approach is powerful, and it could be further employed with datasets typically available in the air pollution monitoring networks.

The Impact of Exposure Measurement Error on the Estimated Concentration-Response Relationship between Long-Term Exposure to PM2.5 and Mortality

BACKGROUND

Exposure measurement error is a central concern in air pollution epidemiology. Given that studies have been using ambient air pollution predictions as proxy exposure measures, the potential impact of exposure error on health effect estimates needs to be comprehensively assessed.

OBJECTIVES

We aimed to generate wide-ranging scenarios to assess direction and magnitude of bias caused by exposure errors under plausible concentration-response relationships between annual exposure to fine particulate matter [PM ≤2.5μm in aerodynamic diameter (PM2.5)] and all-cause mortality.

METHODS

In this simulation study, we use daily PM2.5 predictions at 1-km2 spatial resolution to estimate annual PM2.5 exposures and their uncertainties for ZIP Codes of residence across the contiguous United States between 2000 and 2016. We consider scenarios in which we vary the error type (classical or Berkson) and the true concentration-response relationship between PM2.5 exposure and mortality (linear, quadratic, or soft-threshold-i.e., a smooth approximation to the hard-threshold model). In each scenario, we generate numbers of deaths using error-free exposures and confounders of concurrent air pollutants and neighborhood-level covariates and perform epidemiological analyses using error-prone exposures under correct specification or misspecification of the concentration-response relationship between PM2.5 exposure and mortality, adjusting for the confounders.

RESULTS

We simulate 1,000 replicates of each of 162 scenarios investigated. In general, both classical and Berkson errors can bias the concentration-response curve toward the null. The biases remain small even when using three times the predicted uncertainty to generate errors and are relatively larger at higher exposure levels.

DISCUSSION

Our findings suggest that the causal determination for long-term PM2.5 exposure and mortality is unlikely to be undermined when using high-resolution ambient predictions given that the estimated effect is generally smaller than the truth. The small magnitude of bias suggests that epidemiological findings are relatively robust against the exposure error. In practice, the use of ambient predictions with a finer spatial resolution will result in smaller bias. https://doi.org/10.1289/EHP10389.

Flexible Bayesian Ensemble Machine Learning Framework for Predicting Local Ozone Concentrations

3D-grid-based chemical transport models, such as the Community Multiscale Air Quality (CMAQ) modeling system, have been widely used for predicting concentrations of ambient air pollutants. However, typical horizontal resolutions of nationwide CMAQ simulations (12 × 12 km2) cannot capture local-scale gradients for accurately assessing human exposures and environmental justice disparities. In this study, a Bayesian ensemble machine learning (BEML) framework, which integrates 13 learning algorithms, was developed for downscaling CMAQ estimates of ozone daily maximum 8 h averages to the census tract level, across the contiguous US, and was demonstrated for 2011. Three-stage hyperparameter tuning and targeted validations were designed to ensure the ensemble model's ability to interpolate, extrapolate, and capture concentration peaks. The Shapley value metric from coalitional game theory was applied to interpret the drivers of subgrid gradients. The flexibility (transferability) of the 2011-trained BEML model was further tested by evaluating its ability to estimate fine-scale concentrations for other years (2012-2017) without retraining. To demonstrate the feasibility of using the BEML approach to strictly "data-limited" situations, the model was applied to downscale CMAQ outputs for a future-year scenario-based simulation that considers effects of variations in meteorology associated with climate change.

