fig15
Figure 15. Effect of PCG information on the beat-level parameter estimation performance of the deep learning model. (A and B) correspond to DL + PCG and DL w/o PCG, respectively. Each configuration is evaluated on n = 61 valid test beats. Each dot represents an individual beat; the width of the half violin indicates the kernel density distribution of MAPE values, the horizontal line represents the median, and the diamond indicates the mean.
