Heart failure & cardiomyopathies
Artificial Intelligence Electrocardiogram and Left Ventricular Systolic Dysfunction in Kenya
Authors' conclusion
In this study, the AI-ECG algorithm demonstrated the potential clinical utility for screening of LVSD risk with high sensitivity and negative predictive value and may be particularly scalable in a resource-limited setting.
Study context
- Population
- Adults ≥18 years seeking routine care at outpatient facilities in Kenya
- Participants
- 1,444 adults
- Comparison
- Artificial intelligence electrocardiogram (AI-ECG) vs echocardiography for left ventricular systolic dysfunction (LVSD)
- Outcome
- Diagnostic performance (sensitivity, specificity, PPV, NPV) of AI-ECG for LVSD (LVEF <40%) confirmed on echocardiography
Methodological detailscross-sectional, 8 outpatient facilities, 3 prespecified risk strata, Enrollment June–December 2024; echocardiography within 7 days of ECG
- Cross-sectional
- 8 outpatient facilities
- 3 prespecified risk strata
- Enrollment June–December 2024; echocardiography within 7 days of ECG
Performance
As reported in the abstract
LVSD (LVEF <40%)
| Test or model | Sensitivity | Specificity | PPV | NPV | AUC |
|---|---|---|---|---|---|
| AI-ECG | 95.6% (95% CI, 91.8-97.7) | 79.4% (95% CI, 77.0-81.5) | 43.2% (95% CI, 38.7-47.9) | 99.1% (95% CI, 98.3-99.5) | 0.96 (95% CI, 0.95-0.97) |
AI-ECG
Sensitivity95.6% (95% CI, 91.8-97.7)
Specificity79.4% (95% CI, 77.0-81.5)
PPV43.2% (95% CI, 38.7-47.9)
NPV99.1% (95% CI, 98.3-99.5)
AUC0.96 (95% CI, 0.95-0.97)
- LVSD identified
- 204 (14.1%)
Further findings
AUC across CV risk strata
- Performance consistent across CV risk strata (AUC)
- 0.96-0.98
Original publication
Artificial Intelligence Electrocardiogram and Left Ventricular Systolic Dysfunction in Kenya. JAMA Cardiol. Published 1 June 2026. doi:10.
Summarised from the published abstract; every number is checked against it