Source link : https://health365.info/finely-tuned-broad-language-fashions-can-extract-pd-l1-biomarker-main-points-from-digital-well-being-data/
Credit score: AI-generated symbol
A brand new find out about revealed in AI in Precision Oncology examines the facility of enormous language fashions (LLMs) to swiftly extract PD-L1 biomarker main points from an digital well being file (EHR).
PD-L1 biomarker trying out guides most cancers remedy choices. Those effects are tricky to get right of entry to as a result of laboratory stories are unstructured and require scientific experience to interpret.
Aaron Cohen, MD, from Flatiron Well being and NYU Langone Faculty of Drugs, and co-authors, carried out open-source LLMs to extract seven biomarker main points in terms of PD-L1 trying out from the Flatiron Well being US national, EHR-derived database.
“LLMs, fine-tuned with high-quality labeled data, accurately extracted complex PD-L1 test details from EHRs despite considerable variability in cancer type, documentation and time,” reported the investigators.
“This study is a prime example of AI’s power to find needles in haystacks within massive medical records. By extracting critical biomarker data like PD-L1, AI is helping us provide better care, save valuable time, and improve patient outcomes. I commend these authors for their important work and outstanding manuscript. This is exactly the kind of research we aim to publish in AI in Precision Oncology,” says Douglas Flowers, MD, the magazine’s editor-in-chief.
Additional information:
Aaron B. Cohen et al, Huge Language Style Extraction of PD-L1 Biomarker Trying out Main points From Digital Well being Information, AI in Precision Oncology (2025). DOI: 10.1089/aipo.2024.0043
Supplied via
Mary Ann Liebert, Inc
Quotation:
Finely tuned broad language fashions can extract PD-L1 biomarker main points from digital well being data (2025, March 5)
retrieved 5 March 2025
from https://medicalxpress.com/information/2025-03-finely-tuned-large-language-pd.html
This record is matter to copyright. Except for any honest dealing for the aim of personal find out about or analysis, no
phase is also reproduced with out the written permission. The content material is equipped for info functions simplest.
Author : admin
Publish date : 2025-03-05 20:41:26
Copyright for syndicated content belongs to the linked Source.