Heart failure & cardiomyopathiesResearch article
INSIGHT Model
- explainable artificial intelligence screening tool
- Electronic health record data
- Natural language processing with Named Entity Recognition
- Clinical ModernBERT
- Local interpretable model-agnostic explanations
Authors' conclusion
In an unselected HF population with 1% prevalence of suspected ATTR-CM, INSIGHT-CM identified patients among whom 1 in 5 fit the pattern for suspected ATTR-CM, offering meaningful opportunities to improve timely diagnosis and treatment.
Study context
- Population
- Unselected heart failure (HF) population, electronic health record data (April 2020-March 2023)patients with HF
- Participants
- 100,290 participants
Main finding
As reported in the abstract
Identification of suspected ATTR-CM by INSIGHT-CM
| Group | sensitivity | specificity | positive predictive value | negative predictive value | area under the receiver-operating characteristic curve |
|---|---|---|---|---|---|
| INSIGHT-CM | 77.5% (95% CI: 75.9%-82.8%) | 94.7% (95% CI: 94.4%-95.0%) | 19.9% (95% CI: 17.6%-22.36%) | 99.6% (95% CI: 99.5%-99.7%) | 92.4% (95% CI: 90.5%-94.3%) |
- suspected ATTR-cardiomyopathy (ATTR-CM) n
- 1,108
- prevalence
- 1.2%
Original publication
INSIGHT Model. JACC Adv. Published 17 August 2026. doi:10.1016/j.jacadv.2026.103122
Summarised from the published abstract; every number is checked against it