Exploring the association between smartphone-based place visitation data and neighborhood-level coronary heart disease in the United States

Abstract

Coronary Heart Disease (CHD) is the leading cause of death in the United States, affecting over 20.5 million adults. Previous studies link health behaviors – such as dietary behavior, physical activity, smoking, and alcohol consumption – to CHD risk. These studies typically use surveys and interviews, which, despite their benefits, are resource-intensive and limited by small sample sizes. Using large-scale national level anonymized smartphone-based location data, our study examines whether health behaviors that are proxy measured by place visitation are associated with CHD prevalence across US census tracts. This study utilized data from multiple sources, including demographic and socioeconomic characteristics, health outcomes, and smartphone-based place visitation data. Health behavior measures were derived from aggregated smartphone location data at the census tract level, focusing on categories such as food retails, drinking places, and physical activity locations. Three sets of regression analyses were conducted: one using only demographic variables, the second including socioeconomic variables, and another incorporating the derived health behavior measures. Linear and spatial regression analyses were employed to assess the relationship between neighborhood-level CHD prevalence and these behaviors. Findings indicate a significant association between health behaviors that are proxy measured by place visitation data and the prevalence of CHD at the neighborhood level. The models incorporating these behaviors demonstrated improved fitness and highlighted specific behavioral factors such as increased visits to physical activity facilities and healthy food retail associated with lower CHD rates. Conversely, higher visits to less healthy food retail were associated with increased CHD rates. Smartphone-based visitation data offers a novel method to assess health behaviors at a large scale, providing valuable insights for targeting CHD interventions more effectively at the neighborhood level. This approach could enhance our understanding and management of CHD, informing public health strategies and interventions to mitigate this major health challenge.