What statistical tests would you use to validate that a newly deployed model is not exhibiting demographic bias across different patient populations?
