Real-world implementation of AI-assisted PERT-workflow and its association with care processes and budget impact: a retrospective cohort study.
Authors
Affiliations (2)
Affiliations (2)
- Department of Cardiology and Vascular Medicine, West German Heart and Vascular Centre, Essen, University of Duisburg-Essen, Essen, Germany.
- Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany.
Abstract
Pulmonary embolism (PE) remains a major cause of morbidity and mortality. In intermediate-high and high-risk PE, timely escalation beyond anticoagulation is critical, yet real-world pathways are often delayed. Pulmonary embolism response teams (PERTs) improve decision-making, but activation remains inconsistent. Whether artificial intelligence (AI)-enabled triage strengthens PERT and improves acute PE care has not been elucidated so far. We retrospectively screened 2500 patients presenting to our university hospital with a symptomatic central or segmental PE between 08/2022 and 07/2024. Clinical and economic outcomes were compared during a 1-year timeframe before and after implementation of an AI algorithm that analyses CT scans, flags suspected PE, and supports PERT triage. 424 patients met inclusion criteria. AI-supported triage increased PERT activation from 9.2% to 84.5% (<i>P</i> < 0.001). Time from CT completion to PE diagnosis decreased from 2.1 to 1.3 h (<i>P</i> < 0.01). Time from PE diagnosis to interventional therapy was reduced by 50% (<i>P</i> < 0.05). Hospital/ICU length-of-stay and 30-day mortality showed non-significant reductions. Model-based economic analysis showed lower 1-year per-patient costs and annual savings associated with AI-supported PERT. Implementation of an AI-supported PERT workflow was associated with significant improvements in PE care processes, including higher PERT activation rates and shorter diagnostic and treatment intervals. Numerical reductions in mortality, hospital/ICU length-of-stay did not reach statistical significance. The exploratory budget impact model suggested potential cost savings, mainly driven by trends in reduced resource utilization; however, these findings require prospective validation using adequately powered clinical endpoints.