用联邦学习生成医疗时间序列数据，保护隐私还能提升模型训练效果。
Federated Timeline Synthesis: Scalable and Private Methodology For Model Training and Deployment
- 将病历转化为语言无关的时序令牌，分机构训练并仅上传模型权重。
- 合成大量临床轨迹，使生成模型性能接近真实数据训练结果。
- 适合医疗领域隐私敏感场景，支持预测、预警和虚拟试验设计。
我们提出联邦时间线合成（FTS）框架，用于在分布式电子健康记录（EHR）时序数据上训练生成式基础模型。核心是将患者病史表示为经令牌化的患者健康时间线（PHTs），即编码时间、分类与连续临床信息的语言无关序列。各机构在其本地PHTs上训练自回归Transformer，并仅向中央服务器传输模型权重。服务器利用这些生成器合成大规模轨迹，训练全局生成器（GG），通过蒙特卡洛模拟实现零样本未来PHTs推理。我们在MIMIC-IV数据集上评估了五项临床有意义的预测任务，结果显示由GG生成的合成数据训练的模型表现与真实数据训练模型相当。FTS提供强隐私保障、跨机构可扩展性，并可扩展至多样化预测与仿真任务，尤其适用于医疗领域的反事实推断、早期预警检测及合成试验设计。
原文摘要 · Abstract (English)
We present Federated Timeline Synthesis (FTS), a novel framework for training generative foundation models across distributed timeseries data applied to electronic health records (EHR). At its core, FTS represents patient history as tokenized Patient Health Timelines (PHTs), language-agnostic sequences encoding temporal, categorical, and continuous clinical information. Each institution trains an autoregressive transformer on its local PHTs and transmits only model weights to a central server. The server uses the generators to synthesize a large corpus of trajectories and train a Global Generator (GG), enabling zero-shot inference via Monte Carlo simulation of future PHTs. We evaluate FTS on five clinically meaningful prediction tasks using MIMIC-IV data, showing that models trained on synthetic data generated by GG perform comparably to those trained on real data. FTS offers strong privacy guarantees, scalability across institutions, and extensibility to diverse prediction and simulation tasks especially in healthcare, including counterfactual inference, early warning detection, and synthetic trial design.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。