Simulated Bias in Artificial Medical Images (SimBA) - JAMIA 2024

Artificial intelligence (AI) models trained using medical images for clinical tasks often exhibit bias in the form of subgroup performance disparities. However, since not all sources of bias in real-world medical imaging data are easily identifiable, it is challenging to comprehensively assess their impacts. We introduced Simulated Bias in Artificial Medical Images (SimBA), an analysis framework for systematically and objectively investigating the impact of biases in medical images on AI models.
<br>
This repository contains SimBA data in which a localized morphological deformation in the brain (the “bias effect”) is spuriously correlated with a “disease” classification task. AI models trained on this data are susceptible to shortcut learning and consequently, subgroup performance disparities. The three datasets in this repository include counterfactual dataset scenarios where the bias effect is “near" to the disease region, "far" from the disease region, and where the bias effect was not present.