Supplementary materials for: "Evaluation of ChatGPT in Generating Dermatology Vignettes from Clinical Images"

Supplementary Text 1: Image Selection, Prompts, Rubric, and Vignettes Generated from Images
Supplementary Text 2: Additional Analysis, Supplementary Figure 1, Table 3, and Vignettes Generated from Diagnosis Labels
Supplementary Table 1: Ratings across five metrics by three dermatologists and their averages for vignettes generated based on 25 clinical images sourced from ISIC
Supplementary Table 2.1: Manual annotation of skin lesion category on answer options for vignettes generated with images (diagnosis + image)
Supplementary Table 2.2: Manual annotation of skin lesion category on answer options for vignettes generated without images (diagnosis-only)

补充文本1：图像选取、提示词（Prompt）、评分细则及基于图像生成的示例情境（Vignettes）
补充文本2：补充分析、补充图1、表3及基于诊断标签生成的示例情境（Vignettes）
补充表1：三名皮肤科医师针对取自国际皮肤影像协作组（ISIC）的25张临床图像生成的示例情境，在五项评估指标上的评分及其平均分
补充表2.1：针对结合图像（诊断+图像）生成的示例情境，其备选答案的皮肤病变类别人工标注结果
补充表2.2：针对仅基于诊断（无图像）生成的示例情境，其备选答案的皮肤病变类别人工标注结果