King’s College London researchers utilized artificial intelligence (AI) to evaluate antimicrobial resistance in ICU patients and detect bloodstream infections that cause sepsis. The study aims to enhance outcomes for critically ill patients.
The development of defenses against treatment by microbes, known as antimicrobial resistance, is a major problem for global healthcare. According to estimates, it costs the NHS at least £180 million a year and results in 1.2 million deaths worldwide.
Bloodstream infections can develop antibiotic resistance and result in sepsis, a potentially fatal illness. Patients are at a significant risk of rapidly developing organ failure, shock, and even death after the infection has progressed to a sepsis stage.
Patients vary in their levels of antimicrobial resistance, influenced by prior antibiotic use, genetic factors, and even diet, which can impact their microbiome.
The team demonstrated how AI and machine learning can provide same-day triaging for patients in the intensive care unit (ICU), especially in settings with limited resources. This marks a significant advancement in this sector. The technology is far less expensive than manual testing.
The simplicity and scalability of this innovative machine learning approach indicate its potential for widespread implementation, offering a robust solution to address these critical healthcare issues on a larger scale and ultimately improve patient outcomes.
Yanzhong Wang, Professor, Statistics in Population Health, King’s College London
The time-consuming laboratory tests used for current ICU patient assessments necessitate cultivating germs in a lab, which can take up to five days. Given the fragility of ICU patients who may be afflicted with life-threatening conditions, this can significantly affect the results of care.
If clinicians had access to this information sooner, they could make better judgments about patient care, including antibiotic administration. Positive patient outcomes are strongly correlated with the appropriate use of antibiotics.
Our study provides further evidence on the benefits of AI in healthcare, this time relating to the crucial issues of antimicrobial resistance and bloodstream infections. It comes at an important time, as the NHS is investing in shared data resources, helping to make patient care more collaborative and efficient. Our use of machine learning provides a new way of tackling the important clinical issue of antimicrobial resistance. We hope that the AI will provide a useful tool for clinicians in making important decisions, particularly in relation to ICU.
Davide Ferrari, Study First Author, King’s College London
Dr. Lindsey Edwards, an expert in Microbiology at King’s College London added: “An important way to tackle the grave threat of antimicrobial resistance is to protect the antibiotics we already have, which goes hand in hand with the urgent need for fast diagnostics. Often patients with a drug-resistant infection will present to ICU in a critical condition and may not survive long enough for the current gold standards of diagnostics to determine what they are infected with.”
“So, clinicians are faced with a difficult situation where they must prescribe ‘in a blinded fashion’ a broad-spectrum antibiotic to save the patient,” Dr. Lindsey Edwards added.
However, this will also kill many of the beneficial microbes in the patient’s microbiome, without killing the harmful pathogen. It could even make the pathogen more resistant to the drug.
Dr. Lindsey Edwards, Microbiology Expert, King College London
Dr. Lindsey Edwards said, “The findings of this study are incredibly promising as using AI to speed up the diagnostics of infection to allow for prescription of the correct antibiotic could not only have a huge impact on the patient’s survival and their care outcomes; but could help to preserve the antibiotics we already have developed and prevent the development of further antibiotic resistance.”
This study, which used data from 1,142 patients at Guy's and St. Thomas' NHS Foundation Trust, has opened the door for more ongoing research employing datasets of over 20,000 people.
A more sophisticated approach to this study, specifically in a multi-hospital scenario using the well-liked Federated Machine Learning technology, could satisfy the legal requirements for a real implementation of this AI technique in the NHS front line.