def main(model: str, text: str = None, cuda: int = None):
if text is None:
text = "The patient was prescribed 1 capsule of Advil for five days."
if model == 'crf':
print('using CRF model')
model = Model.load_external('medacy_model_clinical_notes')
else:
print('using BERT model')
if cuda is not None:
from medacy_bert_model_clinical_notes import medacy_bert_model_clinical_notes
medacy_bert_model_clinical_notes.PIPELINE_ARGS[
'cuda_device'] = cuda
model = Model.load_external('medacy_bert_model_clinical_notes')
anons = model.predict(text)
for anon in anons:
print(anon)
df_filtered['tweet_text'] = df_filtered['tweet_text'].apply(
lambda x: remove_emoji(x))
#Getting the annotations from each of the tweet's text
#Part of the following code was obtained from here:
#https://github.com/NLPatVCU/medaCy/
from medacy.model.model import Model
df_medacy_annotations = pd.DataFrame(columns=[
'Tweet_id', 'Text_section', 'Span_start', 'Span_end', 'Annotation_type',
'Extras'
])
df_medacy_tweets_tagged = pd.DataFrame(columns=['Tweet_id', 'Tweet_full_text'])
model = Model.load_external('medacy_model_clinical_notes')
print("Configuring the Medacy tagger. Please wait...")
for index, row in df_filtered.iterrows():
annotation = model.predict(row['tweet_text'])
if len(annotation) > 0:
df_medacy_tweets_tagged.loc[len(df_medacy_tweets_tagged.index)] = [
row['tweet_id'], row['tweet_text']
]
for i in range(len(annotation)):
df_medacy_annotations.loc[len(df_medacy_annotations)] = [
row['tweet_id'], annotation.annotations[i][3],
annotation.annotations[i][1], annotation.annotations[i][2],
