What makes roads safer? New study uses AI to find out

Most people use Google Street View to find their way. Dr. Quynh Nguyen, an epidemiologist and statistician at the University of Maryland School of Public Health, uses it to pinpoint places where your trip could end abruptly. In a study published June 6 in the British medical journal (BMJ) by Injury preventionNguyen identifies the use of AI tools to identify key environmental elements that influence car-related collisions, as well as bicycle and pedestrian-related crashes.

“Car accidents are the leading cause of death for young people between the ages of 5 and 29. So it is critical to understand how the physical environment can increase or decrease the number of fatal collisions and which communities are most affected,” said Nguyen, a professor whose work uses technology and big data sources to examine inequalities health area.

Nguyen and fellow researchers used Google Street View (GSV), an AI tool that provides 360-degree images of streets around the world, to determine the relationship between car crashes and the built environment in locations where accidents occur. Using virtual maps, researchers examined specific road features, such as street lighting or greenery, on a national scale.

“Because we were able to collect such a large amount of GSV data from across the country, we got accurate results based on which built-up features influence car crashes. It was clear that places with more greenery, street lighting, single-lane roads and sidewalks associated with fewer fatal car crashes,” says Nguyen.

Sidewalks had the greatest impact on reducing accidents. Places with more sidewalks had 70% fewer traffic accidents, and places with a single lane, which are often found in rural areas, had 50% fewer accidents.

For pedestrians and cyclists, street lights and stop signs offered greater safety; they were associated with fewer car crashes involving both groups. Conversely, areas with road construction had a negative effect with more collisions.

“Many of the public health problems that communities face are often solvable,” said Xiaohe Yue, a data analyst at the UMD School of Public Health (SPH) and co-author of the study. “Emerging technologies and access to extensive data sources have been instrumental in finding solutions to some of the public health challenges facing populations.”

Researchers hope the findings will inform transportation and infrastructure policy by providing proven practical options for decision makers to improve road safety for motorists, pedestrians and cyclists.

“We hope our work will encourage urban planners and developers to think more carefully about the built environment and thus design safer streets and communities,” said co-author Heran Mane, a data analyst working with Yue at SPH.

Nguyen sees the emergence of a completely new research path.

“We are seeing an increase in the use of data science and AI to enable larger, more efficient and timely studies like this,” said Nguyen. “This research is a demonstration of how we can use AI to improve public health, and we know there is much more to come.”

Nguyen and colleagues want to expand the types of built environment indicators examined in the United States, and also examine these characteristics in other countries.