def extract_sldi(txt_file, output=True):
'''
沒有修改部份內容，使用須注意
負責解析(mapping) txt 檔案中的 term 到 CUI
輸入檔案為 sldi 模式
'''
concepts, error = mm.extract_concepts(filename=txt_file,
word_sense_disambiguation=True)
cui_list = list()
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if True: #concept.semtypes == "[clnd]" or concept.semtypes == "[dsyn]" :
if output:
print("Index:" + concept.index)
if output:
print("CUI:" + concept.cui, "SMT:" + concept.semtypes,
"POS:" + concept.pos_info,
"PREFER_N:" + concept.preferred_name,
"NEGATED:" + mmip.trigger_parser(concept.trigger),
"TRIGGER:" + concept.trigger)
cui_list.append(concept.cui)
return cui_list
def debug_extract_cui(patient_id, target_date, input_path=''):
if input_path == '':
input_txt_path = 'Clinical_Note/' + str(patient_id) + '/output/' + str(
target_date) + '_o.txt'
else:
input_txt_path = input_path
print("-----Processing file : " + input_txt_path + "-----")
sents, lines = read_line(input_txt_path)
index_list = range(1, lines + 1)
# 將剛剛讀出的行內容List, 傳給 mm.extract_concepts 取出概念
# concepts, error = mm.extract_concepts(
# sents, index_list, word_sense_disambiguation=True, user_define_acronyms=False, file_format= 'sldiID', )
concepts, error = mm.extract_concepts(
filename=input_txt_path,
word_sense_disambiguation=True,
user_define_acronyms=False,
file_format='sldiID',
)
number_for_everyone = 0
# 每個 concept 都有的數值, 紀錄現在印到了哪個 index
# 每個 index 只會印一次
whether_print = True
# 如果印出了 index 就不要印了 設為否
cui_list = list()
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if hasattr(concept, 'mm'):
if number_for_everyone != int(concept.index[3:6]):
whether_print = True
number_for_everyone += 1
if True: #is_target_smt(concept) :
# if True :
if whether_print:
if True:
print("Index:" + str(int(concept.index[3:6])))
whether_print = False
if True:
print("CUI:" + concept.cui, "SMT:" + concept.semtypes,
"POS:" + concept.pos_info,
"PREFER_N:" + concept.preferred_name,
"NEGATED:" + mmip.trigger_parser(concept.trigger),
"TRIGGER:" + concept.trigger)
cui_list.append(concept.cui)
elif hasattr(concept, 'ua'):
print("----UDA----")
print(" index :" + concept.index,
"short_form :" + concept.short_form,
"long_form :" + concept.long_form,
"POS :" + concept.pos_info)
print("-----------")
return cui_list
def cui_compare(patient_id, ndays, startday, method="smt", mode="and"):
'''
patient_id :輸入要比較的病人ID
method :使用的方法 semtaic type or similarity
ndays :往前比較幾天
startday :從第幾天開始
mode :比較的模式 and/or
'''
sm_type = {
"[clnd]": 0,
"[dsyn]": 0,
"[acab]": 0,
"[anab]": 0,
"[fndg]": 0,
"[inpo]": 0,
"[mobd]": 0,
"[neop]": 0,
"[patf]": 0,
"[sosy]": 0,
"[aapp]": 0,
"[antb]": 0,
"[bacs]": 0,
"[chem]": 0,
"[enzy]": 0,
"[hops]": 0,
"[horm]": 0,
"[imft]": 0,
"[inch]": 0,
"[lbpr]": 0,
"[medd]": 0,
"[nnon]": 0,
"[orch]": 0,
"[phsu]": 0,
"[topp]": 0,
"[vita]": 0,
}
# how to access use : sm_typpe['[clnd]']
smt_appeared_cur = list()
smt_appeared_before_ndays = list()
smt_appeared_all = list()
# ----------------------------------------------------------
date = ndays + 1
# cnt = today, ndays = n天前
for cnt in range(date, 0, -1):
print("***** 這是第", cnt, "天*****")
txt_file = str(patient_id) + '/' + str(patient_id) + '-' + str(
cnt) + '_o.txt'
sents, lines = read_line(txt_file)
index_list = range(1, lines + 1)
# 將剛剛讀出的行內容List, 傳給 mm.extract_concepts 取出概念
concepts, error = mm.extract_concepts(sents,
index_list,
word_sense_disambiguation=True,
derivational_variants=True,
user_define_acronyms=True,
file_format='sldiID')
number_for_everyone = 0
# 每個 concept 都有的數值, 紀錄現在印到了哪個 index
# 每個 index 只會印一次
whether_print = True
# 如果印出了 index 就不要印了 設為否
cui_list = list()
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if hasattr(concept, 'mm'):
if number_for_everyone != int(concept.index):
whether_print = True
number_for_everyone += 1
if is_target_smt(concept):
# 如果是目標內的smt 才做以下事情
# if True :
print("Index:" + concept.index)
if cnt == date:
pass
# smt_appeared_all = smt_appeared_cur
else:
pass
# is_appeared(concept.semtypes, smt_appeared_before_ndays)
print("CUI:" + concept.cui, "SMT:" + concept.semtypes,
"POS:" + concept.pos_info,
"NEGATED:" + mmip.trigger_parser(concept.trigger)[5],
"TRIGGER:" + concept.trigger)
cui_list.append(concept.cui)
elif hasattr(concept, 'ua'):
print("----UDA----")
print(" index :" + concept.index,
"short_form :" + concept.short_form,
"long_form :" + concept.long_form,
"POS :" + concept.pos_info)
print("-----------")
if cnt == date:
# print("第",cnt,"天的semantic type：",smt_appeared_cur)
sm_type = dict.fromkeys(sm_type, 0)
print("dict rest.")
