Update Mimic4Dataset.py
Browse files- Mimic4Dataset.py +12 -41
Mimic4Dataset.py
CHANGED
@@ -239,36 +239,6 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
239 |
verif=False
|
240 |
return verif
|
241 |
|
242 |
-
def open_dict(self,cond, proc, out, chart, lab, med):
|
243 |
-
if cond:
|
244 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/condVocab", 'rb') as fp:
|
245 |
-
condDict = pickle.load(fp)
|
246 |
-
else :
|
247 |
-
condDict=None
|
248 |
-
if proc:
|
249 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/procVocab", 'rb') as fp:
|
250 |
-
procDict = pickle.load(fp)
|
251 |
-
else :
|
252 |
-
procDict=None
|
253 |
-
if out:
|
254 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/outVocab", 'rb') as fp:
|
255 |
-
outDict = pickle.load(fp)
|
256 |
-
else :
|
257 |
-
outDict=None
|
258 |
-
if chart:
|
259 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/chartVocab", 'rb') as fp:
|
260 |
-
chartDict = pickle.load(fp)
|
261 |
-
elif lab:
|
262 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/labsVocab", 'rb') as fp:
|
263 |
-
chartDict = pickle.load(fp)
|
264 |
-
else :
|
265 |
-
chartDict=None
|
266 |
-
if med:
|
267 |
-
with open("./data/dict/"+self.config.name.replace(" ","_")+"/medVocab", 'rb') as fp:
|
268 |
-
medDict = pickle.load(fp)
|
269 |
-
else :
|
270 |
-
medDict=None
|
271 |
-
return condDict, procDict, outDict, chartDict, medDict
|
272 |
###########################################################RAW##################################################################
|
273 |
|
274 |
def _info_raw(self):
|
@@ -462,11 +432,11 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
462 |
ins_encoder.fit(insVocab)
|
463 |
with open(filepath, 'rb') as fp:
|
464 |
dico = pickle.load(fp)
|
|
|
465 |
df = pd.DataFrame.from_dict(dico, orient='index')
|
466 |
-
|
467 |
for i, data in df.iterrows():
|
468 |
concat_cols=[]
|
469 |
-
dyn_df,cond_df,demo=concat_data(data,self.config.name.replace(" ","_"),self.feat_cond,self.feat_proc,self.feat_out, self.feat_chart, self.feat_meds,self.feat_lab
|
470 |
dyn=dyn_df.copy()
|
471 |
dyn.columns=dyn.columns.droplevel(0)
|
472 |
cols=dyn.columns
|
@@ -474,7 +444,6 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
474 |
for t in range(time):
|
475 |
cols_t = [str(x) + "_"+str(t) for x in cols]
|
476 |
concat_cols.extend(cols_t)
|
477 |
-
|
478 |
demo['gender']=gen_encoder.transform(demo['gender'])
|
479 |
demo['ethnicity']=eth_encoder.transform(demo['ethnicity'])
|
480 |
demo['insurance']=ins_encoder.transform(demo['insurance'])
|
@@ -482,7 +451,8 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
482 |
demo=demo.drop(['label'],axis=1)
|
483 |
X= generate_ml(dyn_df,cond_df,demo,concat_cols,self.concat)
|
484 |
X=X.values.tolist()[0]
|
485 |
-
|
|
|
486 |
size_concat = self.size_cond+ self.size_proc * interv + self.size_meds * interv+ self.size_out * interv+ self.size_chart *interv+ self.size_lab * interv + 4
|
487 |
size_aggreg = self.size_cond+ self.size_proc + self.size_meds+ self.size_out+ self.size_chart+ self.size_lab + 4
|
488 |
|
@@ -517,8 +487,9 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
517 |
def _generate_examples_deep(self, filepath):
|
518 |
with open(filepath, 'rb') as fp:
|
519 |
dico = pickle.load(fp)
|
|
|
520 |
for key, data in dico.items():
|
521 |
-
stat, demo, meds, chart, out, proc, lab, y = generate_deep(data, self.config.name.replace(" ","_"), self.feat_cond, self.feat_proc, self.feat_out, self.feat_chart, self.feat_meds,self.feat_lab
|
522 |
|
523 |
if self.verif_dim_tensor(proc, out, chart, meds, lab):
|
524 |
if self.data_icu:
|
@@ -546,7 +517,8 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
546 |
features = datasets.Features(
|
547 |
{
|
548 |
"label": datasets.ClassLabel(num_classes=2,names=["0", "1"]),
|
549 |
-
"
|
|
|
550 |
}
|
551 |
)
|
552 |
return datasets.DatasetInfo(
|
@@ -572,7 +544,7 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
572 |
if not desc.empty:
|
573 |
cond_text.append(desc['description'].to_string(index=False))
|
574 |
template = 'The patient is diagnosed with {}.'
