Fig. 4: Real-world study.
From: Addressing data heterogeneity in distributed medical imaging with heterosync learning
A ROC curves of the 16 learning methods’ testing efficacy for each hospital. B Performance stability across 16 learning methods for the PTC dataset. Violin plots show the distribution of AUC scores derived from five independent experimental replicates (n = 815 for each replicate). Each violin displays the kernel density estimation of the distribution. The embedded box shows the interquartile range (25th–75th percentiles), and whiskers extend to the minimum and maximum values. Source data are provided as a Source data file. Figures aregenerated using Python and subsequently composited in Adobe Illustrator 2023.
