Researchers at the University of Liverpool have demonstrated a novel approach to deal with the infections. The team has described the use of artificial intelligence (AI) to effectively treat infections and help address antimicrobial resistance (AMR).
Antimicrobial resistance occurs when a bacteria or a virus evolves to resist the once-effective treatment. Such resistance could pose a significant threat to public health and could render common infections untreatable.
To address AMR, the World Health Organization (WHO) has an Access, Watch, Reserve (AWaRe) framework to promote the sustainable use of antibiotics. Antimicrobial Susceptibility Testing (AST) is the key tool to assess the antimicrobial activity against a pathogen.
However, AST is a traditional diagnostic method that uses the Ivory Tower approach. This approach is like one-size-fits-all; if an antibiotic is effective against an infection, it is effective in all human bodies.
A new study, published in Nature Communications, proposes a personalized method that uses real-time data to help clinicians deal with infections more accurately. This approach uses AI to test prediction models for 12 antibiotics using real patient data.
“It could help clinicians and health systems achieve the core aim of antimicrobial treatment—effectively treating the individual while minimizing harm to that individual and the wider population,” says the study.
As this model offers variation in treatments from agent to agent and organism to organism, it reduces the chance of bacteria becoming resistant to antibiotics.
New microfluidic device sheds light on shape-dependent tumor growth
“This research is important and timely for World AMR Awareness Week because it shows how combining routine health data with lab tests can help keep antibiotics working,” said the lead author Dr Alex Howard.
“By using AI to predict when people with urine infections have antibiotic-resistant bugs, we show how lab tests can better direct their antibiotic treatment. This approach could improve the care of people with infections worldwide and help prevent the spread of antibiotic resistance.“
Given that the model approach is relatively new, it has several limitations. First, the AST results are limited to 12 antibiotic drugs. Second, certain data sets might be missing, which could lead to differences in treatment efficacy. Third, there could be other range of factors, other than AWaRe classification, that could influence the individual choice of antimicrobial therapy.
Nonetheless, the results have illustrated a significant leap in addressing antimicrobial resistance. The team of researchers has asserted that microbiology laboratories could use personalized AST to address the global AMR problem.
AI tool reads biopsy images
Journal Reference
- Howard, A., Hughes, D. M., Green, P. L., Velluva, A., Gerada, A., Maskell, S., Buchan, I. E., & Hope, W. (2024). Personalised antimicrobial susceptibility testing with clinical prediction modelling informs appropriate antibiotic use. Nature Communications, 15(1), 1-13. DOI: 10.1038/s41467-024-54192-3