PROTEUS: Can AI improve decision-making around stress echocardiography?

AI augmentation of stress echocardiography did not meet the pre-specified non-inferiority endpoint for appropriate referral to angiography compared with standard decision-making in the PROTEUS trial. However, this effect appeared to be driven by a lower-than-expected angiography referral rate in the study and the investigators observed some positive signs for the AI intervention.

The multicentre, randomised-controlled PROTEUS trial (NCT05028179) tested whether AI-augmented interpretation of stress echocardiography improved the accuracy of clinician selection to refer patients for angiography. “The AI device we used captured left ventricular motion between rest and stress as an additional assessment,” explained Dr Ross Upton (Ultromics, UK) [1].

Patients who were referred for stress echocardiography for the investigation of ischaemic heart disease were randomised to standard-of-care or AI-augmented stress echocardiography. The primary endpoint was evidence of severe coronary artery disease within 6 months after stress echocardiography in patients who were referred for angiography. “Non-inferiority was accepted when the lower bound of the 95% confidence interval of the AUROC difference between intervention and control did not surpass -0.05,” explained Dr Upton.

The primary endpoint was not met, with an AUROC of 0.55 for the control group and 0.63 for the intervention group (difference 0.09; 95% CI -0.22 to 0.39). “The referral rate was only 8% in the current trial, whereas the data was powered on data from the EVAREST trial [NCT03674255], in which 15% of the population were referred for angiography,” said Dr Upton. According to the authors, this low angiography referral rate was most likely responsible for the trial not meeting its primary endpoint. Finally, a subgroup analysis suggested that sites that performed fewer stress echocardiographies may benefit from AI-augmented stress echocardiography (AUROC 0.33 vs 0.58; difference 0.25; 95% CI -0.02 to 0.62).

“Despite the low number of angiography referrals, the results indicate that AI augmentation may have utility in low volume stress echocardiography centres,” concluded Dr Upton.

1. 
   
   1. Upton R, et al. PROTEUS: a prospective randomised controlled trial evaluating the use of artificial intelligence in stress echocardiography. HOTLINE 12, ESC Congress 2024, 30 Aug–02 Sept, London, UK.

Copyright ©2024 Medicom Medical Publishers



Posted on 7 November 20247 November 2024