How to make roads safer? New AI research

Many people use Google Street View to get directions. Dr. Quynh Nguyen, an epidemiologist and statistician at the University of Maryland School of Public Health, uses it to find sudden dead ends along the way. According to a study published on June 6, British Medical Journal (BMJ) of Preventing injuriesNguyen will use AI tools to identify key environmental factors that influence not only motor vehicle-related crashes, but also bicycle- and pedestrian-related accidents.

“Motor vehicle crashes are the leading cause of death among young people ages 5 to 29, so it's important to understand how the physical environment increases or decreases fatal crashes and which communities are most affected,” said Nguyen, a professor who works to address health disparities using technology and big data.

Nguyen and his colleagues used Google Street View (GSV), an AI tool that provides a 360-degree view of roads around the world, to determine the relationship between car crashes and the built environment where the crashes occurred. The researchers used virtual mapping to examine specific road features, such as street lights and green spaces, on a national scale.

“Because we were able to analyze a huge amount of GSV data from across the country, we were able to get precise results on which architectural elements affect car accidents. It was clear that areas with more greenery, streetlights, single-lane roads, and sidewalks had fewer fatal car accidents,” Nguyen said.

Sidewalks had the biggest impact on reducing accidents: areas with more sidewalks saw a 70% drop in traffic accidents, and areas with one-lane roads (common in rural areas) saw a 50% drop in accidents.

For pedestrians and cyclists, street lights and stop signs increased safety and led to fewer car crashes for both groups. Conversely, areas with ongoing road construction had a negative impact, with more crashes.

“Many of the public health problems facing communities are solvable,” said Xiaohe Yue, a data analyst in the UMD School of Public Health (SPH) and co-author of the study. “Emerging technologies and access to a broad range of data sources have helped us find solutions to some of the public health problems that plague our population.”

The researchers hope that their findings will provide decision makers with proven, practical options to improve road safety for drivers, pedestrians and cyclists, and inform transport and infrastructure policy.

“We hope that our research will encourage urban planners and developers to consider the built environment more carefully and design safer roads and communities,” said co-author Helan Manet, a data analyst who works with Yue at SPH.

Nguyen sees entirely new research avenues emerging.

“Data science and AI are increasingly being used to enable large-scale, efficient and timely studies like this one,” Nguyen said. “This study is one example of how AI can be used to improve public health, and I'm confident there will be much more to come.”

Nguyen and his colleagues aim to expand the types of built environment indicators surveyed across the U.S., as well as explore these characteristics in other countries.