-
Notifications
You must be signed in to change notification settings - Fork 9
/
icd9_reader.py
36 lines (30 loc) · 1.28 KB
/
icd9_reader.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
import pandas as pd
import pickle
import numpy as np
class ICD9Reader():
def __init__(self, patient2hadmid_picklepath):
df = pd.read_csv('../mimic3/PROCEDURES_ICD.csv')
df = df[df.HADM_ID.notnull()]
df['HADM_ID'] = df['HADM_ID'].astype(int)
counts = df.ICD9_CODE.value_counts()
counts = counts[counts > 5]
self.icd9s = {item: i for i, item in enumerate(counts.index)}
with open(patient2hadmid_picklepath, 'rb') as f:
patient2hadmid = pickle.load(f)
self.patient2icd9 = {}
for patient in patient2hadmid.keys():
hadmid = patient2hadmid[patient]
icd9_hadmid = list(df[df.HADM_ID == hadmid].ICD9_CODE)
icd9_hadmid = list(filter(lambda x: x in self.icd9s, icd9_hadmid))
self.patient2icd9[int(patient)] = icd9_hadmid
del patient2hadmid
print("Finished building object for ICD9Reader: ", len(self.icd9s))
def get_ic9_onehot(self, fname):
def _get_patient_id_from_filename(fname):
return int(fname.split('_')[0])
a = [0]*len(self.icd9s)
l = self.patient2icd9[_get_patient_id_from_filename(fname)]
for item in l:
if item in self.icd9s:
a[self.icd9s[item]] = 1
return np.array(a)