A new study from the University of Utah has found that repeated exposure to mixtures of common air pollutants during early pregnancy is associated with nearly three times greater odds of early preterm birth. Using machine learning, researchers identified that even moderate levels of multiple pollutants occurring together can increase health risks, potentially challenging current air quality assessment methods.
The research analyzed data from 44,874 first-time mothers in Utah. By utilizing an epidemiologic machine learning framework known as a self-organizing map, the team was able to identify patterns within complex environmental exposures that would otherwise be difficult to detect.
Researchers examined high-resolution air pollution data from Utah between 2013 and 2016, specifically looking at temperature alongside three pollutants: nitrogen dioxide (NO2), ozone (O3), and fine particulate matter (PM2.5). The algorithm identified 12 distinct mixtures of these elements.
A significant finding involved the late first trimester, where a mixture of ozone and fine particulate matter showed the strongest relationship to early preterm birth. Specifically, women exposed to this mixture during week 11 of pregnancy faced 53% greater odds of preterm birth later in the pregnancy. Furthermore, those experiencing repeated exposures between weeks 9 and 14 saw nearly triple the odds of preterm birth.
The study highlights a potential gap in current public health monitoring. Because the Air Quality Index (AQI) typically focuses on the single pollutant posing the greatest harm, the EPA might classify certain levels as safe even when mixtures of pollutants pose risks.
Lead author Brenna Kelly noted that because people are exposed to multiple substances simultaneously and repeatedly, the combination of slightly elevated chemicals may be significant. Michelle Debbink, an associate professor of obstetrics and gynecology at the University of Utah and study coauthor, stated that early pregnancy is a critical window because the placenta and blood-supplying arteries are still developing.
Debbink explained that exposure during this period could impair development or cause inflammation and damage that accumulates over time, leading to complications like preeclampsia which may necessitate preterm delivery.
The researchers expressed hope that this machine learning framework will be used to study how other environmental hazards impact human health. Coauthor Simon Brewer, a professor in the University of Utah's School of Environment, Society & Sustainability, noted that the work demonstrates the potential of artificial intelligence in addressing complex environmental problems and assessing their impacts.
