Auditing Significance, Metric Choice, and Demographic Fairness in Medical AI Challenges
This work addresses critical issues in medical AI evaluation for challenge organizers and participants, though it is incremental as it builds on existing challenge frameworks.
The paper tackled limitations in medical AI challenge leaderboards by developing RankInsight, a toolkit that addresses statistical significance, metric choice, and demographic fairness, revealing that nnU-Net outperforms other models with high certainty and that metric changes can reverse top model rankings.
Open challenges have become the de facto standard for comparative ranking of medical AI methods. Despite their importance, medical AI leaderboards exhibit three persistent limitations: (1) score gaps are rarely tested for statistical significance, so rank stability is unknown; (2) single averaged metrics are applied to every organ, hiding clinically important boundary errors; (3) performance across intersecting demographics is seldom reported, masking fairness and equity gaps. We introduce RankInsight, an open-source toolkit that seeks to address these limitations. RankInsight (1) computes pair-wise significance maps that show the nnU-Net family outperforms Vision-Language and MONAI submissions with high statistical certainty; (2) recomputes leaderboards with organ-appropriate metrics, reversing the order of the top four models when Dice is replaced by NSD for tubular structures; and (3) audits intersectional fairness, revealing that more than half of the MONAI-based entries have the largest gender-race discrepancy on our proprietary Johns Hopkins Hospital dataset. The RankInsight toolkit is publicly released and can be directly applied to past, ongoing, and future challenges. It enables organizers and participants to publish rankings that are statistically sound, clinically meaningful, and demographically fair.