Short-Term PM2.5 Exposure and Health Impacts: Insights from the AMRIT Low-Cost Sensor Network in the Indo-Gangetic Plains of India

Air pollution remains a critical global challenge, exerting significant health and economic burdens, particularly in developing nations. A key obstacle to effective air quality management is the limited availability of monitoring infrastructure, compounded by insufficient spatial coverage. This shortfall is largely due to the high costs associated with establishing and maintaining regulatory air quality monitoring stations. However, the advent of Internet-of-Things (IoT)-based monitoring devices equipped with low-cost sensors (LCS) has opened new avenues for air quality assessment. These devices, which leverage cellular networks for connectivity, offer a cost-effective alternative and can be deployed across geographic locations. Over India, this development presents a valuable opportunity to enhance our understanding of air quality beyond the urban regions, where monitoring stations are scarce.
In this context, a state-wide sensor-based Ambient Air Quality Monitoring (SAAQM) network has been established under the project “Ambient Air Quality Monitoring Over Rural Areas using Indigenous Technology” (AMRIT) to investigate the current levels of fine particulate matter (PM_2.5) pollution across the state of Bihar, located in the eastern part of Indo-Gangetic Plains, India. This initiative has expanded the air quality monitoring infrastructure from 35 regulatory-grade monitors to 539 SAAQM nodes, enhancing monitoring infrastructure coverage by a factor of 15. This enables a more granular assessment of PM_2.5 exposure and associated health impacts, reaching down to the census tract level.
To generate continuous sub-daily PM_2.5 concentration maps at a spatial resolution of 0.5 km², we employed a machine learning framework that integrates meteorological variables and satellite imagery-based predictors with SAAQM observations. Our findings indicate that the dense SAAQM network is considerably more effective in capturing localized variations in PM_2.5 exposure, which are often overlooked by other high-resolution global datasets. We found significant spatiotemporal heterogeneity in the PM_2.5 exposure distribution, with elevated PM_2.5 exposure levels over the northern tracts of Bihar. The population-weighted concentration ranged from 55 (±14) μg/m³ in pre-monsoon (March-May), 110 (±30) μg/m³ in post-monsoon (October-November), and 136 (±40) μg/m³ in winter months (December-February). Furthermore, we evaluated the associated health impacts using an India-specific exposure-response function, identifying a short-term mortality rate of 50 per lakh population (95% CI: 33–64) in the region. These findings emphasize the critical importance of hyperlocal air quality monitoring in understanding PM_2.5 exposure and mitigating its health risks.