[2412.02868] PrecLLM: A Privacy-Preserving Framework for Efficient Clinical Annotation Extraction from Unstructured EHRs using Small-Scale LLMs

Computer Science > Artificial Intelligence arXiv:2412.02868 (cs) [Submitted on 3 Dec 2024 (v1), last revised 22 Mar 2026 (this version, v3)] Title:PrecLLM: A Privacy-Preserving Framework for Efficient Clinical Annotation Extraction from Unstructured EHRs using Small-Scale LLMs Authors:Yixiang Qu, Yifan Dai, Shilin Yu, Pradham Tanikella, Malvika Pillai, Walter Chen, Jialiu Xie, Yishan Ren, Duan Wang, Yikai Wang, Sid Sheth, Guanting Chen, Yufeng Liu, Travis Schrank, Trevor Hackman, Didong Li, Di Wu View a PDF of the paper titled PrecLLM: A Privacy-Preserving Framework for Efficient Clinical Annotation Extraction from Unstructured EHRs using Small-Scale LLMs, by Yixiang Qu and 16 other authors View PDF HTML (experimental) Abstract:Large Language Models (LLMs) have demonstrated remarkable proficiency in automated text annotation within natural language processing. However, their deployment in clinical settings is severely constrained by strict privacy regulations and the prohibitive computational cost of processing voluminous unstructured Electronic Health Records (EHRs). In this study, we developed a resource-efficient preprocessing technique that can be adopted in existing LLM procedures. This approach is particularly useful for smaller LLMs, which are often more accuracy-challenged, and forms a compact LLM framework optimized for local deployment in computational environments with stringent privacy requirements and restricted access to high-performance GPUs (PrecLLM). The p...