Define the dead and alive patient cohorts for medGAN

[yy6linda]

create two feature sets for the dead and alive patient cohorts

[yy6linda]

I used ivan's model which generated a sparse matrix with 42549 features, I used todense methods for feeding the model into the medgan
Here I didn't use the full alive set, all the death set is kept and alive set is randomly selected based on death: alive = 1:5
Death: (8309,42549)
Alive (613577,42549)
when training death.npy, due to the small size of death.npy, use replace = True

 File "train.py", line 96, in <module>
    saveMaxKeep=args.save_max_keep)
  File "/data/users/yanyao/myproj/synpuf/omop/app/1000-1000-death-SynthEHR/model.py", line 391, in train
    batchIdx = np.random.choice(idx, size=batchSize, replace=False)
  File "mtrand.pyx", line 948, in numpy.random.mtrand.RandomState.choice
ValueError: Cannot take a larger sample than population when 'replace=False'
"nohup.out" 3065L, 5097612C    

[yy6linda]

nohup python train.py --data_file './death.npy' --n_pretrain_epoch 1000 --n_epoch 1000&

synthetic death.npy
/data/users/yanyao/myproj/synpuf/omop/app/1000-1000-death-SynthEHR
synthetic alive.npy
/data/users/yanyao/myproj/synpuf/omop/app/1000-1000-alive-SynthEHR

[yy6linda]

only use condition:
(8309, 28664)
(41545, 28664)
Inside 28664, is 28539 condition_source_value, 116 year of birth, 3 gender concept_id , 6 race_concept_id