Help Center
How does the accuracy of Stethophone AI compare to traditional medical devices?
We have validated Stethophone algorithms for AI heart sound analysis in rigorous retrospective and clinical performance testing. Stethophone AI has demonstrated high-performance results for the primary indicators (murmur detection and heart sound timing). Its sensitivity, specificity, and accuracy are on par with popular commercially available digital stethoscopes (see the validation results in the table below).
Murmur detection
| Metric | Dataset | Test result |
|---|---|---|
| Sensitivity | Multi-Device | 93.0 95% CI: (91.9 to 94.2) |
| Specificity | Multi-Device | 94.5 95% CI: (93.6 to 95.4) |
| Accuracy | Multi-Device | 93.9 95% CI: (93.1 to 94.6) |
| ROC AUC | Multi-Device | 97.9 95% CI: (97.5 to 98.3) |
Heart sound timing
| Metric | Dataset | Test result |
|---|---|---|
| S1 PPV (precision) | Multi-Device | 96.9 95% CI: (96.5 to 97.3) |
| S1 sensitivity (recall) | Multi-Device | 97.9 95% CI: (97.7 to 98.1) |
| S2 PPV (precision) | Multi-Device | 97.1 95% CI: (96.7 to 97.4) |
| S2 sensitivity (recall) | Multi-Device | 97.7 95% CI: (97.5 to 97.9) |