Can big data and artificial intelligence help policymakers tackle population health issues? Google Street View could help detect diseases and identify trends to improve policy response times, according to a new study from researchers at New York University.
In the context of work published in Proceedings of the National Academy of Sciencesthe authors analyzed two million images taken from the Google platform showing the American metropolis, in order to evaluate their usefulness in a public health context.
Relying on these snapshots alone could lead to inaccurate information and erroneous interventions, they say, but combining this information with other data increases its potential tenfold.
“There is a lot of excitement about using new data sources to get a holistic view of health, including the integration of machine learning and data science methods,” said Rumi Chunara, one of the study’s lead authors.
“Our study highlights the potential of digital data sources, such as images from Street View, to improve public health research, while also highlighting the limitations of the data and the complex dynamics at work between the environment, individual behaviors, and public health outcomes,” added Miao Zhang, the study’s first author.
Health at the street level
In recent years, researchers have begun collecting images of various streets in order to link a specific environment, and its infrastructure, to specific issues such as mental health problems, the transmission of infectious diseases, or obesity; a task that would be particularly difficult to do manually.
“We know that the built environment of a city can affect our health, whether it’s the availability of sidewalks and green spaces for walking, or the existence of grocery stores that offer healthy foods,” Chunara said.
“Some studies indicate that the presence of sidewalks is linked to lower obesity rates… But is that all?”
Analyzing the millions of images, the researchers used artificial intelligence to detect sidewalks and crosswalks. They then compared that information to local data on obesity, diabetes and the proportion of citizens who exercise, all from the Centers for Disease Control and Prevention. The goal? To determine whether the built environment predicts public health outcomes.
The authors of the study found that neighborhoods with more pedestrian crossings also had lower rates of obesity and diabetes, but there does not appear to be a link between the presence of sidewalks and health effects, unlike previous studies.
“This could be because a lot of the sidewalks in New York are located in places where people don’t use them, like along a highway, on a bridge, in a tunnel – and so sidewalk density might not reflect walking capacity as effectively as crosswalks,” Zhang said.
Furthermore, relying on AI to detect sidewalks and pedestrian crossings has led to a series of errors compared to official documents concerning these developments. According to the researchers, this can be explained by the presence of a car in the wrong place, in a photo, or by shadows obscuring infrastructure.
What is the cause and effect relationship?
If crosswalks have indeed been linked to lower rates of obesity and diabetes, the researchers wanted to know what might explain this association. Their analysis of CDC data revealed that physical activity, not just the crosswalks, was responsible for the lower proportion of people suffering from these ailments.
In a test, the authors found that increasing physical activity led to a four-times greater drop in obesity rates and a 17-times greater drop in diabetes rates than simply installing more pedestrian crossings.
Based on their findings, the researchers believe that creating public health policies should not only rely on new sources of data, but also take into account other evidence.
“A more nuanced approach, linked to the big datais necessary to use this data, in particular by drawing on existing expertise,” judges Mr. Chunara, who suggests rather the addition of physical activity classes, for example, rather than the construction of urban infrastructures, in order to combat obesity and diabetes.
