Bulletin of Computer and Data Sciences

Large language models are increasingly attractive for healthcare question answering, longitudinal note generation, discharge communication, and specialty-specic decision support, but their direct deployment in hospitals remains constrained by privacy regulation, compute limitations, heterogeneous local data, and the risk of destructive domain adaptation.This paper introduces MedQA-FoRA-MultiHospital, a new multihospital benchmark designed for privacypreserving adaptation of medical language models under realistic non-identically distributed client partitions.The benchmark contains 50,000 medical questionanswer pairs and 15,000 clinical report-generation instances distributed across ve simulated hospital clients with distinct specialty proles: cardiology, respiratory medicine, neurology, general medicine, and pediatrics.Building on the micromesomacro philosophy of ecient federated low-rank netuning, we formulate Adaptive FoRA, a heterogeneity-aware extension that preserves the frozen backbone, inserts structured low-rank operators across all major linear maps, and aggregates client updates through divergence-sensitive weighting.Rather than inventing unveried empirical scores, this manuscript contributes a complete benchmark specication, a mathematically explicit training framework, communication and parameter analyses, a privacy threat model, a full experimental protocol, and a reproducibility package structure suitable for subsequent leaderboard development.The paper is intentionally written as a stand-alone benchmark-andmethodology manuscript: it denes the dataset, the task suite, the optimization design, the ablation roadmap, and the evaluation criteria required for rigorous future implementation.We argue that the proposed benchmark lls an important gap between generic medical instruction datasets and privacy-preserving federated adaptation studies by unifying question answering, structured reasoning, and clinical report generation within a single non-IID multihospital setting.