Heart failure & cardiomyopathies
Automated Detection of Reduced Ejection Fraction Using an ECG-Enabled Digital Stethoscope: A Large Cohort Validation
An erratum to this article was published (JACC Adv, Aug 2025 ↗)
Authors' conclusion
The CNN model, utilizing a digital stethoscope, offers a noninvasive and scalable method for early detection of individuals with EF ≤40%. This technology has the potential to facilitate early diagnosis and treatment of heart failure, thereby improving patient outcomes.
Study context
- Population
- Adults undergoing echocardiography from 4 U.S. health care networksadults undergoing echocardiography
- Participants
- 2,960 adults
- Exposure
- The algorithm's performance was compared against echocardiographic ejection fractions (EF)
Main finding
As reported in the abstract
Detection of EF ≤40%
| Group | Area under the receiver operating characteristic curve | Sensitivity | Specificity | Positive predictive value | Negative predictive value |
|---|---|---|---|---|---|
| Convolutional neural network (CNN) model | 0.85 | 77.5% | 78.3% | 20.3% | 98.0% |
Convolutional neural network (CNN) model
Area under the receiver operating characteristic curve0.85
Sensitivity77.5%
Specificity78.3%
Positive predictive value20.3%
Negative predictive value98.0%
Observational data: associations, not effects.
Further findings
False positives (abnormal artificial intelligence screen but EF >40%)
- Of false positives, EF between 41%-49%
- 25%
- Of false positives, conduction/rhythm abnormalities
- 63%
Subgroup analyses
- Consistent performance across various demographics and comorbidities
Original publication
Automated Detection of Reduced Ejection Fraction Using an ECG-Enabled Digital Stethoscope: A Large Cohort Validation. JACC Adv. Published 20 February 2025. doi:10.
Summarised from the published abstract; every number is checked against it