Why were Phenotyping tasks not chosen as prediction tasks?

[johntiger1]

I am comparing your benchmark against the one by Harutyunyan et al. (https://github.com/YerevaNN/mimic3-benchmarks). Is there a reason why your benchmark suite doesn't look at phenotyping? I know you explained why you picked certain tasks in the paper.

But I am curious, it seems like phenotyping could have been done easily (it's just another column, right?)

Thanks