Reviewed by Clare Marlin, MS, BSN, CIC, Shirley Ryan Ability Lab
Using risk prediction as an approach for catheter-associated urinary tract infection (CAUTI) prevention, the research team developed, and validated, an artificial intelligence (AI) model of prognosis and prediction for patients at high-risk for CAUTI development up to 48 hours after decatheterization through application of a random forest machine learning algorithm (utilizes multiple decision trees to determine an outcome).
Model development and internal validation used data from two hospitals, while a third hospital dataset served as external validation, all in Taiwan. Eligible adult patients (2-75 years old) with an order for urinary catheterization, including the time of catheterization to 48 hours after decatheterization, were analyzed. The research team defined a CAUTI among such patients as the documentation of a urinary tract infection or placement of a urinary bacterial culture order without a simultaneous diagnosis of pneumonia.
The research team foresees the application of the model when a patient is encountered by a clinician, which indicates implementation of the model within the electronic health record. The second application is when a patent with urinary catheterization is assessed by a clinician. Within this application, the model is an element of a multimodal CAUTI prevention strategy. In addition to the research team’s creation and sharing of a nomogram for clinician use, the team has also created a publicly available web application for the model’s implementation.
Reference:
Sufriyana, H., Chen, C., Chiu, H.-S., Sumazin, P., Yang, P.-Y., Kang, J.-H., & Su, E. C.-Y. (2025). Estimating individual risk of catheter-associated urinary tract infections using explainable artificial intelligence on clinical data. American Journal of Infection Control, 53, 368–374. https://doi.org/10.1016/j.ajic.2024.10.027