Institution-Specific LLM Prompting Recovers PHI That De-identification Systems and Their Gold Standards Both Miss
Source
arxiv.org
Author
Daniel Palacios, Matthew Brady Neeley, Angel Adetomike Otto, Shalini Dhamodharan, John P. Woodhouse, Chi-fan Lin, Mark Zobeck, Zhandong Liu, Hyun-Hwan Jeong
Date
Why it matters
Purpose-built de-identification tools and the gold standards used to grade them both miss locally-defined PHI — hospital abbreviations, building names, internal codes. Adding institution-specific context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → to the prompt recovers those spans and gives you a precision/recall dial.
Terms in this piece · Glossary
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
articleAuditable Emergency Triage for Maternal and Newborn Care in IndiaShobhit Jagga, Aman Dalmia, Niharika Priyadarshini, Neelima Devadas, Amrita K Prasen, Nikhil Nalin, Santhosh SJ, Sreeram Nurani Ramasubramanian, Muhammed Afeer K, Anubhav Arora
articleDo large language models scrutinise what they review? A multimodal audit of scoring calibration, error detection, and author-identity effectsEmad Alharbi
articleBeyond Prompt Engineering: A Systematic Analysis of Prompt Lexical Sensitivity and Its Impacts on QualityQipeng Xie, Zi Liang, Jiafei Wu, Yufei Chen, Weizheng Wang, Wenao Ma, Zhong Ming, Haiqin Yang, Kaishun Wu