psyBERTpt

psyBERTpt is one of the few language models for clinical NER available in Portuguese, such as BioBERTpt PUC models.
psyBERTpt is a NER model built from the fine tuning of BERT-portuguese-large-cased.
psyBERTpt used, in the fine tuning stage, a corpus composed of 300 medical records of admission to a specialized psychiatric hospital, which was annotated by 5 doctors and nurses specialized in psychiatry in the following NER categories:

Self-Destructive Behavior
Diagnosis
Drug
Pharmaceutical
Psychic Function
Family History
Patient History
Observation
Symptom and Psychological Complaint

psyBERTpt achieved a precision of 0.65, recall of 0.69 and f1-score of 0.67.

The scientific article that presents the psyBERTpt is waiting to be published =)

My thanks to everyone who was part of this project:
Ms. Luiz Henrique Pereira Niero (me) ([email protected])
Prof. Dr. Ivan Rizzo Guilherme ([email protected])
Prof. Dr. Lucas Emanuel Silva e Oliveira ([email protected])
Prof. Dr. Gerardo Maria de Araújo Filho ([email protected])
Dr. Gabriel
Enf. Giovanna Favareto
Dra. Leisa
Dra. Loíse
Hospital Dr. Adolfo Bezerra de Menezes

Cannot merge
This branch has merge conflicts in the following files:
  • README.md

Sign up or log in to comment