Explaining provincial disease network via graph-theoretic analysis of disease occurrence in Iran

Purpose: Social networks are increasingly being used to study the epidemiology of human diseases. These methods are usually limited to studying a single or a small group of diseases within a small community and may not fully capture effects of environmental factors that drift the epidemiology of diseases at the community level (rather than individual level). By introducing an ecological variant of the social network, we described provincial disease network (PDN) to study the similarities in regional occurrence of diseases. Methods: In this network, nodes (provinces) are connected via edges together. Provinces that have similar pattern of disease prevalence and/or incidence tie stronger together. We sought to find modular organization of Iran's PDN and to identify factors (literacy, rural population percentage, smoking, geographical distance, and age distribution) that could predict the strength of interprovincial connections in the PDN. Results: Provinces in Iran's PDN were segregated into five different modules. Geographic distance, differences in the literacy percentage and percentage of population in the 0-16 years age group showed significant inverse correlation with strength of connections in Iran's PDN. Conclusions: Network-based approaches could provide important insights by identifying modular architecture within the PDN and underlying factors that drive similarity between disease patterns among regional entities.