Researchers at the University of Zurich used artificial intelligence to assist in identifying bacteria that are resistant to antibiotics. The group, headed by Adrian Egli, a Professor at the Institute of Medical Microbiology at UZH, is the first to look at the potential applications of GPT-4, a potent AI model created by OpenAI, for the analysis of antibiotic resistance. The Journal of Clinical Microbiology published this study.
Kirby-Bauer disk diffusion test of gut bacteria: Paper sheets soaked with antibiotics are placed on a petri dish. The antibiotic concentration decreases with increasing distance. The closer bacteria grow to the test sheets, the more resistant they are (red circles). If the gradients of two different antibiotics meet, their effectiveness can increase (yellow arrows). Image Credit: University of Zurich.
The Kirby-Bauer disk diffusion test, a widely used laboratory test that aids physicians in determining whether antibiotics are effective against a specific bacterial illness, was interpreted by the researchers using artificial intelligence.
The researchers developed the “EUCAST-GPT-expert” based on GPT-4, which interprets antimicrobial resistance pathways under stringent EUCAST (European Committee on Antimicrobial Susceptibility Testing) requirements. The method was tested on hundreds of bacterial samples, integrating the most recent data with expert criteria to assist detect drug resistance.
Human Experts Are More Accurate – But AI is Faster
Antibiotic resistance is a growing threat worldwide, and we urgently need faster, more reliable tools to detect it. Our research is the first step toward using AI in routine diagnostics to help doctors identify resistant bacteria more quickly.
Adrian Egli, Professor and Study Lead, Institute of Medical Microbiology, University of Zurich
Although it was not flawless, the AI system did a good job of identifying some forms of resistance. Although it was effective at identifying bacteria resistant to specific medicines, it occasionally identified germs as resistant when they were not, which could cause treatment delays.
The AI technology could assist in standardizing and expediting the diagnosis procedure, but human experts were more accurate in identifying resistance.
Useful Tool to Support Medical Staff
The study emphasizes AI's revolutionary potential in healthcare despite its limitations. AI may help lower the subjectivity and variability present in manual readings by providing a consistent method for interpreting complicated diagnostic tests, thus improving patient outcomes.
Adrian Egli highlights that before this AI technology may be utilized in hospitals, more research and development are required.
“Our study is an important first step, but we are far from replacing human expertise. Instead, we see AI as a complementary tool that can support microbiologists in their work,” Adrian Egli said.
Curbing the Global Development of Antibiotic Resistance
The study suggests that AI could help the international effort to combat the emergence of antibiotic resistance. As AI-based diagnostics continue to advance, they may assist labs throughout the globe in identifying drug-resistant illnesses more quickly and accurately, preserving the potency of currently available medications.
Journal Reference:
Giske, C. G., et al. (2024) GPT-4-based AI agents the new expert system for detection of antimicrobial resistance mechanisms? Journal of Clinical Microbiology. doi.org/10.1128/jcm.00689-24.