CareGraph：可审计的混合AI框架用于个性化纵向健康智能
原标题：CareGraph: An Auditable Hybrid AI Framework for Evidence-Grounded Personalized Longitudinal Health Intelligence
AI 摘要
arXiv 论文提出 CareGraph，一个可审计的混合 AI 框架，用于基于证据的个性化纵向健康智能。该框架将实验室、电子病历、可穿戴设备等多源健康数据转化为趋势、缺失上下文和可追溯的解释，并通过确定性规则与 LLM 结合，在合成数据上验证了高准确率和安全性。研究表明 CareGraph 比单一 GPT-5.6 基线更快、更简洁，且更符合纵向目标，为患者和临床医生提供可检查的健康情报支持。
正文节选
CareGraph: An Auditable Hybrid AI Framework for Evidence-Grounded Personalized Longitudinal Health Intelligence Abstract Health information already exists across laboratory portals, electronic health records, medication lists, symptoms, lifestyle reports, and wearable devices, yet these sources rarely form a coherent and traceable explanation for an individual. CareGraph is proposed as an auditable hybrid AI framework and extensible backbone for evidence-grounded personalized longitudinal health