Fig. 4

Detection validation against \(P_{\textrm{es}}\) reference. The validity of both approaches is examined by comparing automatically detected inspirations in the sEMG and manual annotations in the esophageal pressure. The upper plot overviews the total number of correctly and incorrectly detected patient efforts. The last entry shows how many of the detected breaths were assigned to an uncertain reference where the experts disagreed. The lower plot shows binary metrics to evaluate the detection performance. For sensitivity and positive predictive value, the distribution over patients and recordings is given. Black lines denote the median value, and white lines visualize the interquartile range