ECG-based artificial intelligence for obstructive sleep apnea detection: a mechanism-oriented diagnostic meta-analysis.
Sleep BreathSep 16, 2026 (epub)
Ayşe Bulut, Ömer Engin Bulut
✦ AI-curated · Sources linked
In 30 seconds
This meta-analysis evaluated the diagnostic performance of AI models using ECG signals for detecting obstructive sleep apnea (OSA) across 13 studies. The pooled sensitivity and specificity were 0.88 and 0.89, respectively, indicating high diagnostic accuracy. The study highlights that AI may identify physiological responses rather than direct airway obstruction.
Key findings
- Pooled sensitivity for AI models in OSA detection was 0.88.
- Pooled specificity for AI models in OSA detection was 0.89.
- The summary receiver operating characteristic curve showed an AUC > 0.90.
- AI models demonstrated consistent performance across various architectures.
Why it matters
Understanding the diagnostic capabilities of AI in OSA detection can enhance clinical decision-making and improve patient outcomes. High sensitivity and specificity indicate that these models could serve as reliable tools in identifying OSA.