Health AI Central
Peer-Reviewed Publication
Mayo Clin Proc Digit Health2025;3(4):100296.December 1, 2025Journal Article

Identifying Bias at Scale in Clinical Notes Using Large Language Models.

Donald U Apakama1,2,3,4, Kim-Anh-Nhi Nguyen5, Daphnee Hyppolite6, Shelly Soffer7, Aya Mudrik8, Emilia Ling4, Akini Moses4, Ivanka Temnycky4, Allison Glasser6, Rebecca Anderson6, Prathamesh Parchure5, Evajoyce Woullard5, Masoud Edalati5, Lili Chan1,3, Clair Kronk2, Robert Freeman3, Arash Kia5, Prem Timsina1,5, Matthew A Levin1,5,9,10, Rohan Khera11, Patricia Kovatch10, Alexander W Charney3,10, Brendan G Carr4, Lynne D Richardson2,4,12, Carol R Horowitz2,12, Eyal Klang1,3, Girish N Nadkarni1,3
1The Windreich Department of Artificial Intelligence and Human Health, Icahn School of Medicine at Mount Sinai, New York, NY.
2Institute for Health Equity Research, Icahn School of Medicine at Mount Sinai, New York, NY.
3The Charles Bronfman Department of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY.
4Department of Emergency Medicine, Icahn School of Medicine at Mount Sinai, New York, NY.
5Institute for Healthcare Delivery Science, Icahn School of Medicine at Mount Sinai, New York, NY.
6Office of the Chief Medical Officer, Mount Sinai Health System, New York, NY.
7Institute of Hematology, Davidoff Cancer Center, Rabin Medical Center Petah-Tikva, Israel.
8Ben-Gurion University of the Negev, Be'er Sheva, Israel.
9Department of Anesthesiology, Perioperative and Pain Medicine, Icahn School of Medicine at Mount Sinai, New York, NY.
10Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY.
11Department of Medicine, Yale University School of Medicine New Haven, CT.
12Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, NY.
PubMed PMC DOI PDF

Abstract

OBJECTIVE: To evaluate whether generative pretrained transformer (GPT)-4 can detect and revise biased language in emergency department (ED) notes, against human-adjudicated gold-standard labels, and to identify modifiable factors associated with biased documentation. PATIENTS AND METHODS: We randomly sampled 50,000 ED medical and nursing notes from the Mount Sinai Health System (January 1, 2023,…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free AccountSign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.