Urban air quality modeling and health impact analysis using geospatial methods and machine learning algorithms

This study utilized geospatial techniques and machine learning (ML) algorithms, viz. Random Forest and XGBoost, for predicting the air quality and the AirQ+ model for assessing health risks in urban environments. We analyzed the annual variations in sulfur dioxide (SO₂) and nitrogen dioxide (NO₂) levels of five Indian metropolitan cities from 2019 to 2022. Preliminary analysis indicated the highest levels of NO₂ and SO₂ in Delhi and Kolkata as compared to other metropolises. Kolkata had an 11% increase in SO₂ concentrations in 2022 compared to 2019, while Delhi had a 20% increase in NO₂ concentrations in 2022 compared to 2019. The air pollutant levels predicted by ML algorithms were analyzed in the AirQ+ model for health risks. The health impact assessment conducted using the AirQ+ model revealed concerning trends. In 2023, particulate matter (PM_2.5) was attributed to 20.26% of respiratory disease cases per 100,000 population in Delhi, followed by NO₂, accounting for 11.01%. In Kolkata, SO₂ was responsible for 3.21% of respiratory disease cases. By implementing this approach, policymakers can estimate the air pollution levels and potential respiratory disease health risks. This knowledge can help them formulate targeted interventions, such as implementing pollution control measures, managing health risks, and issuing health advisories, to protect public health and improve air quality in cities.