A gradient-boosting model predicts each county's life expectancy from three socioeconomic
measures only: median household income, the share of residents without health insurance,
and primary care physicians per 100,000 residents. Every county's prediction comes from a
model trained on the other counties (5-fold cross-validation), so no county is scored
by a model that has seen it.
The map shows the gap between actual and predicted life expectancy. Counties more than two
years above or below their prediction are labeled as living longer or shorter than predicted.
Health behaviors such as smoking and physical inactivity are deliberately left out of the
model; each county profile instead compares them with counties in the same income decile,
which hints at why a county beats or misses its prediction. These comparisons are
correlations, not causes.
County Health Rankings & Roadmaps - 2025 Release (v3)
2024 health data for 3,159 U.S. counties covering health outcomes, health behaviors,
clinical care, and socioeconomic factors. A collaboration between the Robert Wood Johnson
Foundation and the University of Wisconsin Population Health Institute.