Alignment Between Cardiologists and AI-Driven Diagnostic Systems: Mixed Methods Study
Cardiologists using AI for heart imaging found the technology performed worse in complex patients who need accurate diagnosis most, highlighting where AI systems need improvement.
In the PROTEUS RCT, AI-cardiologist diagnostic alignment for stress echocardiography was modest (60%), with the AI system performing worse in comorbid patients most in need of accurate diagnosis. Cardiologists use AI as a second opinion rather than a decision-maker, highlighting the need for AI systems that better integrate clinical context and avoid perpetuating diagnostic bias.
What the study was
- Study design
- Mixed methods (RCT data analysis + qualitative survey)
- Population
- Patients with suspected CAD undergoing stress echocardiography (n=854, PROTEUS RCT); UK cardiologists (n=61 survey)
- Sample size
- 854
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Journal of Medical Internet Research
Why it surfaced
Multicenter RCT + qualitative data from practicing cardiologists; identifies real-world AI deployment gaps for one of the most common cardiac diagnostics; Harvard-affiliated authorship.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.