arXiv：cs.LG（机器学习，全量分类）· Abdullah Al Noman, Fahmid Al Rifat, Tahrima Hashem, Syed Muhammad Ibne Zulfiker, Rishov Paul, Tanzima HAshem·· 14 小时前AI 评分31
MIMIC-IV 死亡率预测公平性评估：超越人口统计平衡的多指标与交叉群体分析
Beyond Demographic Balance: Multi-Metric and Intersectional Evaluation of Fairness in MIMIC-IV Mortality Prediction
AI 导读
研究重新审视 MIMIC-IV 上 ICU 死亡率预测的公平性评估选择，比较多种公平性干预在预测效用与子群体误差指标上的表现。作者引入一种轻量适配策略，在不依赖死亡率结果的情况下联合平衡族裔—性别—保险的代表性，并在边际与三元交叉子群体层面评估其行为。结果显示，干预措施在准确率/AUROC、敏感性和假阳性率上可能得到显著不同的评价，边际人口统计摘要会掩盖其内部交叉群体的异质误差分布。
正文
Abstract:Fairness conclusions in clinical prediction can depend strongly on both the metrics reported and the demographic resolution at which performance is evaluated. We revisit these evaluation choices for ICU mortality prediction on MIMIC-IV, comparing predictive-utility and subgroup-error metrics across several fairness interventions. As a complementary case study, we introduce a lightweight adaptation strategy that jointly balances ethnicity--gender--insurance representation without conditioning on mortality outcomes, allowing demographic representation balancing to be examined separately from outcome-conditioned or direct error-rate interventions. We evaluate its behavior at both marginal and corresponding three-way intersectional subgroup levels, while accounting for the statistical support of finer-grained estimates. The results show that interventions can receive substantially different assessments across accuracy/AUROC, sensitivity, and false-positive rate, and that marginal demographic summaries can conceal heterogeneous error profiles within their constituent intersections, including among larger subgroups. These findings highlight the importance of evaluating fairness interventions at both complementary metric and subgroup resolutions, while accounting for the intervention target and the reliability of subgroup estimates.
| Comments: | NEurlPS TAE workshop 2026 accepted |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.01645 [cs.LG] |
| (or arXiv:2610.01645v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.01645 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tahrima Hashem [view email]
[v1]
Thu, 1 Oct 2026 13:09:42 UTC (1,415 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org