Presented our research on auditing fairness in dermatological AI systems, focusing on the intersection of group fairness and counterfactual fairness in clinical decision support. This work, co-authored with Lauraine Tiogning-Djiogue, examined how AI diagnostic systems may produce disparate outcomes across demographic groups, particularly regarding skin cancer detection across gender and skin tone.

Key contributions discussed:

• Novel framework for auditing both group-level and counterfactual fairness in medical AI
• Analysis of bias in dermatological image classification systems
• Generation of biologically plausible counterfactual examples for fairness evaluation
• Empirical evaluation on dermatology datasets with intersectional demographic attributes
• Practical implications for clinical deployment of AI diagnostic tools