An Investigation Into Race Bias in Random Forest Models Based on Breast DCE-MRI Derived Radiomics Features
This addresses bias in AI models for breast cancer diagnosis, specifically for patients of different races, and is incremental as it extends bias research from deep learning to classical methods.
The study investigated race bias in random forest models using radiomics features from breast DCE-MRI data, finding that these features contain race-identifiable information with 60-70% accuracy in predicting race and that models trained on race-imbalanced data show biased performance, performing better on the race they were trained on.
Recent research has shown that artificial intelligence (AI) models can exhibit bias in performance when trained using data that are imbalanced by protected attribute(s). Most work to date has focused on deep learning models, but classical AI techniques that make use of hand-crafted features may also be susceptible to such bias. In this paper we investigate the potential for race bias in random forest (RF) models trained using radiomics features. Our application is prediction of tumour molecular subtype from dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) of breast cancer patients. Our results show that radiomics features derived from DCE-MRI data do contain race-identifiable information, and that RF models can be trained to predict White and Black race from these data with 60-70% accuracy, depending on the subset of features used. Furthermore, RF models trained to predict tumour molecular subtype using race-imbalanced data seem to produce biased behaviour, exhibiting better performance on test data from the race on which they were trained.