Figure 1.

Overview of the proposed framework. Curation often discards missing data, resulting in a discrepancy between the collected $\Omega$ and source datasets ${\Omega }_{\text{source}}$ used for training. Current practice is denoted in blue: naïve training and evaluating on ${\Omega }_{\text{source}}$ leads to biased estimates of performance and informativeness on target data. The orange path illustrates the proposed ICYM²I: a double inverse probability weighting (IPW) mechanism that yields accurate performance and informativeness estimates under the target distribution.