Social Determinants Match Genetic Risk Scores for Four of Six Diseases in All of Us Analysis

The analysis integrated over 100 survey and community-level variables with genetic and EHR data from the NIH All of Us program across six conditions. For breast and prostate cancer, genetic scores retained greater weight, yet social factors still added measurable predictive value. The observational design cannot establish temporality or causation, yet the pattern directly challenges the prevailing emphasis on DNA-first risk models that often sideline living conditions.

Mainstream reporting on polygenic scores rarely quantifies how neighborhood resources, loneliness, or health behaviors perform in the same models. This omission sustains a genetic-determinism frame that understates modifiable drivers and equity gaps. Prior All of Us work on ancestry-specific scores already showed performance decay across populations; adding social variables narrows that gap without requiring new sequencing.

Next steps require longitudinal linkage of repeated social measures to incident events and formal tests of whether intervening on loneliness or access alters outcomes beyond genetic strata. Absent such trials, risk models will continue to overstate immutable biology while underfunding structural interventions.