Phenotyping population-level chronic condition prevalence: The importance of forcing factors from the ecological...

The primary health challenges currently facing the United States (U.S.) and many other countries around the world are largely due to patients with chronic conditions, which act either independently or synergistically. The current study assesses the ability of the Ecological Framework of Population Health to predict U.S. county-level prevalences of eight common chronic conditions.

Analytic analysis of population-level surveillance data METHODS: This study utilizes several U.S. county-level datasets representing over 30 predictive variables of the ecological framework, a model that includes measures of culture, politics, policy, socioeconomics, lifestyle behaviors, and both chronic condition risk factors and diagnoses. A non-linear artificial intelligence statistical approach was used to assess the ability of these variables (i.e., features) to predict the prevalence of eight leading chronic conditions at the U.S. county-level.

Artificial intelligence models demonstrated good to excellent performances in the independent test set (0.73 < R² < 0.96) in predicting U.S. county-level prevalence of chronic conditions. Findings indicate that upstream domains (culture, politics, policy and environment) explain substantial variance in the prevalence of chronic conditions before downstream domains (behavior and risk) are introduced.

Despite a significant amount of attention given to the health challenges associated with chronic conditions, little progress has been made in reversing trends. The findings presented here propose a new approach to this complex issue that focuses on the forcing factors that lie upstream from health behaviors to improve downstream health outcomes.