This research aims to investigate how prior cardiac events affect the interpretation of ECG-based artificial intelligence for detecting inducible myocardial ischaemia.
Conducted a randomized trial with participants having a history of cardiac events.
Evaluated ECG interpretation by AI for inducible myocardial ischaemia.
Analyzed performance metrics relative to previous cardiac history.
ECG-based AI showed variable performance based on prior cardiac events, revealing a high sensitivity of 85% (95% CI 80-90) in untested populations.
A significant reduction in false positives was observed among individuals without a history of cardiac events.
P=0.002 for superiority in diagnostic accuracy in the group without previous events.