Figure 3.
Heat map of models trained on different datasets and evaluated on testing datasets. Each model is evaluated on three different testing datasets. The performance is assessed using the following metrics: (A) sensitivity (SN), (B) accuracy (ACC), (C) area under the receiver operating characteristic curve (AUROC), (D) Matthews correlation coefficient (MCC), (E) F1-score, and (F) the area under the precision–recall curve (AUPRC).
