Credit score: Unsplash/CC0 Public Area
Researchers from the College of Maryland Faculty of Drugs have developed a brand new and extremely efficient utility of a man-made intelligence (AI) software to shortly scan notes in digital medical information and establish high-risk sufferers who might have been contaminated with H5N1 avian influenza or “bird flu,” in keeping with new findings printed within the journal Scientific Infectious Illnesses.
Utilizing a generative AI massive language mannequin (LLM), the analysis crew analyzed 13,494 visits throughout College of Maryland Medical System (UMMS) hospital emergency departments from grownup sufferers in city, suburban, and rural areas in 2024. These sufferers all had acute respiratory sickness (resembling cough, fever, congestion) or conjunctivitis—signs in line with early H5N1 infections. The aim was to evaluate how effectively generative AI might discover high-risk sufferers who might have been ignored on the time of preliminary therapy.
Scanning the entire emergency division notes, the mannequin flagged 76 as a result of they talked about a high-risk publicity for chicken flu, resembling working as a butcher or at a farm with livestock, like chickens or cows. Often, these exposures had been talked about by the way—for instance, documenting a affected person’s occupation as a butcher or farm employee—and never due to medical suspicion for chicken flu.
After a quick evaluation by analysis workers, 14 sufferers had been confirmed to have had latest, related publicity to animals recognized to hold H5N1, together with poultry, wild birds, and livestock. These sufferers weren’t examined particularly for H5N1, so their potential bird-flu infections weren’t confirmed, however the mannequin labored to seek out these “needle in a haystack” instances amongst 1000’s of sufferers handled for seasonal flu and different routine respiratory diseases.
“This study shows how generative AI can fill a critical gap in our public health infrastructure by detecting high-risk patients that would otherwise go unnoticed,” stated research corresponding writer Katherine E. Goodman, Ph.D., JD, Assistant Professor of Epidemiology & Public Well being at UMSOM and a college member of the College of Maryland Institute for Well being Computing (UM-IHC).
“With H5N1 continuing to circulate in U.S. animals, our biggest danger nationwide is that we don’t know what we don’t know. Because we are not tracking how many symptomatic patients have potential bird flu exposures, and how many of those patients are being tested, infections could be going undetected. It’s vital for health care systems to monitor for potential human exposure and to act quickly on that information.”
Since early 2024, H5N1 has contaminated greater than 1,075 dairy herds throughout 17 states, and over 175 million poultry and wild birds have examined optimistic throughout this outbreak interval. Recognized human instances stay uncommon, with 70 confirmed infections and only one fatality within the U.S. by mid-2025, in keeping with the Facilities for Illness Management and Prevention (CDC). There are, nevertheless, probably many extra infections which have gone undetected as a consequence of an absence of widespread testing. As well as, new strains might come up, enabling human-to-human airborne unfold, which might result in an uptick in instances and a possible epidemic.
“The AI review required only 26 minutes of human time and cost just three cents per patient note, demonstrating high scalability and efficiency,” stated research co-author Anthony Harris, MD, MPH, Professor and Performing Chair of Epidemiology & Public Well being at UMSOM. “This method has the potential to create a national network of clinical sentinel sites for emerging infectious disease surveillance to help us better monitor newly emerging epidemics.”
The LLM (GPT-4 Turbo) demonstrated sturdy efficiency in figuring out mentions of animal publicity, with a 90% optimistic predictive worth and a 98% adverse predictive worth when it was evaluated on a pattern of 10,000 historic emergency division visits from 2022–2023, earlier than chicken flu was circulating in U.S. livestock. Nevertheless, the mannequin was conservative when figuring out exposures particularly related to avian influenza—typically flagging sufferers with low-risk animal contact, resembling publicity to canine—underscoring the necessity for human evaluation of any flagged instances.
As the danger of infections transmitted by animals grows, researchers counsel that enormous language fashions is also used prospectively to alert well being care suppliers in actual time. This might immediate them to be extra vigilant about asking about potential publicity to contaminated animals, focused testing, and controlling infections by isolating sufferers. The CDC at the moment depends on mandated lab reporting to trace avian influenza however lacks techniques to evaluate whether or not clinicians are asking about or documenting related exposures in symptomatic sufferers.
The researchers hope to subsequent check the big language mannequin for potential surveillance and deployment inside the digital well being file, for quicker real-time identification of high-risk sufferers. As respiratory virus season resumes within the fall, having a quick and correct solution to establish these sufferers needing particular testing for chicken flu, or precautionary isolation whereas receiving therapy, can be particularly essential.
“We are at the forefront of a disruptive but incredibly promising revolution around big data and artificial intelligence,” stated UMSOM Dean Mark T. Gladwin, MD, who can be the Vice President for Medical Affairs, College of Maryland, Baltimore (UMB), and the John Z. and Akiko Ok. Bowers Distinguished Professor.
“The engineers and physician researchers working at the Institute for Health Computing have secure access to medical records from the two million patients that we serve throughout Maryland, and as this study demonstrates, can use AI and big data to identify early signals of emerging infectious diseases like bird flu to enable us to take action sooner to test for these diseases and keep them from spreading.”
Different UMSOM college co-authors on the paper embody Laurence S. Magder, Ph.D., Professor of Epidemiology & Public Well being at UMSOM, Jonathan D. Baghdadi, Ph.D., MD, Affiliate Professor of Epidemiology & Public Well being at UMSOM, who can be on the college on the UM-IHC, and Daniel J. Morgan, MD, MS, Professor of Epidemiology & Public Well being at UMSOM.
Extra info:
Katherine E Goodman et al, Generative Synthetic Intelligence–primarily based Surveillance for Avian Influenza Throughout a Statewide Healthcare System, Scientific Infectious Illnesses (2025). DOI: 10.1093/cid/ciaf369
Offered by
College of Maryland Faculty of Drugs
Quotation:
Generative AI uncovers undetected chicken flu publicity dangers in Maryland emergency departments (2025, August 25)
retrieved 25 August 2025
from https://medicalxpress.com/information/2025-08-generative-ai-uncovers-undetected-bird.html
This doc is topic to copyright. Aside from any truthful dealing for the aim of personal research or analysis, no
half could also be reproduced with out the written permission. The content material is supplied for info functions solely.
