Data approach and bias treatment

Thanks for the talk! I am curious about how do you select your approach to deal with the dataset in this paper, and what do you think about the individual differences related to drug and ways to deal with the bias?