A hybrid approach to predict daily NO2 concentrations at city block scale

Estimating the ambient concentration of nitrogen dioxide (NO2) is challenging because NO2 generated by local fossil fuel combustion varies greatly in concentration across space and time. This study demonstrates an integrated hybrid approach combining dispersion modeling and land use regression (LUR) to predict daily NO2 concentrations at a high spatial resolution (e.g., 50 m) in the New York tri-state area. The daily concentration of traffic-related NO2 was estimated at the Environmental Protection Agency's NO2 monitoring sites in the study area for the years 2015-2017, using the Research LINE source (R-LINE) model with inputs of traffic data provided by the Highway Performance and Management System and meteorological data provided by the NOAA Integrated Surface Database. We used the R-LINE-predicted daily concentrations of NO2 to build mixed-effects regression models, including additional variables representing land use features, geographic characteristics, weather, and other predictors. The mixed model was selected by the Elastic Net method. Each model's performance was evaluated using the out-of-sample coefficient of determination (R2) and the square root of mean squared error (RMSE) from ten-fold cross-validation (CV). The mixed model showed a good prediction performance (CV R2: 0.75-0.79, RMSE: 3.9-4.0 ppb). R-LINE outputs improved the overall, spatial, and temporal CV R2 by 10.0%, 18.9% and 7.7% respectively. Given the output of R-LINE is point-based and has a flexible spatial resolution, this hybrid approach allows prediction of daily NO2 at an extremely high spatial resolution such as city blocks.

Exposure Measurement Error in Air Pollution Studies: The Impact of Shared, Multiplicative Measurement Error on Epidemiological Health Risk Estimates

Spatiotemporal air pollution models are increasingly being used to estimate health effects in epidemiological studies. Although such exposure prediction models typically result in improved spatial and temporal resolution of air pollution predictions, they remain subject to shared measurement error, a type of measurement error common in spatiotemporal exposure models which occurs when measurement error is not independent of exposures. A fundamental challenge of exposure measurement error in air pollution assessment is the strong correlation and sometimes identical (shared) error of exposure estimates across geographic space and time. When exposure estimates with shared measurement error are used to estimate health risk in epidemiological analyses, complex errors are potentially introduced, resulting in biased epidemiological conclusions. We demonstrate the influence of using a three-stage spatiotemporal exposure prediction model and introduce formal methods of shared, multiplicative measurement error (SMME) correction of epidemiological health risk estimates. Using our three-stage, ensemble learning based nitrogen oxides (NOx) exposure prediction model, we quantified SMME. We conducted an epidemiological analysis of wheeze risk in relation to NOx exposure among school-aged children. To demonstrate the incremental influence of exposure modeling stage, we iteratively estimated the health risk using assigned exposure predictions from each stage of the NOx model. We then determined the impact of SMME on the variance of the health risk estimates under various scenarios. Depending on the stage of the spatiotemporal exposure model used, we found that wheeze odds ratio ranged from 1.16 to 1.28 for an interquartile range increase in NOx. With each additional stage of exposure modeling, the health effect estimate moved further away from the null (OR=1). When corrected for observed SMME, the health effects confidence intervals slightly lengthened, but our epidemiological conclusions were not altered. When the variance estimate was corrected for the potential "worst case scenario" of SMME, the standard error further increased, having a meaningful influence on epidemiological conclusions. Our framework can be expanded and used to understand the implications of using exposure predictions subject to shared measurement error in future health investigations.

Air Pollution Monitoring for Health Research and Patient Care. An Official American Thoracic Society Workshop Report

Air quality data from satellites and low-cost sensor systems, together with output from air quality models, have the potential to augment high-quality, regulatory-grade data in countries with in situ monitoring networks and provide much-needed air quality information in countries without them. Each of these technologies has strengths and limitations that need to be considered when integrating them to develop a robust and diverse global air quality monitoring network. To address these issues, the American Thoracic Society, the U.S. Environmental Protection Agency, the National Aeronautics and Space Administration, and the National Institute of Environmental Health Sciences convened a workshop in May 2017 to bring together global experts from across multiple disciplines and agencies to discuss current and near-term capabilities to monitor global air pollution. The participants focused on four topics: 1) current and near-term capabilities in air pollution monitoring, 2) data assimilation from multiple technology platforms, 3) critical issues for air pollution monitoring in regions without a regulatory-quality stationary monitoring network, and 4) risk communication and health messaging. Recommendations for research and improved use were identified during the workshop, including a recognition that the integration of data across monitoring technology groups is critical to maximizing the effectiveness (e.g., data accuracy, as well as spatial and temporal coverage) of these monitoring technologies. Taken together, these recommendations will advance the development of a global air quality monitoring network that takes advantage of emerging technologies to ensure the availability of free, accessible, and reliable air pollution data and forecasts to health professionals, as well as to all global citizens.

