Heart failure & cardiomyopathies
Improving transthyretin cardiac amyloidosis detection from electrocardiograms through the Willem artificial intelligence platform
Authors' conclusion
This work expands Willem AI capabilities, enabling automated ATTR-CM detection from routine 12-lead ECGs in a real-world referral population. The Willem AI model may support informed clinical management and scalable triage to trigger confirmatory testing and accelerate diagnosis.
Study context
- Population
- Adults evaluated for ATTR-CMindividuals, 20,754 electrocardiograms (ECGs) after quality control · patients with ATTR-CM 585 · controls 2,316
- Participants
- 2,901 adults
- Exposure
- Willem artificial intelligence (AI) ATTR-CM detection from 12-lead ECGs
- patients with ATTR-CMn = 585
- controlsn = 2,316
Methodological detailstime-slicing convolutional neural network, training/validation and test data sets stratified by amyloidosis status, sex, and age groups
- Time-slicing convolutional neural network
- Training/validation and test data sets stratified by amyloidosis status, sex, and age groups
Main finding
As reported in the abstract
Willem AI diagnostic performance · test cohort
AUC
0.88
95% confidence interval, 0.85–0.91
Sensitivity
80.7%
Specificity
78.5%
Observational data: associations, not effects.
Further findings
Performance by ATTR type
- ATTRv AUC
- 0.91
- ATTRwt AUC
- 0.88
Sensitivity in early presentations
- Asymptomatic sensitivity
- 68.4%
- New York Heart Association class I sensitivity
- 73.9%
Performance in red-flag scenarios
- Carpal tunnel syndrome AUC
- 0.92
- Lumbar spinal stenosis AUC
- 0.92
- Polyneuropathy AUC
- 0.91
Original publication
Improving transthyretin cardiac amyloidosis detection from electrocardiograms through the Willem artificial intelligence platform. Heart Rhythm. Published 1 April 2026. doi:10.
Summarised from the published abstract; every number is checked against it