Some public health problems are so large that they can be calculated from space using satellite images using a special algorithm compiled by artificial intelligence. Millions of people in the world are obese, in the United States it is a third of the adult population. Scientists were puzzled by the problem of its active distribution and decided to study the issue using new technologies.
Obesity is influenced by many factors, including genetics, demographics, behavior, etc. Behavioral traits that encourage unhealthy food choices and sedentary lifestyles appear to be reflected in the social and built environment. The number of fast food outlets and the presence of green areas with pedestrian paths have a real impact on the lifestyle of people. It is this quantitative relationship between behavioral traits and infrastructure that scientists decided to study, according to the JAMA Network.
At the same time, in a new study, scientists assessed the level of obesity in a specific region, but overweight townspeople were not involved in the experiments! Instead, they turned to artificial intelligence for help. The experts created an algorithm that relied on signals from the environment and infrastructure of the area in which these people live. Data on the prevalence of obesity among adults they received from the Centers for Disease Control and Prevention "500 Cities" project. The model takes into account "points of interest" - gas stations, shopping centers, parks, eateries, etc. It can be used to judge where and how often people go, and what factors of the urban environment are associated with obesity.
To find the best way to create the algorithm, the researchers loaded 150,000 Google Maps satellite images into the neural network (a technology aimed at efficient image recognition). The data covered 1,695 neighborhoods in six US cities: Los Angeles, Memphis, San Antonio, Seattle, Tacoma, and Bellevue. The study was conducted from February 14 to October 31, 2017. Then, among a vast array of information, they identified patterns and paths of movement of people. The network helped researchers focus on the most important features of the images, such as the number of green spaces, fitness clubs, cafes, fast food eateries, different types of housing, and more.
As a result, the researchers estimated the "obesity area" in cities even better than statistics would. Research conclusion: on average, 64.8% of obesity prevalence cases are associated with a specific urban infrastructure. That is, in order to somehow get away from an unhealthy lifestyle, it is necessary to develop green areas, build pedestrian paths and arrange sports grounds in open areas. There is nothing fundamentally new in the study, but there is confirmation of the fact that the environment largely determines the health of the population.