Comprehensive Summary
This study, presented by Liu et al., examined the application artificial intelligence machine learning model Random Forest modeling compared to statistical logistic regressions on diabetes mellitus (DM) risk factors in the US, “diabetes belt.” The data analyzed was taken from a CDC survey, the 2019 Behavioral Risk Factor Surveillance System (BRFSS) where participants (n = 398,243, age = 18+) were separated based on residence in or outside of the diabetes belt. The predominant factors explaining the prevalence of diabetes in the belt were low socioeconomic status, low physical activity, and high blood pressure while also confirming the fact that adults living in the belt had a relatively higher rate of DM. Liu et al. acknowledged the impact of social determinants on health status and the disadvantage other comorbidities can have on diabetes risk.
Outcomes and Implications
By acknowledging and studying the impact of health disparities on the general populations, public health initiatives can better target the needs of a community and clinicians can have better awareness of the population a patient is coming from. The health behaviors and traditions of a community can greatly influence well-being and targeted efforts to improve the SES and quality of life of those living in a “diabetes belt” could lower their risk. Social conditions could be another large determinant of the diabetes burden on a population and providing resources derived from data-based need may best address these disparities.