Table 2.
AUC, sensitivity, and specificity for identification of reflux events on 24-hour pH/impedance studies by the machine learning system, currently available automated analysis software (Reflux Reader v6.1), and expert physician reader
| AUC (95% CI) | Sensitivity | Specificity | |
| Machine learning system | 0.87 (0.85–0.89) | 68.7%a | 80.8%b |
| Current software | 0.40 (0.37–0.42) | 61.1% | 18.6% |
| Expert physician reader | 0.83 (0.81–0.86) | 79.4% | 87.3% |
AUC, area under the curve; CI, confidence interval.
a
Based on a false positive rate for the expert physician reader of 13%.
b
Based on a true positive rate for the expert physician reader of 79%.