dartrevan/CLEF-2017-train-scripts

This code uses for training recurrent neural network models for normalizing free-text descriptions of causes of death into icd-10 codes. Code written in python 2.7

Installing required libraries:

pip install -r requirements.txt

After installing libraries nltk punkt model must be downloaded in python interpreter: import nltk nltk.download('punkt')

Usage:

python predict.py --data path/to/training/data \
                  --embeddings path/to/word2vec/binary
                  --embeddings_dim embedding_vector_size
                  --encoder_size encoder_rnn_size
                  --decoder_size decoder_rnn_size
                  --use_sim indicates that similarities vector must be used
                  --dropout_rate drate to use after on similarities vector

Citing:

KFU at CLEF eHealth 2017 Task 1: ICD-10 Coding of English Death Certificates with Recurrent Neural Networks Z Miftakhutdinov, E Tutubalina - 2017

http://ceur-ws.org/Vol-1866/paper_64.pdf

BibTex:

@inproceedings{
    miftakhutdinov2017kfu,
    title={KFU at CLEF eHealth 2017 Task 1: ICD-10 Coding of English Death Certificates with Recurrent Neural Networks},
    author={Miftakhutdinov, Z and Tutubalina, Elena},
    year={2017},
    organization={CLEF}
}