AI-Assisted Heart Scans Show Promise for Less Experienced Clinicians in PROTEUS Trial
Key Insights
The PROTEUS trial evaluated AI's role in stress echocardiogram interpretation for patients with suspected coronary artery disease (search), showing no significant difference in the primary endpoint.
AI assistance may improve decision-making among less experienced clinicians and in subgroups with challenging-to-interpret images.
The study suggests AI could potentially standardize accuracy levels across clinicians with varying experience in stress echocardiogram interpretation.
A recent randomized trial, known as PROTEUS, conducted across 20 hospitals in the UK, investigated the efficacy of artificial intelligence (AI) in aiding clinical decision-making during stress echocardiograms for patients with suspected heart disease. While the primary endpoint showed no significant differences compared to standard practice, the study revealed potential benefits for less experienced clinicians and specific patient subgroups with complex cases.
The findings, presented at the ESC Congress 2024, suggest that AI integration could enhance diagnostic accuracy and treatment decisions, particularly in scenarios where clinical interpretation is more challenging. The trial enrolled 2,341 patients to evaluate whether AI-augmented decision-making could improve the selection of patients for invasive coronary angiograms and reduce acute coronary events within six months.
PROTEUS Trial Details
The PROTEUS trial enrolled patients aged 18 and older referred to stress echocardiography (SE) clinics for suspected coronary artery disease (search) (CAD (search)) between November 2021 and June 2023. Participants were randomized 1:1 to either standard clinical decision-making or AI-augmented decision-making using EchoGo Pro (search) software, which provides automated interpretation of SE images. The primary outcome assessed the appropriateness of patient selection for invasive coronary angiograms, with secondary endpoints including acute coronary events within 6 months.
Key Findings and Clinical Implications
The analysis revealed that AI-assisted decision-making did not demonstrate non-inferiority compared to clinical decision-making alone in correctly selecting patients for coronary angiography. However, further analysis indicated that AI could benefit less experienced clinicians and specific subgroups with images known to be more difficult to interpret.
Dr. Ross Upton from the University of Oxford, the lead author, noted, "The results of this trial suggest that AI has the potential to bring all operators, regardless of experience, up to the same level of accuracy." This is particularly relevant given the reported variability in clinician performance in interpreting stress echocardiograms based on experience level.
The Need for Improved CAD Diagnosis
Coronary artery disease (search) (CAD (search)) is a major global health concern, responsible for over 9 million deaths worldwide and 63,000 deaths annually in the UK. Stress echocardiography (SE) is a commonly used diagnostic imaging test for CAD, but its accuracy varies widely (60% to 94%) depending on clinician expertise and image quality. The PROTEUS trial sought to determine if AI could improve the consistency and accuracy of SE interpretation, potentially leading to better patient outcomes.