Version 15 (modified by 13 months ago) (diff) | ,
---|
Medical text analysis
Annotation manual
The goal is to mark the occurrence of known medical entities or other important information in the text of health records.
You will familiarize yourself with the BRAT annotation tool. After logging in with credentials assigned to you (https://nlp.fi.muni.cz/trac/research/wiki/BratManual), you will be able to highlight the entities to be annotated, after which a pop-up will appear, letting you choose the category of the entity you highlighted.
Within the pop-up, you will choose the category of the annotation and in the Notes field you will be able to enter or correct additional content relevant to the annotation (in the case of Abbreviations, you may correct the meaning of it, in the case of other entities, you can express the degree of certainty of your category choice if unsure).
Entities to be annotated
- AnatomicalSite
- names of body parts and locations on the body
- every AnatomicalSite annotation is either of these two:
- AnatomicalSite_name: the name itself, e.g.
našla v pravém prsu bulku
- AnatomicalSite_laterality: further specification of location, e.g. v
našla v pravém prsu bulku
- AnatomicalSite_name: the name itself, e.g.
- DiseaseDisorder
- names of diseases and disorders, e.g.
léčena xareltem inf mononukleoza v 15 letech
- names of diseases and disorders, e.g.
- SignSymptom
- medical occurrences which are not names of diseases and disorders but can indicate their presence or absence, e.g.
při bolestech svalů, teplotě
If unsure whether it is an official disorder name or only a symptom, annotate as SignSymptom.
- medical occurrences which are not names of diseases and disorders but can indicate their presence or absence, e.g.
- Procedure
- Medication
- name of a drug or substance
- every Medication annotation is either
- Medication_name: the name of the product, e.g.
adjuv HT - Tamoxifen od 6/2018 s ef
- Medication_strength: the strength of the product, e.g.
večer: Lexaurin 1,5 mg p.o. dle potřeby
- Medication_dosage: an expression of recommended dosage, e.g.
METAMIZOL STADA [TBL NOB] 500 mg / p.o. [při bolesti 1 tbl]
- Medication_name: the name of the product, e.g.
- Lab
- measurements related to the body, e.g. blood test values
- every Lab annotation is either
- Lab_name: the name of the measured aspect, e.g.
S_Estradiol: 18 pmol/l, S_FSH: 3.8 IU/l
- Lab_value: the value, mostly numeric, measured, e.g.
S_Estradiol: 18 pmol/l, S_FSH: 3.8 IU/l
- Lab_unit: the unit of measurement, e.g.
S_Estradiol: 18 pmol/l, S_FSH: 3.8 IU/l
- Lab_name: the name of the measured aspect, e.g.
- DateTime
- any date or time-related information, e.g.
HER2 negat, premenopauza, dg. 9/2020
- any date or time-related information, e.g.
- Negation
- words marking negation, e.g. something was not found, not proven, negative etc., e.g.
adjuvantní radioterapie není indikována
- words marking negation, e.g. something was not found, not proven, negative etc., e.g.
Attachments (3)
- ehr-image.png (86.8 KB) - added by 14 months ago.
-
typesystem.png (458.0 KB) - added by 14 months ago.
Apache cTAKES type system
- bratann.png (43.6 KB) - added by 13 months ago.
Download all attachments as: .zip