Study finds racial and ethnic differences in air quality locations in the US

US Environmental Protection Agency (EPA) Air Quality monitors are disproportionately located in predominantly white neighborhoods, according to research from the University of Utah. The EPA network consistently failed to catch air quality in color communities in six large pollutants, in particular lead and sulfur dioxide, followed by ozone and carbon monoxide.

EPA -controlling monitors are the most important data source that stimulates decisions about reducing pollution, city planning and initiatives for public health. Without equal monitor distribution, the data on pollution concentrations can incorrectly display, which means that marginalized groups are at risk.

“It is the question behind the question. Researchers, policy makers, we all use data about air quality, but whose air measures it?” said Brenna Kelly, PhD student at the U and main author of the research. “Although these data are of high quality, that does not mean that it is high quality for everyone.”

Research has shown that marginalized communities have the highest exposure rates of air pollution, but the studies assumed that the data equalized all neighborhoods. This study is the first to assess differences in the monitors themselves for all American census groups, in a neighborhood scale. Although there were differences for all non-white groups, were the largest for native Hawaiians and other Pacific Islanders, followed by the indigenous population of the Indian and Alaska.

Research and analysis of air quality often require artificial intelligence (AI) tools to process the enormous amounts of data. Although bias is known in AI algorithms, the study is a different ethical problem for BIG-data users-the chance that the data sets themselves are inherently biased.

“If there was an inequality for just one type of monitor, it might be coincidental or just bad design,” said co-author Simon Brewer, university teacher Geography and executive committee of the One-U’s responsions AI initiative of the U. fairly divided. “

The study was published in Jama Network Open.

Air quality is hyperlocal and can change dramatically from street to street. The authors mapped monitor locations and neighborhood demography to the level of the census, one of the smallest units of the American Census Bureau for residential patterns. With the help of the EPA air quality system -regulating monitoring repository, they identified monitors for six large air pollutants that are dangerous for human health – lead, ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide and particles.

They used the census of the American Community Survey from 2022 to estimate the racial and ethnic composition for every census block in the country. The researchers found systemic monitoring differences for each of the pollutants of the criteria. Compared to the white non-Spanish population, all groups were associated with less lead, ozone, nitrogen dioxide and particle monitors.

Kelly became curious about the EPA monitoring network of the EPA, while continuing her doctoral research in public health sciences, aimed at the risks of exposure to air pollution to pregnant people. Epidemiology research identifies the factors that contribute to diseases within a population. So far there is an underlying assumption that the data everywhere represent air quality problems, she explained.

“It is not only that we miss one type of pollutants for one group, it is that we understand less about everything for all these groups. That is worrying,” said Kelly. “If I want to relate exposure to air pollution to a disease, I have to measure it well. If I have a better understanding of the air quality for a group of people, it will result in biased results.”

Air quality and population of the population are only two of the many areas that struggle with the challenges of using Big Data and AI in a responsible manner. The one-u responsible for AI initiative is the recent effort of yours to bring experts together to develop best practices.

“This study is particularly relevant in an increasingly data-driven society. One of the goals of the responsible AI initiative is to study the fair application of artificial intelligence methods,” said Brewer. “Our results suggest that prejudices in the data can be just as important to consider as any algorithmic bias.”