Canada Research Chair in Medical AI @ uOttawa / CHEO RI & Scholar-in-Residence at the Office of the Information and Privacy Commissioner of Ontario
On 21st February we will be organizing a webinar looking at methods to address bias (e.g., based on race, sex, socio-economic status, and so on) in real-world datasets (RWD). The presentation will summarize an analysis of different methods to evaluate and mitigate bias in RWD, and some recommendations on which ones work and when they work. Bias mitigation using synthetic data generation works quite well for low to medium bias. In cases of extreme bias there are not methods that work consistently. You can see the results at the webinar.
