31.10.2024

A pilot study at the University of Zurich is the first to demonstrate the potential of artificial intelligence (AI) in identifying antibiotic-resistant bacteria. Researchers used the GPT-4 model to support the diagnostic process in the laboratory and thus relieve the burden on medical professionals in the fight against increasing antibiotic resistance worldwide.

Under the leadership of Professor Adrian Egli from the Institute of Medical Microbiology, the University of Zurich is presenting a solution with this study that could speed up and optimize the diagnostic process for bacterial infections in the future.

Image: Example illustration of the Kirby-Bauer disk diffusion test; Copyright: UZH

Kirby-Bauer disk diffusion test of gut bacteria with paper sheets soaked with. The antibiotic concentration decreases with increasing distance. The closer bacteria grow to the test sheets, the more resistant they are (red circles).

Products and exhibitors related to this topic

Exhibitors and products related to this topic can be found in the database of MEDICA:

Focus on the Kirby-Bauer disk diffusion test

The study focused on the Kirby-Bauer disk diffusion test, a common method for determining the effectiveness of antibiotics in infections. The "EUCAST-GPT Expert" developed by the researchers is based on GPT-4 and follows the strict guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST). The system was tested on hundreds of bacterial samples and proved to be a useful tool for detecting resistance to essential antibiotics.

Speed versus precision: human expertise remains essential

"Antibiotic resistance is a growing threat worldwide, and we urgently need faster, more reliable tools to detect it," explains study leader Egli. Despite good results, the AI occasionally showed weaknesses: While it correctly identified specific types of resistance, there were occasional misclassifications where microbes were categorized as resistant when this was not the case. These discrepancies illustrate that human experts continue to play a central role in diagnostics. However, the AI-based system offers the potential to standardize the diagnostic process and ensure greater consistency.

Support for medical professionals

The study presented by UZH emphasizes the potential that AI has for clinical diagnostics in the healthcare sector. The standardized interpretation of complex tests could contribute to a reduction in variability and subjectivity in the analysis and thus increase the reliability of the diagnosis for doctors. Even if the AI system cannot currently replace human expertise, Professor Egli sees this as an important support for microbiologists: "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."

Contributing to the global containment of antibiotic resistance

Antibiotic resistance is a global problem that requires a comprehensive solution, according to the University of Zurich. In future, AI could help laboratories around the world to identify drug-resistant bacteria more precisely and quickly. Such systems could help to preserve the effectiveness of existing antibiotics for longer and thus strengthen efforts to combat the development of resistance.

MEDICA-tradefair.com; Source: University of Zurich

More about the UZH at: www.uzh.ch/en