Table 11.

BAC of Each Model (4-Fold Cross-Validation)

Model	BAC
yolov5n	0.71 ± 0.07
yolov5s	0.72 ± 0.05
yolov5m	0.58 ± 0.04
yolov5l	0.69 ± 0.05
yolov5x	0.70 ± 0.10

Abbreviation: BAC, balanced accuracy.

The classification task consists of correctly labeling each image as “none,” “mild,” “moderate,” or “severe” urticaria based on the number of hives detected by the model (itch severity was not taken into account in the present work). For each fold, we calculated BAC and then aggregated the results (mean and SD).