Quality reviews for emergency department revisits frequently miss cases outside narrow 72-hour windows due to clinical review fatigue. In an exploratory evaluation of 99 diagnosis pairs from revisits within 1–14 days, baseline GPT-4 flagged 94% of cases for audit—far exceeding human clinicians. By channeling model outputs through a clinical knowledge graph algorithm, researchers achieved an 83–100% positive predictive value relative to physician consensus, demonstrating a viable way to expand review coverage without overwhelming hospital quality teams.
There are 4 persisted snapshots in the last 24 hours. Peak heat was 0 at 9/12, 20:00; latest heat is 0.