Heart failure & cardiomyopathiesRandomised controlled trial
A large language model for complex cardiology care
- Randomised controlled trial
- retrospective patient data
- blinded subspecialist evaluation
Study context
- Population
- Complex patients suspected of genetic cardiomyopathy, real-world cases from a subspecialist cardiology practice; 9 general cardiologists
- Participants
- 9 general cardiologists
- Comparison
- LLM-assisted care vs usual care by general cardiologists
- Primary endpoint
- Blinded subspecialist preference (evaluation rubric of triage, diagnosis and management quality)
Primary result
As reported in the abstract
Overall subspecialist preference
P = 0.02
Further findings
Clinically significant errors
P = 0.033
Missing content
P = 0.0021
Cardiologist-reported benefit of Articulate Medical Intelligence Explorer (AMIE)
- Cardiologists who used AMIE reported AMIE helped their assessment
- 57.0%
- Cardiologists who used AMIE reported AMIE saved time
- 50.5%
Safety
Clinically significant errors
P = 0.033
Original publication
A large language model for complex cardiology care. Nat Med. Published 6 February 2026. doi:10.1038/s41591-025-04190-9
Summarised from the published abstract; every number is checked against it