Predict Patient Self-reported Race from Skin Histological Images

2025-07-29

AI Summary

This study investigates whether deep learning models can predict patient-reported race from whole-slide skin pathology images, thereby exposing algorithmic bias in computational pathology linked to social determinants of health. Method: We propose the first multi-center framework integrating explainable attention mechanisms to identify race-associated morphological shortcuts; epidermal structure is validated as a key predictive cue via regional ablation and three confounding-control strategies. Contribution/Results: The model achieves AUCs of 0.799 (White) and 0.762 (Black) subgroups, with an overall AUC of 0.663. Performance drops significantly upon epidermal removal, confirming its central role. Our findings underscore the critical importance of rigorous data governance and incorporation of histological priors to mitigate AI bias. This work provides both methodological innovation—via interpretable, multi-site modeling—and empirical evidence supporting trustworthy computational pathology.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Bias, Fairness & Privacy
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Responsible Web: Algorithmic accountability and transparency on the web
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems
• Security and Privacy: Data transparency and provenance

Abstract

Artificial Intelligence (AI) has demonstrated success in computational pathology (CPath) for disease detection, biomarker classification, and prognosis prediction. However, its potential to learn unintended demographic biases, particularly those related to social determinants of health, remains understudied. This study investigates whether deep learning models can predict self-reported race from digitized dermatopathology slides and identifies potential morphological shortcuts. Using a multisite dataset with a racially diverse population, we apply an attention-based mechanism to uncover race-associated morphological features. After evaluating three dataset curation strategies to control for confounding factors, the final experiment showed that White and Black demographic groups retained high prediction performance (AUC: 0.799, 0.762), while overall performance dropped to 0.663. Attention analysis revealed the epidermis as a key predictive feature, with significant performance declines when these regions were removed. These findings highlight the need for careful data curation and bias mitigation to ensure equitable AI deployment in pathology. Code available at: https://github.com/sinai-computational-pathology/CPath_SAIF.

Problem

• Investigates AI's ability to predict patient race from skin histology images
• Identifies potential biases in deep learning models for pathology
• Explores morphological features linked to race in dermatopathology slides

Innovation

• AI predicts race from skin histology images
• Attention mechanism identifies race-associated features
• Epidermis key feature for racial prediction