Abstract Background The incidence of pulmonary embolism (PE) continues to rise globally, significantly impacting morbidity and mortality. Conservative therapies like oral anticoagulation might fall short in patients with intermediate high and high risk, requiring additional therapeutical options. Especially in patients with complex morbidities and high fatality rates, timinig of advanced therapeutical strategies like interventional catheter based thrombectomy or local lysis are of particular importance. Pulmonary Embolism Response Teams (PERT) have become central in managing acute PE. There is a pressing need for innovative tools to support PERT teams. The impact of integration of innovative artificial intelligence (AI) algorithms in PE diagnosis and optimizing PERT management has not been elucidated so far. Purpose Aim of this retrospective trial is to analyse to which extent an AI enhanced PERT approach can increase PERT activation rates, reduce time to interventional treatment decisions and overall patient outcomes. Methods We screened 2500 patients that presented to our clinic with a positive pulmonary embolism finding on a CT-scan between 08/2022 and 07/2024. Patients were stratified accoring to PE location and PESI score. Time from admission to clinic, ordering of CT-scan and PERT activation as well as treatment algorithms were analyzed. We compared those data for one year before implementation of an AI algorithm and for one year after implementation of our AI-PERT approach. Statistical analysis was performed using the student’s t-test. Results Of the 2500 screened patients with a positive PE finding, 181 patients had a central and 243 had a segmental pulmonary emoblism with 45% of patients being female. The median pulmonary embolism severity score (PESI) was 115.5 with a 30-day mortality rate of 18%. PERT-activation rates went up almost 10-fold from 9.2% up to 84.5% after implementation of our AI-PERT. Our AI-PERT concept increased interventional treatment strategies of intermediate-high PE cases from 6.5% (13 out of 200 cases) to 10.2% (23 out of 226 cases) and decreased elapsed time from diagnosis to interventional therapy from 2158 minutes to 1133 minutes (p 0.05). Conclusions Our data show that implementation of an AI-enhanced PERT network approach has significantly improved guideline recommended activation rates of PERT and expedited treatment decisions in acute PE management. By streamlining workflows, the AI-PERT model has increased the frequency of interventional therapies in patients requiring advanced therapeutical options and substantially reduced the time to treatment initiation. These findings highlight the potential of AI integration in acute PE care, suggesting that AI-supported PERT protocols may play a critical role in improving patient outcomes and reducing hospital stays. Future studies should investigate long-term impacts and broader applications of AI in PE management to further validate and optimize these promising results.
Messiha et al. (Sat,) studied this question.