Additional file 1 of Performance of GPT-4 for automated prostate biopsy decision-making based on mpMRI: a multi-center evidence study

Additional file 1. Methods. Fig. S1 Flowchart of patient screening, exclusions, and eligibility. Fig. S2 Subgroup analysis of confusion matrices showing pathological outcomes vs. Generative Pretrained Transformer (GPT)-4 biopsy recommendations across 3 cohorts. Fig. S3 Confusion matrices displaying pathological outcomes vs. Generative Pretrained Transformer (GPT)-4 recommendations for biopsy (n = 912). Fig. S4 Interface of the PostAIGuide platform. Table S1 Two examples of Chain of Thought (CoT) responses. Table S2 Two typical examples of “dialogues” with Generative Pretrained Transformer (GPT)-4 and the generated reports. Table S3 Quantitative results from the confusion matrix. Table S4 Three relatively challenging questions and GPT-4’s responses.