TAIPEI (Taiwan News) — A team from National Taiwan University Hospital has won top honors at the Infectious Diseases Society of America, taking the Abstract Award first prize for an AI-powered system that predicts bacterial antibiotic resistance.

The study, led by Lee Chien-chang (李建璋), director of the health ministry’s information department and a professor in the hospital’s Department of Emergency Medicine, was selected from more than 3,000 submissions worldwide. It also ranked first among four papers chosen for the US association’s Committee Choice Abstract Award.

Lee said the team analyzed more than 400,000 clinical bacterial samples and integrated large language models to develop the system. The model can identify bacterial species and provide precise recommendations on antibiotic selection, dosage, and formulation, helping doctors improve treatment outcomes, per CNA.

Lee said the system can also predict whether bacteria are resistant to antibiotics in under 30 seconds, covering about 80% of bacterial strains. This allows doctors to choose effective treatments quickly and helps reduce the risk of antibiotic resistance.

Sepsis is triggered by an infection that sparks an excessive immune response, causing widespread inflammation that can lead to organ failure, shock, or death. In Taiwan, there are about 110,000 cases annually, with up to 50% of severe cases resulting in death. 

Typically, tests to determine which antibiotics work best take three to four days, during which doctors must rely on experience to select initial treatment. According to the hospital, with rising bacterial resistance, about 30% of people with sepsis do not respond to the first treatment, delaying critical care.

Lee added the team will continue integrating clinical data with AI algorithms to showcase Taiwan's strength in combining healthcare and technology.