Adversarial debiasing for age-equitable diabetes prediction: performance–fairness trade-offs and partition dependency in machine learning - Takeaways - MDSpire

Adversarial debiasing for age-equitable diabetes prediction: performance–fairness trade-offs and partition dependency in machine learning

  • By

  • Vinod Kumar Yata

  • Sravanthi Jena

  • Meera Indracanti

  • Shivaprasad Chitta

  • Narasaiah Kolliputi

  • July 17, 2026

Share

  • 1

    Adversarial debiasing with a gradient reversal layer was evaluated for age-bias mitigation in diabetes risk prediction.

  • 2

    The study used the Pima Indians Diabetes Database, analyzing performance across three age groups: 50 years.

  • 3

    The adversarial model improved recall for the >50 age group while increasing the recall parity gap for the <30 age group.

  • 4

    Mean recall parity gap showed modest reduction across five random seeds, but with high variability exceeding the mean difference.

  • 5

    Adversarial debiasing can enhance predictive recall for underrepresented groups but does not ensure consistent fairness improvements.

Original Source(s)

Adversarial debiasing for age-equitable diabetes prediction: performance–fairness trade-offs and partition dependency in machine learning

  • by Vinod Kumar Yata, Sravanthi Jena, Meera Indracanti, Shivaprasad Chitta, Narasaiah Kolliputi

  • July 17, 2026

Related Content

Predictive value and clinical guidance of FNA-Tg combined with BRAF V600E for PTC cervical lymph node metastasis: a case report

Who can access testosterone hormone medication in the U.S.?

Determinants of metabolic dysfunction associated steatotic liver disease and liver fibrosis in patients with type 2 diabetes mellitus: a cross-sectional study