Abstract:OPS 03: Machine learning in environmental epidemiology, Room 315, Floor 3, August 26, 2019, 4:30 PM - 5:30 PM Background: NO2 is an air pollutant that leads to multiple adverse health outcomes. Various modeling approaches have been proposed to estimate NO2, using statistical regressions, machine learning algorithms, hybrid models and other approaches, with predictor variables ranging from land-use terms, satellite-derived column NO2 concentration, and meteorological variables. Those complementary fitting methods and predictor variables have the potential to improve model performance. However, few studies have been proposed to integrate multiple fitting methods to estimate NO2. Methods: We propose an ensemble model to integrate multiple machine learning algorithms, including neural network, random forest, and gradient boosting, with a variety of predictor variables as input variables. This NO2 model covers the entire contiguous United States from 2000 to 2016. After model training, we predicted daily NO2 levels at 1 km × 1 km grid cells, as well as associated monthly uncertainty level. We also downscaled the 1-km-level prediction to 100-meter-level. Results: After cross-validation, this ensemble-based model produced good R2, with mean R2 0.788, and mean spatial R2 0.844. The relationship between daily monitored NO2 and predicted NO2 is almost linear. The distribution of NO2 exhibits clear spatial clustering, with high concentration clustering around urban areas, especially major cities, and along highways. Temporally, NO2 level underwent a profound decline over the study area, with annual level in 2016 about 50% of the 2000 level. Conclusion: This NO2 estimation has very high spatiotemporal resolution (daily and 1 km × 1 km), covers a large spatial area (contiguous United States), and provides good exposure assessment to epidemiologists to analyze the long-term and short-term health effect of NO2. We also conclude that the most appropriate predictor variables and fitting algorithm are context-based. It is time to consider how to integrate different predictor variables and fitting algorithms together and achieve an optimized modeling for air pollution estimation.

Leveraging low-cost sensors to predict nitrogen dioxide for epidemiologic exposure assessment

National Scale Spatiotemporal Land-Use Regression Model for PM2.5, PM10 and NO2 Concentration in China

Low-cost sensor networks and land-use regression: Interpolating nitrogen dioxide concentration at high temporal and spatial resolution in Southern California

Modeling fine-grained spatio-temporal pollution maps with low-cost sensors

Forecasting Smog-Related Health Hazard Based on Social Media and Physical Sensor.

The rise of low-cost sensing for managing air pollution in cities

Developing high-resolution PM2.5 exposure models by integrating low-cost sensors, automated machine learning, and big human mobility data

Assessing NO2 Exposures with High Spatiotemporal Resolution Across the Contiguous United States Using Ensemble Model

Leveraging Citizen Science and Low-Cost Sensors to Characterize Air Pollution Exposure of Disadvantaged Communities in Southern California

Low-cost PM2.5 sensors can help identify driving factors of poor air quality and benefit communities

From air quality sensors to sensor networks: Things we need to learn

Characterising Low-Cost Sensors in Highly Portable Platforms to Quantify Personal Exposure in Diverse Environments

Air Pollution Exposure Monitoring Using Portable Low-Cost Air Quality Sensors

Machine Learning for a Low-cost Air Pollution Network

Using low-cost sensor technologies and advanced computational methods to improve dose estimations in health panel studies: results of the AIRLESS project

Exposure prediction approaches used in air pollution epidemiology studies: Key findings and future recommendations

Investigating Use of Low-Cost Sensors to Increase Accuracy and Equity of Real-Time Air Quality Information

Modeling spatial variation of gaseous air pollutants and particulate matters in a Metropolitan area using mobile monitoring data

Comprehensive Monitoring of Air Pollution Hotspots Using Sparse Sensor Networks

A high-resolution computationally-efficient spatiotemporal model for estimating daily PM2.5 concentrations in Beijing, China

Feasibility of low-cost particulate matter sensors for long-term environmental monitoring: Field evaluation and calibration