FedPref: Federated Preference Learning for Structured Radiology Report Extraction
FedPref lets institutions with unequal, unpooled data benefit from collaboration without ever sharing reports or annotations, and frozen public language models propose alternative JSON extractions, local annotations rank them, and sites collaboratively train compact Qwen3-8B adapters while sharing only model updates.
Flint Xiaofeng Fan, Cheston Tan, Y. Ong et al.
· 0 citations