AI Breakthrough in ICU: Faster Detection of Antimicrobial Resistance and Sepsis

Researchers at King’s College London have developed an AI-driven method for same-day antimicrobial resistance assessments in ICUs, promising improved patient outcomes and better management of life-threatening sepsis.

Scientists have harnessed the power of artificial intelligence to provide same-day assessments of antimicrobial resistance in intensive care units, revolutionizing the fight against life-threatening sepsis.

Antimicrobial resistance, the ability of microorganisms to develop defenses against treatments, remains a formidable challenge for global health care. This resistance is responsible for an estimated 1.2 million deaths worldwide annually. When bloodstream infections become resistant to antibiotics, they can escalate to sepsis — a dangerous condition likely to result in organ failure, shock and even death.

“An important way to tackle the grave threat of antimicrobial resistance is to protect the antibiotics we already have,” Lindsey Edwards, a senior lecturer in microbiology at King’s College London, said in a news release.

“Our study provides further evidence on the benefits of AI in health care, this time relating to the crucial issues of antimicrobial resistance and bloodstream infections,” added first author Davide Ferrari, a doctoral student at King’s College London.

Current diagnostic methods in ICUs require lengthy laboratory tests that can take up to five days, a delay that can be detrimental to critically ill patients. The new AI-driven approach, developed through collaboration between King’s College London researchers and clinicians at Guy’s and St Thomas’ NHS Foundation Trust, aims to provide same-day triaging for ICU patients. This rapid assessment is vital, especially in resource-limited environments.

Using data from 1,142 patients, the researchers demonstrated that AI and machine learning could quickly identify sepsis-causing infections, significantly impacting patient care and antibiotic stewardship.

“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,” Ferrari said.

“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,” added Edwards.

However, the broad-spectrum approach can kill beneficial microbes without eliminating the harmful pathogen, potentially exacerbating antibiotic resistance. The AI method aims to mitigate this issue by enabling more precise and timely antibiotic administration.

The study, recently published in PLOS Digital Health, paves the way for larger-scale research involving over 20,000 individuals and aims for real-world deployment of this AI technology within the NHS using Federated Machine Learning.

“The simplicity and scalability of this innovative machine learning approach indicate its potential for widespread implementation, offering a robust solution to address these critical health care issues on a larger scale and ultimately improve patient outcomes,” Yanzhong Wang, a professor of statistics in population health expert, said in the news release.

This research marks a significant advancement in the use of AI for health care, promising improved survival rates and optimized use of antibiotics, ultimately protecting both patient health and existing antibiotic treatments.