Examining reliance patterns on AI advice in medical imaging: a mixed-methods, randomised, crossover experiment.
Showing doctors which specific areas of AI predictions matter most boosted their diagnostic accuracy on brain imaging by 84% compared to AI advice alone.
In a preregistered randomized crossover study of 125 clinicians reading head CTs, under-reliance on correct AI advice was more detrimental than over-reliance on incorrect advice; explainable AI (scan annotations) improved diagnostic performance vs no advice (OR=1.84, P<0.0001) and was perceived as more useful than basic AI prediction. This record was retained from the prior triage attempt for PubMed pipeline handoff.
What the study was
- Study design
- Randomized crossover experiment (preregistered)
- Category
- ai_ml_clinical_diagnostics
- Maturity
- Potentially Practice-Changing
- Journal
- Lancet Digit Health
Why it surfaced
High-quality preregistered Lancet Digital Health RCT identifying under-reliance as the primary AI-clinician risk in radiology and demonstrating explainable AI superiority; findings directly actionable for clinical AI deployment guidelines and human-AI teaming research.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.