Adversarial debiasing for age-equitable diabetes prediction: performance-fairness trade-offs and partition dependency...

Machine learning models used for diabetes risk prediction may encode age-related biases that reduce diagnostic accuracy for specific demographic groups. Adversarial debiasing with a gradient reversal layer (GRL) offers a theoretically principled approach to learning representations that are invariant to a protected attribute; however, its practical effectiveness under realistic conditions of subgroup imbalance in healthcare datasets has not been fully characterised.

Does adversarial debiasing with a GRL improve age-equitable diabetes prediction, and how do its fairness effects vary across different data partitions?

Adversarial debiasing was evaluated for age-bias mitigation in diabetes prediction using the publicly available Pima Indians Diabetes Database (n = 768). All eight dataset predictors were used; three age groups (<30, 30-50, and >50 years) were derived from age for fairness evaluation. An adversarial neural model with a gradient reversal layer was compared against a logistic regression baseline. Features were standardised using a scaler fitted on training data only. The train-test split was stratified by diabetes outcome. Overall performance metrics (accuracy, recall, ROC-AUC) and the recall parity gap across age groups were computed on a primary labelled test partition (n = 154); robustness was assessed across five independent random seeds (0-4).

On the primary test partition, the adversarial model improved recall for the smallest age group [>50 years: 0.5556 → 0.7778, +22.22 percentage points (pp)] while maintaining comparable overall discrimination (ROC-AUC: 0.7852 → 0.7896, +0.45 pp). However, the recall parity gap increased from 0.0996 to 0.2153 (+11.57 pp), reflecting a concurrent decline in recall for the <30-year group (-6.25 pp). Across five random seeds, the mean recall parity gap showed a modest mean reduction (0.3282 → 0.3033, -2.49 pp), but with high variability (SD > 0.27) exceeding the mean difference. The adversarial model reduced the fairness gap in three of five seeds, increased it in one, and produced no change in one.

Adversarial debiasing can improve predictive recall for underrepresented demographic subgroups but does not guarantee consistent fairness improvements across data partitions, particularly when subgroup sample sizes are small. Multi-seed evaluation is essential for reliable fairness assessment; single train-test splits are insufficient.