config.yaml

PHI:
  AGE:
    AGE: keep
  DATE:
    DATE: shift
    BIRTHDATE: shift
  CONTACT:
    PHONE: replace
    FAX: replace
    EMAIL: replace
    URL: replace
    IPADDRESS: replace
  ID:
    PATIENTID: replace
    STUDYID: replace
    OTHER: replace
  LOCATION:
    ORGANIZATION: replace
    HOSPITAL: replace
    STREET: replace
    CITY: replace
    ZIP: replace
    COUNTRY: replace
    STATE: replace
    OTHER: replace
  NAME:
    PATIENT: replace
    STAFF: replace
    OTHER: replace
  PROFESSION:
    PROFESSION: replace
    STATUS: replace
  OTHER:
    OTHER: replace

OFFSET:
  # Offset table must have the format: "patientid,timeoffset" -> timeoffset must be specified in miliseconds
  TABLE_PATH: /path/to/offset_table
  DATE_POSITIVE_BOUNDARY: 15
  DATE_NEGATIVE_BOUNDARY: -14
  AGE_POSITIVE_BOUNDARY: 3
  AGE_NEGATIVE_BOUNDARY: -2

RESULTS:
  JSON_INPUT_PATH: /path/to/json_bundle
  JSON_OUTPUT_PATH: /path/to/json_result
  TXT_OUTPUT_PATH: /path/to/text_result

INPUT:
  FORMAT: cas
  PATH: /path/to/cas
  TS: /path/to/ts
  # the filenames must have the structure "Patient:patientid_ResourceType:resourceid.dataending" for example "Patient:0a1a3cd_Observation:7d7bcb.xmi"
  # the name of the resource type can be customized with this parameter
  RESOURCE_TYPE: Observation

NAMES:
  MODEL: bert-base-multilingual-cased-ner-hrl_bert-base-multilingual-cased-ner-hrl
  TOKENIZER: Davlan/bert-base-multilingual-cased-ner-hrl
  DATASET: Dataset_Name