Utility of a Low-Cost, Dense Sensor Network for the Study of Air Quality Impact upon Human Health in an Urban Area

Many varieties of healthcare data can be used in studies of the public health impact of air quality changes. The spatiotemporal resolution of both healthcare and air quality data may vary considerably, presenting difficulties in matching the datasets for detailed study that is often desired. For example, inpatient admissions data for hospitals is frequently available at the zip code level, but air quality data rarely has similar spatial heterogeneity, particularly in urban regions that may have complex gradients of pollutants from multiple sources. The construction of a dense sensor monitoring network using less costly air quality sensors, combined with satellite and in situ observations that meet federal air quality surveillance requirements, could be utilized to address this information gap. Concerns regarding the accuracy of data collected by low-cost sensors, which may vary according to sensor manufacturer, pollutant type or source, and meteorological conditions, create barriers for use of such data in health/air quality impact studies.

Birmingham, Alabama (pop. 212,167), ranks 9^th in the list of U.S. metropolitan areas for asthma prevalence. The city is influenced by pollution from natural (smoke from prescribed wildfires in the region) and human (industrial) sources. High-resolution spatial data on healthcare resource utilization in adults ages 65 years and older is available, but the current surface-level monitoring network does not collect samples at the same degree of spatial granularity. Study investigators have partnered with a grass-roots advocacy organization to construct a dense AQ monitoring network using low-cost sensors (for gas and particulate matter) in the city. The gas sensor suite consists of Alphasense ozone, nitrogen-dioxide, nitric oxide and carbon-monoxide sensors. A variety of products used for monitoring fine particulate matter (Sharp, Plantower, Alphasense) will be tested. The sensor network will be hosted by community volunteers that have been recruited in partnership with the advocacy group. Plans are to site one of the sensor nodes at the same site as existing pollution monitoring sites.

Low-cost sensors have demonstrated high variability in performance between sensors in past studies, due to factors such as meteorological conditions and proximity to urban emission sources. Multiple methodologies for accurate field calibration of the low-cost sensors in the Birmingham area will be discussed, along with statistical methodologies and techniques that can be used to incorporate these data into health studies for the community population. Results of the project will support the need for community-level AQ data that may be used by the advocacy group along with policy- and decision-makers to raise awareness and engage community members in strategies for sustainable improvements in local air quality.