def write_diagnoses(root_patient_folder,
hadm_id,
output_dir,
type="low",
pheno_map=None):
if type == "low":
events = read_diagnoses(root_patient_folder)
events = events.loc[events["HADM_ID"] == hadm_id]
events["EVENT_ID"] = events['ICD9_CODE']
events = events[['SUBJECT_ID', 'HADM_ID', 'EVENT_ID']]
write_static_event(events, output_dir, "diag")
variables = var_map.VARIABLE.unique()
for subject_dir in tqdm(os.listdir(args.subjects_root_path),
desc='Iterating over subjects'):
dn = os.path.join(args.subjects_root_path, subject_dir)
try:
subject_id = int(subject_dir)
if not os.path.isdir(dn):
raise Exception
except:
continue
try:
# reading tables of this subject
stays = read_stays(os.path.join(args.subjects_root_path, subject_dir))
diagnoses = read_diagnoses(
os.path.join(args.subjects_root_path, subject_dir))
events = read_events(os.path.join(args.subjects_root_path,
subject_dir))
except:
sys.stderr.write(
'Error reading from disk for subject: {}\n'.format(subject_id))
continue
episodic_data = assemble_episodic_data(stays, diagnoses)
# cleaning and converting to time series
events = map_itemids_to_variables(events, var_map)
events = clean_events(events)
if events.shape[0] == 0:
# no valid events for this subject
continue
for subject_dir in os.listdir(args.subjects_root_path):
dn = os.path.join(args.subjects_root_path, subject_dir)
try:
subject_id = int(subject_dir)
if not os.path.isdir(dn):
raise Exception
except:
continue
sys.stdout.write('Subject {}: '.format(subject_id))
sys.stdout.flush()
try:
sys.stdout.write('reading...')
sys.stdout.flush()
stays = read_stays(os.path.join(args.subjects_root_path, subject_dir))
diagnoses = read_diagnoses(os.path.join(args.subjects_root_path, subject_dir))
events = read_events(os.path.join(args.subjects_root_path, subject_dir))
except:
sys.stdout.write('error reading from disk!\n')
continue
else:
sys.stdout.write('got {0} stays, {1} diagnoses, {2} events...'.format(stays.shape[0], diagnoses.shape[0], events.shape[0]))
sys.stdout.flush()
episodic_data = assemble_episodic_data(stays, diagnoses)
sys.stdout.write('cleaning and converting to time series...')
sys.stdout.flush()
events = map_itemids_to_variables(events, var_map)
events = clean_events(events)
if events.shape[0] == 0: