Institute of Information Science, Academia Sinica

Fairness in Artificial Intelligence (AI) systems has been widely studied including tasks such as AI-based computer vision (CV) systems. However, the bias has not been fully discussed in AI healthcare applications such as skin cancer detection, mortality prediction and healthcare utilization prediction algorithms. For example, the state-of-the-art AI models have recently been shown to produce significant differences in under-diagnosis rates across racial and other demographic groups even when models only have access to the image itself. If the AI model itself cannot make fair decisions across different groups, how can we trust AI and put it into the clinical frontline. In this talk, I will first introduce the bias problem in medical AI and then show you how we can approach it. Finally, I will discuss possible solutions to eliminate the bias in healthcare AI models.