Google screens 9,279 observations for biomarkers

Google Research’s Biomarker Discovery Framework analyzed 9,279 participant-observations across three wearable-data cohorts and identified 41 candidate digital biomarkers for mental health plus 25 for metabolic outcomes. In its research report on the supervised biomarker workflow, the team says adding the generated features to demographic variables improved prediction by ΔR² 0.040 for depression and 0.021 for insulin resistance; 15 domain experts retained 56.9% of its manuscript content, versus 18.8%–30.4% for comparison systems. This is research triage, not clinical validation: the useful pattern is to let agents generate and adversarially test hypotheses while humans own the release decision, much like the replication-first gate for AI scientists.