Cluster-based bagging of constrained mixed-effects models for high spatiotemporal resolution nitrogen oxides prediction over large regions

BACKGROUND

Accurate estimation of nitrogen dioxide (NO2) and nitrogen oxide (NOx) concentrations at high spatiotemporal resolutions is crucial for improving evaluation of their health effects, particularly with respect to short-term exposures and acute health outcomes. For estimation over large regions like California, high spatial density field campaign measurements can be combined with more sparse routine monitoring network measurements to capture spatiotemporal variability of NO2 and NOx concentrations. However, monitors in spatially dense field sampling are often highly clustered and their uneven distribution creates a challenge for such combined use. Furthermore, heterogeneities due to seasonal patterns of meteorology and source mixtures between sub-regions (e.g. southern vs. northern California) need to be addressed.

OBJECTIVES

In this study, we aim to develop highly accurate and adaptive machine learning models to predict high-resolution NO2 and NOx concentrations over large geographic regions using measurements from different sources that contain samples with heterogeneous spatiotemporal distributions and clustering patterns.

METHODS

We used a comprehensive Kruskal-K-means method to cluster the measurement samples from multiple heterogeneous sources. Spatiotemporal cluster-based bootstrap aggregating (bagging) of the base mixed-effects models was then applied, leveraging the clusters to obtain balanced and less correlated training samples for less bias and improvement in generalization. Further, we used the machine learning technique of grid search to find the optimal interaction of temporal basis functions and the scale of spatial effects, which, together with spatiotemporal covariates, adequately captured spatiotemporal variability in NO2 and NOx at the state and local levels.

RESULTS

We found an optimal combination of four temporal basis functions and 200 m scale spatial effects for the base mixed-effects models. With the cluster-based bagging of the base models, we obtained robust predictions with an ensemble cross validation R2 of 0.88 for both NO2 and NOx [RMSE (RMSEIQR): 3.62 ppb (0.28) and 9.63 ppb (0.37) respectively]. In independent tests of random sampling, our models achieved similarly strong performance (R2 of 0.87-0.90; RMSE of 3.97-9.69 ppb; RMSEIQR of 0.21-0.27), illustrating minimal over-fitting.

CONCLUSIONS

Our approach has important implications for fusing data from highly clustered and heterogeneous measurement samples from multiple data sources to produce highly accurate concentration estimates of air pollutants such as NO2 and NOx at high resolution over a large region.

Study Design, Protocol and Profile of the Maternal And Developmental Risks from Environmental and Social Stressors (MADRES) Pregnancy Cohort: a Prospective Cohort Study in Predominantly Low-Income Hispanic Women in Urban Los Angeles

BACKGROUND

The burden of childhood and adult obesity disproportionally affects Hispanic and African-American populations in the US, and these groups as well as populations with lower income and education levels are disproportionately affected by environmental pollution. Pregnancy is a critical developmental period where maternal exposures may have significant impacts on infant and childhood growth as well as the future health of the mother. We initiated the "Maternal And Developmental Risks from Environmental and Social Stressors (MADRES)" cohort study to address critical gaps in understanding the increased risk for childhood obesity and maternal obesity outcomes among minority and low-income women in urban Los Angeles.

METHODS

The MADRES cohort is specifically examining whether pre- and postpartum environmental exposures, in addition to exposures to psychosocial and built environment stressors, lead to excessive gestational weight gain and postpartum weight retention in women and to perturbed infant growth trajectories and increased childhood obesity risk through altered psychological, behavioral and/or metabolic responses. The ongoing MADRES study is a prospective pregnancy cohort of 1000 predominantly lower-income, Hispanic women in Los Angeles, CA. Enrollment in the MADRES cohort is initiated prior to 30 weeks gestation from partner community health clinics in Los Angeles. Cohort participants are followed through their pregnancies, at birth, and during the infant's first year of life through a series of in-person visits with interviewer-administered questionnaires, anthropometric measurements and biospecimen collection as well as telephone interviews conducted with the mother.