smt_appeared_all = list(
set(smt_appeared_cur) | set(smt_appeared_before_ndays))
print("比較：第", date, "天 往前追朔", ndays, "天的Semantic type")
print("全部出現的 semantic type：", smt_appeared_all)
print("第", date, "天的semantic type：", smt_appeared_cur)
print(ndays, "天之間出現的 semantic type：", smt_appeared_before_ndays)
print("第", date, "天 跟前", ndays, "天相比，沒出現過的 SMT (新資訊) : ",
list(set(smt_appeared_cur) - set(smt_appeared_before_ndays)))
print(
"第", date, "天 跟前", ndays, "天相比，已出現過的 SMT (舊資訊) : ",
list(
set(smt_appeared_cur) -
(set(smt_appeared_cur) - set(smt_appeared_before_ndays))))
def get_all_cui_list(patient_id, date, output=True):
'''
回傳 某病人 某幾天病例中的所有 CUI 及 部份資訊：第幾句、位置、否定詞 \n
patient_id (str) : 指定某病人 \n
date (int) : 指定取回CUI 病例天數範圍\n
output (bool): 是否印出執行結果 訊息，預設為 是\n
'''
file_out = str(patient_id) + '/' + str(patient_id) + '_all_' + str(
date) + '_cui.txt'
for cnt in range(1, date + 1):
txt_file = str(patient_id) + '/' + str(patient_id) + '-' + str(
cnt) + '_o.txt'
print("-----Processing file : " + txt_file + "-----")
sents, lines = read_line(txt_file)
index_list = range(1, lines + 1)
# 將剛剛讀出的行內容List, 傳給 mm.extract_concepts 取出概念
concepts, error = mm.extract_concepts(sents,
index_list,
word_sense_disambiguation=True,
derivational_variants=True,
user_define_acronyms=True,
file_format='sldiID')
number_for_everyone = 0
# 每個 concept 都有的數值, 紀錄現在印到了哪個 index
# 每個 index 只會印一次
whether_print = True
# 如果印出了 index 就不要印了 設為否
cui_list = list()
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if hasattr(concept, 'mm'):
if number_for_everyone != int(concept.index):
whether_print = True
number_for_everyone += 1
if (concept.semtypes == "[clnd]"
or concept.semtypes == "[dsyn]"
or concept.semtypes == "[acab]"
or concept.semtypes == "[anab]"
or concept.semtypes == "[fndg]"
or concept.semtypes == "[inpo]"
or concept.semtypes == "[mobd]"
or concept.semtypes == "[neop]"
or concept.semtypes == "[patf]"
or concept.semtypes == "[sosy]"
or concept.semtypes == "[aapp]"
or concept.semtypes == "[antb]"
or concept.semtypes == "[bacs]"
or concept.semtypes == "[chem]"
or concept.semtypes == "[enzy]"
or concept.semtypes == "[hops]"
or concept.semtypes == "[horm]"
or concept.semtypes == "[imft]"
or concept.semtypes == "[inch]"
or concept.semtypes == "[lbpr]"
or concept.semtypes == "[medd]"
or concept.semtypes == "[nnon]"
or concept.semtypes == "[orch]"
or concept.semtypes == "[phsu]"
or concept.semtypes == "[topp]"
or concept.semtypes == "[vita]"):
# if True :
if whether_print:
if output:
print("Index:" + concept.index)
whether_print = False
if output:
print(
concept.cui, "POS:" + concept.pos_info,
"NEGATED:" +
mmip.trigger_parser(concept.trigger)[5])
f = open(file_out, 'a')
f.write(concept.cui + ", " + concept.index + ", " +
concept.pos_info + ", " +
mmip.trigger_parser(concept.trigger)[5] + '\n')
cui_list.append(concept.cui)
elif hasattr(concept, 'ua'):
print("----UDA----")
print(" index :" + concept.index,
"short_form :" + concept.short_form,
"long_form :" + concept.long_form,
"POS :" + concept.pos_info)
print("-----------")
f = open(file_out, 'a')
f.write(concept.long_form + ", " + concept.index + ", " +
concept.pos_info + '\n')
f.close()
def extract_sldiID(patient_id, number_of_note, output=True, no_filterd=False):
'''
update !!
