def get_prescribe_df(no):
'''
환자번호를 넣으면 column은 KCDcode, row는 time-serial의 형태인 dataframe이 나오는 함수
'''
global prescribe_output_path, MEDI_USE_COLS
store_pres = pd.HDFStore(prescribe_output_path,mode='r')
if not '/metadata/usecol' in store_pres.keys():
set_prescribe_row(); store_pres = pd.HDFStore(prescribe_output_path,mode='r')
col_list = get_timeseries_column()
# create empty dataframe
use_prescribe_values = store_pres.select('metadata/usecol').col.values
result_df = pd.DataFrame(columns=col_list,index=use_prescribe_values)
# target patient dataframe
target_df = store_pres.select('data',where='no=={}'.format(no))
for value in target_df.values:
_ , _medi_code, _date, _times = value
if _medi_code in use_prescribe_values:
start_index = result_df.columns.get_loc(_date)
end_index = start_index + _times + 1
result_df.loc[_medi_code].iloc[start_index:end_index] = 1
del target_df
index_name_dict = get_index_name_map()
result_df.index = result_df.index.map(index_name_dict.get)
return result_df
def get_demo_df(no):
per_demo_series = get_demographic_series(no)
per_demo_df = pd.DataFrame(index=per_demo_series.index,
columns=get_timeseries_column())
for column in per_demo_df.columns:
per_demo_df[column] = per_demo_series
return per_demo_df
def get_patient_timeseries_label(no,label_name):
global PREP_OUTPUT_DIR, SAMPLE_PATIENT_PATH
# syntax checking existence for directory
PREP_OUTPUT_DIR = check_directory(PREP_OUTPUT_DIR)
output_path = PREP_OUTPUT_DIR + SAMPLE_PATIENT_PATH
x = pd.HDFStore(output_path,mode='r').select('data/{}'.format(label_name))
result_series = pd.Series(index=get_timeseries_column())
for _,x_date,x_label in x[x.no==no].values:
result_series[x_date]=x_label
return result_series
def get_labtest_df(no):
'''
환자번호를 넣으면 column은 KCDcode, row는 time-serial의 형태인 dataframe이 나오는 함수
'''
global labtest_output_path
store_lab = pd.HDFStore(labtest_output_path,mode='r')
if not '/metadata/usecol' in store_lab.keys():
store_lab.close()
set_labtest_row();
else:
store_lab.close()
store_lab = pd.HDFStore(labtest_output_path,mode='r')
col_list = get_timeseries_column()
# create empty dataframe
use_labtest_values = store_lab.select('metadata/usecol').col.values
result_df = pd.DataFrame(columns=col_list, index=use_labtest_values)
lab_node = store_lab.get_node('data')
for lab_name in lab_node._v_children.keys():
result_lab_series = result_df.loc[lab_name]
target_df = store_lab.select('data/{}'.format(lab_name),where='no=={}'.format(no))
target_df = target_df.groupby(['no','date']).mean().reset_index() # 같은달에 한번 이상 했을 시, 결과의 평균으로 저장
pre_value = None
for value in target_df.values:
_, _date, _result = value
result_df.loc[lab_name].loc[_date] = _result
# if pre_value is not None:
# min_time = pre_value[1];max_time =_date
# min_value = pre_value[2];max_value = _result
# inter_time = get_time_interval(min_time,max_time)
# for i in result_lab_series.loc[pre_value[1]:_date].index:
# intpol_time = get_time_interval(min_time,i) / inter_time
# intpol_value = (max_value-min_value)*intpol_time + min_value
# result_df.loc[lab_name].loc[i] = intpol_value
# pre_value = value
# else :
# result_df.loc[lab_name].loc[:_date] = _result
# pre_value = value
# missing_index = result_lab_series[result_lab_series.isnull()].index
# if missing_index.min() > col_list[0]:
# last_loc = result_df.columns.get_loc(missing_index.min()) - 1
# result_df.loc[lab_name].iloc[last_loc+1:]= result_df.loc[lab_name].iloc[last_loc]
del target_df
store_lab.close()
return result_df
def get_diagnosis_df(no):
'''
환자번호를 넣으면 column은 KCDcode, row는 time-serial의 형태인 dataframe이 나오는 함수
'''
global diagnosis_output_path, KCD_USE_COLS, DIAG_TIME_INTERVAL
store_diag = pd.HDFStore(diagnosis_output_path, mode='r')
if not '/metadata/usecol' in store_diag.keys():
set_diagnosis_row()
store_diag = pd.HDFStore(diagnosis_output_path, mode='r')
col_list = get_timeseries_column()
# create empty dataframe
use_diagnosis_values = store_diag.select('metadata/usecol').col.values
result_df = pd.DataFrame(columns=col_list, index=use_diagnosis_values)
# target paitent dataframe
target_df = store_diag.select('data', where='no=={}'.format(no))
target_df = target_df.sort_values(['KCD_code', 'date'], axis=0)
KCD_code_i = KCD_USE_COLS.index('KCD_code')
date_i = KCD_USE_COLS.index('date')
_prev_value = []
for value in target_df.values:
_, curr_code, curr_date = value
if curr_code in use_diagnosis_values:
if len(_prev_value) > 0 and _prev_value[KCD_code_i] == curr_code:
prev_date = _prev_value[date_i]
if get_time_interval(prev_date,
curr_date) <= DIAG_TIME_INTERVAL:
start_index = result_df.columns.get_loc(prev_date)
end_index = result_df.columns.get_loc(curr_date)
result_df.loc[int(
value[KCD_code_i])].iloc[start_index:end_index] = 1
result_df.loc[curr_code].loc[curr_date] = 1
_prev_value = value
del target_df
index_name_dict = get_index_name_map()
result_df.index = result_df.index.map(index_name_dict.get)
return result_df