CareGraph Brings Auditable Hybrid AI to Longitudinal Health Intelligence

Summary

Researchers present CareGraph, an auditable hybrid AI framework that organizes clinical, self-reported, and wearable evidence into traceable health insights. It combines deterministic rules, graph retrieval, constrained language-model synthesis, validation, and safety gating without diagnosing or selecting treatments. Tests on synthetic cohorts showed strong trend analysis, missing-context detection, and safety performance. Compared with GPT-5.6, CareGraph was faster and produced shorter outputs, although GPT-5.6 cited more raw evidence.