New AI diagnostics can help prevent the spread of resistant bacteria

Resistant bacteria are everywhere in nature

Antibiotics are substances we use to cure or prevent infections caused by bacteria. Many antibiotics are naturally produced by microorganisms in different environments, and some bacteria have evolved the ability to survive even in the presence of antibiotics. This means that antibiotic-resistant bacteria are a natural part of the microflora in, for example, soil and our gut flora. However, problems arise when pathogenic bacteria become resistant, as they can then cause very difficult-to-treat infections. The increasing number of resistant bacterial infections is one of the biggest health problems in the world today, according to the World Health Organization, WHO.

Previous research shows that there are more variants of resistance genes in nature, rather than in for example hospitalised patients. A bacterium develops resistance through changes in its genome: either through mutations in genes it already carries or by acquiring resistance genes from other bacteria. The exchange of genes between bacteria makes the spread of resistance difficult to control. Anna Johnning, Erik Kristiansson and their colleagues are investigating how and under what conditions this spread occurs, to find ways to slow it down. By analysing DNA from different environments, researchers can discover new resistant genes and develop models showing how resistance genes spread between different types of bacteria.

“We have seen that some resistance genes spread rapidly, after their discovery. They spread to different types of bacteria, including those that cause diseases and lead to the need for medical care. We do not yet fully understand why some resistance genes spread rapidly and widely, and under what conditions. We need to understand the factors and processes that influence gene transfer to effectively prevent it,” says Anna Johnning. 

AI tools for diagnostics in healthcare

Multidrug-resistant bacteria are a major problem in healthcare worldwide. In severe infections, doctors often must start antibiotic treatment immediately, before they know for sure which treatment is effective. This can lead to unnecessary use of antibiotics or a treatment that does not work because the bacteria is resistant. They are therefore developing a new AI-based diagnostic tool for hospitals, which will be able to quickly identify resistant bacteria and thus help doctors choose an effective treatment.

With the help of large amount of data already available from diagnostic tests on millions of bacterial infections (so-called disc tests), an AI model has been trained to be able to determine which antibiotics are appropriate to prescribe. The goal is to create a simple, predictive and reliable AI diagnostic tool for routine use in healthcare. With better diagnostic tools, the use of antibiotics should be reduced, which in turn reduces the risk of new resistant bacteria.

“I hope that our work will lead to improved methods and tools that can be implemented in healthcare. We also continue to explore the evolutionary mechanisms of resistance to find new ways to prevent the spread of resistant genes,” says Johnning.

Interdisciplinary co-operation

Research on antibiotic resistance takes place in many scientific areas and is therefore often interdisciplinary. In Gothenburg, for example, there is the Centre for Antibiotic Resistance Research (CARe), which is a platform for collaboration between the University of Gothenburg, Chalmers University of Technology and the healthcare sector, with support from Sahlgrenska University Hospital and Region Västra Götaland. The centre brings together researchers from different disciplines with a common purpose – to combat antibiotic resistance.

“Belonging to a larger research network is positive. It contributes to increased collaboration, new impressions and broadened perspectives, which is positive for the research as a whole and the common goal of combating antibiotic resistance,” Johnning concludes.