TaiwanHealthcare and life sciences2021 – ongoingChina Medical University Hospital
Intelligent Anti-Microbial System (iAMS) for sepsis and antibiotic decisions
A Taiwanese hospital built its own AI for sepsis and antibiotic choices and estimated 22 deaths avoided a year
Problem
Sepsis mortality in Taiwan reaches up to 29.2%, according to the hospital, and conventional culture tests take days to show which bacteria are present and which antibiotics will work. Doctors had to choose antibiotics before results arrived, and existing scoring tools predicted sepsis only moderately well.
Approach
China Medical University Hospital in Taichung built iAMS with an in-house team of about 40 people. The platform connects laboratory instruments, including mass spectrometry, to models that predict sepsis and 7-day mortality risk and antibiotic resistance, and pushes the results and prescribing alerts into the hospital information system. The sepsis module launched in April 2021 and the full platform in June 2021.
Results
Takeaway
Staff opened the system about 11,000 times a month in 2022. The mortality results came from an uncontrolled comparison with the same quarter of 2020.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Production and qualityIT and R&D
- Innovation type
- Process innovationProduct and service innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
The hospital said a 40-member team built iAMS, which launched in June 2021 and recorded 146,438 cumulative uses by the end of 2022. It reported lower sepsis mortality against the same quarter of 2020 and, for about 1,600 emergency bacteremia patients a year, estimated 22 deaths and about NT$7.33 million in insurance costs avoided.
The hospital's AI and IT leaders reported more than 65,000 uses of the sepsis module since April 2021 and model accuracy above 80% in their submission for the 2023 HIMSS Davies Award. They estimated 22 deaths and about NT$7.33 million in annual insurance costs avoided. They also reported that time spent on antibiotic decisions fell by up to 75% and monthly antibiotic costs by 2.66% to 19.66%.
Full details are in the original sources.