DISCUSSION

In this paper, we outline the study rationale and data collection protocol for the MADRES cohort, and we present a profile of demographic, health and exposure characteristics for 291 participants who have delivered their infants, out of 523 participants enrolled in the study from November 2015 to October 2018 from four community health clinics in Los Angeles. Results from the MADRES cohort could provide a powerful rationale for regulation of targeted chemical environmental components, better transportation and urban design policies, and clinical recommendations for stress-coping strategies and behavior to reduce lifelong obesity risk.

Exposure measurement error in air pollution studies: A framework for assessing shared, multiplicative measurement error in ensemble learning estimates of nitrogen oxides

BACKGROUND

Increasingly ensemble learning-based spatiotemporal models are being used to estimate residential air pollution exposures in epidemiological studies. While these machine learning models typically have improved performance, they suffer from exposure measurement error that is inherent in all models. Our objective is to develop a framework to formally assess shared, multiplicative measurement error (SMME) in our previously published three-stage, ensemble learning-based nitrogen oxides (NO_x) model to identify its spatial and temporal patterns and predictors.

METHODS

By treating the ensembles as an external dosimetry system, we quantified shared and unshared, multiplicative and additive (SUMA) measurement error components in our exposure model. We used generalized additive models (GAMs) with a smooth term for location to identify geographic locations with significantly elevated SMME and explain their spatial and temporal determinants.

RESULTS

We found evidence of significant shared and unshared multiplicative error (p < 0.0001) in our ensemble-learning based spatiotemporal NO_x model predictions. Unshared multiplicative error was 26 times larger than SMME. We observed significant geographic (p < 0.0001) and temporal variation in SMME with the majority (43%) of predictions with elevated SMME occurring in the earliest time-period (1992-2000). Densely populated urban prediction regions with complex air pollution sources generally exhibited highest odds of elevated SMME.

CONCLUSIONS

We developed a novel statistical framework to formally evaluate the magnitude and drivers of SMME in ensemble learning-based exposure models. Our framework can be used to inform building future improved exposure models.

Machine Learning Approaches for Outdoor Air Quality Modelling: A Systematic Review

Current studies show that traditional deterministic models tend to struggle to capture the non-linear relationship between the concentration of air pollutants and their sources of emission and dispersion. To tackle such a limitation, the most promising approach is to use statistical models based on machine learning techniques. Nevertheless, it is puzzling why a certain algorithm is chosen over another for a given task. This systematic review intends to clarify this question by providing the reader with a comprehensive description of the principles underlying these algorithms and how they are applied to enhance prediction accuracy. A rigorous search that conforms to the PRISMA guideline is performed and results in the selection of the 46 most relevant journal papers in the area. Through a factorial analysis method these studies are synthetized and linked to each other. The main findings of this literature review show that: (i) machine learning is mainly applied in Eurasian and North American continents and (ii) estimation problems tend to implement Ensemble Learning and Regressions, whereas forecasting make use of Neural Networks and Support Vector Machines. The next challenges of this approach are to improve the prediction of pollution peaks and contaminants recently put in the spotlights (e.g., nanoparticles).

Mining Public Datasets for Modeling Intra-City PM2.5 Concentrations at a Fine Spatial Resolution

Air quality models are important for studying the impact of air pollutant on health conditions at a fine spatiotemporal scale. Existing work typically relies on area-specific, expert-selected attributes of pollution emissions (e,g., transportation) and dispersion (e.g., meteorology) for building the model for each combination of study areas, pollutant types, and spatiotemporal scales. In this paper, we present a data mining approach that utilizes publicly available OpenStreetMap (OSM) data to automatically generate an air quality model for the concentrations of fine particulate matter less than 2.5 μm in aerodynamic diameter at various temporal scales. Our experiment shows that our (domain-) expert-free model could generate accurate PM2.5 concentration predictions, which can be used to improve air quality models that traditionally rely on expert-selected input. Our approach also quantifies the impact on air quality from a variety of geographic features (i.e., how various types of geographic features such as parking lots and commercial buildings affect air quality and from what distance) representing mobile, stationary and area natural and anthropogenic air pollution sources. This approach is particularly important for enabling the construction of context-specific spatiotemporal models of air pollution, allowing investigations of the impact of air pollution exposures on sensitive populations such as children with asthma at scale.