Fischer KB, Valencia DN, Reddy A, Khouzam JP, Karabatak ZF, Reddivari A, Safar A, Reddy MN, Nazir RA, Schwartz BP. Effects of Artificial Intelligence Clinical Decision Support Tools on Complications Following Percutaneous Coronary Intervention.
JOURNAL OF THE SOCIETY FOR CARDIOVASCULAR ANGIOGRAPHY & INTERVENTIONS 2025;
4:102497. [PMID:
40230664 PMCID:
PMC11993894 DOI:
10.1016/j.jscai.2024.102497]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 09/26/2024] [Revised: 11/19/2024] [Accepted: 11/27/2024] [Indexed: 04/16/2025]
Abstract
Background
Artificial intelligence (AI) models have been created that incorporate unique patient characteristics to risk stratify patients undergoing cardiac catheterization with percutaneous coronary intervention (PCI). The most frequent complications following PCI are contrast-induced acute kidney injury (CI-AKI) and postprocedural bleeding, resulting in increased adverse outcomes, length of stay (LOS), and health care costs. Our study investigates the impact of AI clinical decision support tools on these events.
Methods
A retrospective review of patients undergoing PCI at our institution from April 2023 to March 2024 was performed. All patients had an ePRISM (Terumo Health Outcomes AI clinical decision support tool) generated risk assessment and maximum contrast volume recommendation reported during procedure time-out. Statistical analysis was performed to determine the incidence of post-PCI CI-AKI, bleeding events, and LOS.
Results
A total of 642 patients were analyzed. The incidence of CI-AKI significantly declined from a baseline of 10% to an average of 2.18% (P < .0001). Of the remaining CI-AKI, 92.9% occurred in hospitalized patients. The incidence of bleeding complications declined from a baseline incidence of 2.15 per month to an average of 1.54 per month. Our institution's average LOS declined from a baseline of 3.44 to 1.79 days.
Conclusions
AI clinical decision support tools can be effectively incorporated into clinical practice. ePRISM successfully risk-stratified patients undergoing PCI for CI-AKI and bleeding events and gave meaningful recommendations which resulted in a significant reduction in adverse events and LOS.
Collapse