AUROC discrepancy between Figure 1c and Figure 5

[YXNTU]

Hi authors,

Thanks for sharing this impressive work.

While reading the paper, I noticed a numerical discrepancy regarding the performance scores for the Risk Prediction Tasks.

• In Figure 1c, the score for EHR-R1-72B is reported as 0.9328.
• However, in Figure 5, the result for EHR-R1-72B is stated as 0.9523.

Could you clarify if this is a reporting error in the figures, or if I might have misunderstood the evaluation settings?

Thanks!