負責解析(mapping) txt 檔案中的每一個句子中的每一個term 到 一個 concept
輸入檔案為 sldiID 模式
會將所有資訊輸出成csv 到 output 資料夾中
patient_id : 欲做 mapping 的病人ID
number_of_note : 這個病人有幾個病例
此 function 會回傳mapping 到的concpet 所有資訊
（ 如：cui, position, negation, smt_type...)
'''
for cnt in range(1, number_of_note + 1):
txt_file = 'Clinical_Note/' + str(patient_id) + '/output/' + str(
cnt) + '_o.txt'
# 80001/80001-1_o.txt
if no_filterd:
# 不指定，通通進來
output_file_path = 'Clinical_Note/' + str(
patient_id) + '/csv_all/' + str(cnt) + '.csv'
else:
output_file_path = 'Clinical_Note/' + str(
patient_id) + '/csv/' + str(cnt) + '.csv'
csvfile = open(output_file_path, 'w', newline='')
writer = csv.writer(csvfile)
writer.writerow([
'CUI', 'ROW', 'POSITION', 'NEGATION', 'TEXT_TRIGGER', 'SMT',
'TRIGGER'
])
print("-----Processing file : " + txt_file + "-----")
# sents, lines = read_line(txt_file)
# index_list = range(1, lines+1)
# 將剛剛讀出的行內容List, 傳給 mm.extract_concepts 取出概念
concepts, error = mm.extract_concepts(
filename=txt_file,
word_sense_disambiguation=True,
user_define_acronyms=True,
file_format='sldiID',
composite_phrase=4,
ignore_word_order=True,
prefer_multiple_concepts=True,
negated_setting=True,
term_processing=False,
allow_concept_gap=False,
)
number_for_everyone = 0
# 每個 concept 都有的數值, 紀錄現在印到了哪個 index
# 每個 index 只會印一次
whether_print = True
# 如果印出了 index 就不要印了 設為否
cui_list = list()
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if hasattr(concept, 'mm'):
if number_for_everyone != int(concept.index[3:6]):
whether_print = True
number_for_everyone += 1
if is_target_smt(concept) or no_filterd:
# if True :
if whether_print:
if output:
print("Index:" + str(int(concept.index[3:6])))
whether_print = False
if output:
print(
"CUI:" + concept.cui, "SMT:" + concept.semtypes,
"POS:" + concept.pos_info,
"PREFER_N:" + concept.preferred_name,
"NEGATED:" + mmip.trigger_parser(concept.trigger),
"TEXT_TRIGGER:" +
mmip.trigger_parser_getinfo(concept.trigger, 3),
"TRIGGER:" + concept.trigger)
writer.writerow([
concept.cui,
int(concept.index[3:6]), concept.pos_info,
mmip.trigger_parser(concept.trigger),
mmip.trigger_parser_getinfo(concept.trigger, 3),
concept.semtypes, concept.trigger
])
cui_list.append(concept.cui)
elif hasattr(concept, 'ua'):
print("----UDA----")
print(" Index :" + concept.index,
"short_form :" + concept.short_form,
"long_form :" + concept.long_form,
"POS :" + concept.pos_info)
print("-----------")
csvfile.close()
print("---Process complete !---")
return cui_list
from pymetamap import MetaMap
from mytool import mmi_parser as mmip
mm = MetaMap.get_instance('/home/feng/public_mm/bin/metamap20')
txt_file = "clinical_txt/2.txt"
concepts, error = mm.extract_concepts(filename=txt_file,
word_sense_disambiguation=True)
counter = 0
for concept in concepts:
'''
Concept :
Index 儲存行的資訊
semtype 儲存 semantic type 的資訊
'''
if concept.semtypes == "[clnd]" or concept.semtypes == "[dsyn]":
print("Index:" + concept.index)
print("CUI:" + concept.cui, "SMT:" + concept.semtypes,
"POS:" + concept.pos_info, "PREFER_N:" + concept.preferred_name,
"NEGATED:" + mmip.trigger_parser(concept.trigger)[5],
"TRIGGER:" + concept.trigger)
counter = counter + 1
print(counter)