Reviewed by Clare Marlin, MS, BSN, CIC, Shirley Ryan Ability Lab
Central line-associated bloodstream infections (CLABSIs) may prolong a patient’s hospital stay, lead to further illness, and call for treatment with complex medical care and resources. In this retrospective cohort study of patients in adult intensive care units (ICUs), Lahariya and colleagues developed an artificial intelligence (AI) predictive model, distributed as a clinician-utilized AI-based calculator on the internet (“CLABpredICU”), of CLABSI risk development. The calculator was designed for use two days after central line insertion with seven biochemical marker values entered. Once the markers are entered, a predicted percentage of CLABSI development to scaffold ongoing infection prevention interventions is generated.
The seven biochemical markers for entry were identified by the researchers as standard lab collections. Additionally, the markers were determined following analysis of the labs of CLABSI positive vs. CLABSI negative patients with significant differences in coagulation times, kidney function, and electrolyte (i.e. sodium) abnormalities between the infectious and non-infectious patients. To best identify which machine learning analysis was most predictive, the researchers assessed four analyses and identified the support vector machine (SVM) was the best analysis for the model.
The researchers acknowledge the potential for future growth and application of this model with external validation. Such validation includes expansion of the model in clinical application, potential inclusion of additional inputs to assess infection risk, and increased implementation among patients.
Reference:
Lahariya, R., Anand, G., Sarfraz, A., Tiewsoh, J. B. A., and Kumar, A. (2025). CLABpredICU—AI-driven risk prediction for CLABSI in intensive care units based on clinical and biochemical parameters. American Journal of Infection Control, 53, 875–880. https://doi.org/10.1016/j.ajic.2025.05.016