Singapore General Hospital is developing an AI solution to determine the need to prescribe antibiotics, reduce their use and identify the most suitable for each patient.
The Augmented Intelligence in Infectious Diseases (AI2D), developed with DXC technology, is an AI model currently covering pneumonia cases. It was built from deidentified clinical data, including radiographs, clinical symptoms, vital signs and trends in infection response, from approximately 8,000 SGH patients between 2019 and 2020. It included seven broad-spectrum intravenous antibiotics commonly prescribed against pneumonia.
FINDINGS
Last year, the research team led by the SGH Pharmacy division conducted a pilot validation study of the AI model, comparing it to 2,000 cases of pneumonia in 2023.
In an article, SGH and DXC noted that the AI2D helped reduce the number of cases requiring review by three times (624 out of 2,012). It also increased the ability to identify cases requiring intervention from that pile of review cases by almost 12%, compared to 4% with a completely manual review process. Furthermore, in one case, the analysis of the data, which took 20 minutes in a manual review, was done in “less than a second”.
In the study, the AI model demonstrated 90% accuracy in determining whether antibiotic administration is necessary in a given case of pneumonia; The study also found that almost 40% of antibiotics prescribed in these cases may not have been necessary.
WHY IT’S IMPORTANT
SGH said pneumonia accounts for 20% of all infections treated in their hospital, with antibiotic prescriptions the most common of all infections. On average, patients are hospitalized for two to nine days, costing the government up to SG$5,000 (over $3,500) for each government-subsidized patient admitted to a public hospital.
Half of acute care hospitals worldwide may be prescribing the wrong antibiotics to patients, resulting in antimicrobial resistance. For example, at SGH, the hospital unnecessarily prescribed 20% to 30% of broad-spectrum intravenous antibiotics, based on a 2018 antibiotic use guideline. In Singapore, up to 30% of hospital-acquired infections are believed to have become resistant to broad-spectrum antibiotics.
Hospitals are establishing antimicrobial stewardship programs to address this growing global problem by preventing the overuse of antibiotics and identifying where they – especially narrow-spectrum antibiotics – can best be recommended. Such programs can potentially help reduce hospital stays, minimize deaths and readmissions, and save patient and hospital costs.
When implementing antimicrobial stewardship programs, using automation and AI may be the best approach, as teams need real-time insights at the point of prescribing. AI would help identify assessment cases and prioritize cases that require intervention. Automation would allow hospitals to expand the coverage of their antibiotic use audit, providing deep insights into their use.
The research team is now preparing a comparative study involving 200 SGH patients to test the effectiveness of their AI model in reducing antibiotic use. Later, they will work to determine the most effective antibiotic for pneumonia, and then build the same model for urinary tract infections, another common hospital-acquired infection.
THE BIG TREND
Outside of Singapore, Taiwan’s hospital, China Medical University Hospital, has been recognized for its efforts to tackle antimicrobial resistance using AI. First implemented in 2021, its Intelligent Anti-Mbiotic System is a four-in-one AI-powered platform that identifies resistant strains, predicts and monitors sepsis and death, recommends drug doses and compares drug interactions. This specific application of AI in the EMR system was highlighted during the hospital’s rehabilitation for Phase 7 of the HIMSS Model for adoption of electronic health records in November last year.
ON THE RECORD
“Well-intentioned physicians continually balance the risks and benefits of using antibiotics. It is often difficult to definitively say that patients will benefit from them based solely on clinical judgment, patient-specific factors, or the severity of the condition. If the use If antibiotics are not immediately prescribed they really need, it can lead to serious complications. Yet the misuse of antibiotics contributes to antibiotic resistance, which poses a challenge for the future treatment of infections,” said Dr. Piotr. Chlebicki, AI2D project member and senior consultant at the SGH Department of Infectious Diseases.