摘要
 

Bias and discrimination in Artificial Intelligence (AI) systems has been widely studied including tasks like AI-based facial recognition system. However, such bias has not been fully discussed in AI healthcare applications such as skin cancer detection, mortality prediction and healthcare utilization prediction algorithms. The state-of-the-art AI models have been shown to produce significant differences in underdiagnosis rates across racial and other demographic groups even when models only have access to the image itself. In this talk, I will first introduce the bias problem in medical AI and then show you the experiments we conducted to explore the problem. Finally, I will give you the assumptions we made from the results and show you the pipeline we came up with to eliminate the bias in the prediction of the AI imaging model.

---