|
575 |
-
cond_text = template.format(';'.join(cond_text))
|
576 |
else :
|
577 |
cond_text=''
|
578 |
|
@@ -590,21 +562,20 @@ class Mimic4Dataset(datasets.GeneratorBasedBuilder):
|
|
590 |
for mean_val, feat_label in zip(chart_mean, feat_text):
|
591 |
text = template.format(mean_val,feat_label)
|
592 |
chart_text.append(text)
|
593 |
-
chart_text='The chart events mesured are :
|
594 |
else:
|
595 |
chart_text=''
|
596 |
|
597 |
yield int(key),{
|
598 |
'label' : data['label'],
|
599 |
-
'
|
|
|
600 |
}
|
601 |
|
602 |
#############################################################################################################################
|
603 |
def _info(self):
|
604 |
self.path = self.init_cohort()
|
605 |
self.size_cond, self.size_proc, self.size_meds, self.size_out, self.size_chart, self.size_lab, eth_vocab,gender_vocab,age_vocab,ins_vocab=vocab(self.config.name.replace(" ","_"),self.feat_cond,self.feat_proc,self.feat_out,self.feat_chart,self.feat_meds,self.feat_lab)
|
606 |
-
self.outDict,self.chartDict,self.condDict,self.procDict,self.medDict = self.open_dict(self.feat_cond,self.feat_proc,self.feat_out, self.feat_chart, self.feat_lab, self.feat_meds)
|
607 |
-
|
608 |
if (self.encoding == 'concat' or self.encoding =='aggreg'):
|
609 |
return self._info_encoded()
|
610 |
|
|
|
239 |
verif=False
|
240 |
return verif
|
241 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
242 |
###########################################################RAW##################################################################
|
243 |
|
244 |
def _info_raw(self):
|
|
|
432 |
ins_encoder.fit(insVocab)
|
433 |
with open(filepath, 'rb') as fp:
|
434 |
dico = pickle.load(fp)
|
435 |
+
|
436 |
df = pd.DataFrame.from_dict(dico, orient='index')
|
|
|
437 |
for i, data in df.iterrows():
|
438 |
concat_cols=[]
|
439 |
+
dyn_df,cond_df,demo=concat_data(data,self.config.name.replace(" ","_"),self.feat_cond,self.feat_proc,self.feat_out, self.feat_chart, self.feat_meds,self.feat_lab)
|
440 |
dyn=dyn_df.copy()
|
441 |
dyn.columns=dyn.columns.droplevel(0)
|
442 |
cols=dyn.columns
|
|
|
444 |
for t in range(time):
|
445 |
cols_t = [str(x) + "_"+str(t) for x in cols]
|
446 |
concat_cols.extend(cols_t)
|
|
|
447 |
demo['gender']=gen_encoder.transform(demo['gender'])
|
448 |
demo['ethnicity']=eth_encoder.transform(demo['ethnicity'])
|
449 |
demo['insurance']=ins_encoder.transform(demo['insurance'])
|
|
|
451 |
demo=demo.drop(['label'],axis=1)
|
452 |
X= generate_ml(dyn_df,cond_df,demo,concat_cols,self.concat)
|
453 |
X=X.values.tolist()[0]
|
454 |
+
|
455 |
+
interv = (self.timeW//self.bucket) + 1
|
456 |
size_concat = self.size_cond+ self.size_proc * interv + self.size_meds * interv+ self.size_out * interv+ self.size_chart *interv+ self.size_lab * interv + 4
|
457 |
size_aggreg = self.size_cond+ self.size_proc + self.size_meds+ self.size_out+ self.size_chart+ self.size_lab + 4
|
458 |
|
|
|
487 |
def _generate_examples_deep(self, filepath):
|
488 |
with open(filepath, 'rb') as fp:
|
489 |
dico = pickle.load(fp)
|
490 |
+
|
491 |
for key, data in dico.items():
|
492 |
+
stat, demo, meds, chart, out, proc, lab, y = generate_deep(data, self.config.name.replace(" ","_"), self.feat_cond, self.feat_proc, self.feat_out, self.feat_chart, self.feat_meds,self.feat_lab)
|
493 |
|
494 |
if self.verif_dim_tensor(proc, out, chart, meds, lab):
|
495 |
if self.data_icu:
|
|
|
517 |
features = datasets.Features(
|
518 |
{
|
519 |
"label": datasets.ClassLabel(num_classes=2,names=["0", "1"]),
|
520 |
+
"COND" : datasets.Value(dtype='string', id=None),
|
521 |
+
"CHART/LAB" : datasets.Value(dtype='string', id=None),
|
522 |
}
|
523 |
)
|
524 |
return datasets.DatasetInfo(
|
|
|
544 |
if not desc.empty:
|
545 |
cond_text.append(desc['description'].to_string(index=False))
|
546 |
template = 'The patient is diagnosed with {}.'
|
547 |
+
cond_text = template.format('; '.join(cond_text))
|
548 |
else :
|
549 |
cond_text=''
|
550 |
|
|
|
562 |
for mean_val, feat_label in zip(chart_mean, feat_text):
|
563 |
text = template.format(mean_val,feat_label)
|
564 |
chart_text.append(text)
|
565 |
+
chart_text='The chart events mesured are : ' + '; '.join(chart_text)
|
566 |
else:
|
567 |
chart_text=''
|
568 |
|
569 |
yield int(key),{
|
570 |
'label' : data['label'],
|
571 |
+
'COND': cond_text,
|
572 |
+
'CHART/LAB': chart_text,
|
573 |
}
|
574 |
|
575 |
#############################################################################################################################
|
576 |
def _info(self):
|
577 |
self.path = self.init_cohort()
|
578 |
self.size_cond, self.size_proc, self.size_meds, self.size_out, self.size_chart, self.size_lab, eth_vocab,gender_vocab,age_vocab,ins_vocab=vocab(self.config.name.replace(" ","_"),self.feat_cond,self.feat_proc,self.feat_out,self.feat_chart,self.feat_meds,self.feat_lab)
|
|
|
|
|
579 |
if (self.encoding == 'concat' or self.encoding =='aggreg'):
|
580 |
return self._info_encoded()
|
581 |
|