EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Privacy-Preserving Federated Medical NLP Framework Using Large Language Models for Healthcare Analytics

Shweta Dharmadhikari, Ameya Kadam, Bhagyashri Kulkarni, Rupali Bhimrao Sirsat

Computer and Decision Making: An International Journal · 05.11.2026 · 3 (6S) · 2637-2646

DOI: 10.1922/comdem.v3i6s.19013 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 05.11.2026. Verfasst von Shweta Dharmadhikari und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Medical Natural Language Processing (Medical NLP) is increasingly being used to extract clinically relevant information from Electronic Health Records (EHRs), clinical notes, discharge summaries, radiology reports, and other healthcare documents. However, the sensitive and distributed nature of clinical data creates significant challenges for centralized model training, particularly with the growing adoption of transformer-based models and Large Language Models (LLMs). This paper proposes a privacy-preserving Federated Medical NLP framework that enables collaborative training of Medical NLP models across healthcare institutions without requiring the exchange of raw patient data. The proposed framework integrates local clinical data preprocessing, transformer-based Medical NLP models and LLMs, federated optimization, Differential Privacy, Secure Aggregation, and encrypted model-update exchange to support privacy-aware collaborative learning. The framework is designed to address key challenges associated with Federated Medical NLP, including non-IID clinical data distributions, communication overhead, scalability, privacy risks, model security, and interpretability. The proposed architecture supports multiple clinical NLP applications, including Named Entity Recognition, medical text classification, ICD coding, clinical question answering, and clinical report summarization. A systematic evaluation methodology is defined to assess the framework from both Medical NLP and Federated Learning perspectives using predictive performance, communication efficiency, convergence, training time, scalability, privacy budget, and explainability measures. The proposed framework provides a structured foundation for secure and scalable collaborative Medical NLP and establishes a basis for future experimental investigation of privacy-preserving Medical LLMs in distributed healthcare environments.

Zitieren

APA

Dharmadhikari, S., Kadam, A., Kulkarni, B., & Sirsat, R. B. (2026). A Privacy-Preserving Federated Medical NLP Framework Using Large Language Models for Healthcare Analytics. Computer and Decision Making: An International Journal, 3(6S), 2637-2646. https://doi.org/10.1922/comdem.v3i6s.19013

Vancouver

Dharmadhikari S, Kadam A, Kulkarni B, Sirsat RB. A Privacy-Preserving Federated Medical NLP Framework Using Large Language Models for Healthcare Analytics. Computer and Decision Making: An International Journal. 2026;3(6S):2637-2646. doi:10.1922/comdem.v3i6s.19013

BibTeX
@article{Dharmadhikari2026,
  author = {Dharmadhikari, Shweta and Kadam, Ameya and Kulkarni, Bhagyashri and Sirsat, Rupali Bhimrao},
  title = {A Privacy-Preserving Federated Medical NLP Framework Using Large Language Models for Healthcare Analytics},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {2637--2646},
  doi = {10.1922/comdem.v3i6s.19013}
}
Kostenlos starten →