MedEqualizer: A Framework Investigating Bias in Synthetic Medical Data and Mitigation via Augmentation CCS: Applied computing Health informatics CCS: Computing methodologies Machine learning algorithms
Sama Salarian, Yue Zhang, Swati Padhee, Srinivasan Parthasarathy · 2025
Read MedEqualizer: A Framework Investigating Bias in Synthetic Medical Data and Mitigation via Augmentation CCS: Applied computing Health informatics CCS: Computing methodologies Machine learning algorithms by Sama Salarian, Yue Zhang, Swati Padhee, Srinivasan Parthasarathy (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-01054.txt.
Description
Sama Salarian and colleagues examine fairness in synthetic medical data. Made up patient records promise wider research access without exposing real patients, but if the synthetic data skew the mix of demographic groups, studies built on them can mislead clinical decisions.
Using the MIMIC-III hospital data set, they measure how faithfully several GAN based tools cover protected subgroups, using a logarithmic disparity metric. Many subgroups came out badly over or under represented compared with the real data. Their remedy, MedEqualizer, is a model agnostic augmentation step that enriches underrepresented groups before generation begins. With it, the new data sets came out far more balanced across groups.
Catalog record
| CREATOR | Sama Salarian, Yue Zhang, Swati Padhee, Srinivasan Parthasarathy |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 43 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-01054