How Can a New genAI Tool Detect Bird Flu Virus Exposure?

New Delhi, Aug 28 (NationPress) - Researchers in the United States have introduced a groundbreaking generative artificial intelligence (AI) tool aimed at swiftly detecting exposure to the H5N1 virus, the causative agent of avian influenza, commonly known as bird flu. This innovation is expected to enhance national surveillance efforts as the virus continues its spread among various animal populations.

The findings, detailed in the journal Clinical Infectious Diseases, demonstrate that this AI tool can efficiently review notes in electronic medical records to pinpoint high-risk individuals potentially infected with H5N1 bird flu.

Dr. Katherine E. Goodman, Assistant Professor of Epidemiology and Public Health at the University of Maryland School of Medicine, emphasized, "This research illustrates how generative AI can bridge significant gaps in our public health framework by identifying high-risk patients who might otherwise remain undetected."

She further cautioned, "With H5N1 continuing to pose threats in animal populations, our greatest risk nationally is our lack of awareness. Without tracking symptomatic patients who might have been exposed to bird flu, we may miss infections that go unreported. It is crucial for healthcare systems to closely monitor potential human exposure and respond promptly to the information gathered."

The AI tool is noted for its efficiency, taking only 26 minutes to analyze data at a cost of merely 3 cents per patient. The research team highlighted its potential to establish a national network of clinical sentinel sites, significantly improving surveillance for emerging infectious diseases.

In their study, the research team utilized a generative AI large language model to analyze 13,494 emergency department visits from adult patients across urban, suburban, and rural settings in 2024.

These patients exhibited acute respiratory symptoms, such as cough, fever, and congestion, or conjunctivitis, which align with early signs of H5N1 infection. The primary objective was to evaluate the effectiveness of generative AI in identifying high-risk patients who may have been overlooked during initial assessments.

The model flagged 76 cases due to mentions of high-risk exposure to bird flu, including occupations such as butchers or individuals working on farms housing livestock like chickens and cows.

Upon review, 14 patients were confirmed to have had recent, relevant exposure to animals known to carry H5N1, including poultry, wild birds, and livestock. Although these patients were not explicitly tested for H5N1, the model successfully identified those rare cases among thousands treated for seasonal flu and other common respiratory issues.