Dilated Convolutional Attention Network (DCAN), integrating dilated convolutions, residual connections, and label attention, for medical code assignment. It adopts dilated convolutions to capture complex medical patterns with a receptive field which increases exponentially with dilation size.
Download MIMIC-III dataset from physionet.
Organize your data using the following structure
data
| D_ICD_DIAGNOSES.csv
| D_ICD_PROCEDURES.csv
| ICD9_descriptions
└───mimic3/
| | NOTEEVENTS.csv
| | DIAGNOSES_ICD.csv
| | PROCEDURES_ICD.csv
| | *_hadm_ids.csv
ICD9_descriptions is avaiable in this repo, and
*_hadm_ids.csv are avaiable here.
MIMIC_RAW_DSUMS is available here, while the rest file for MIMIC2 can be generated with their code.
If you use Python3 consctruct_datasest.py in ICD9_Coding_of_Discharge_Summaries to create data files, remember to convert dict object to list (line 82&83) and use dict.items() instead of dict.iteritems().
Assign the directories of MIMIC data using MIMIC_3_DIR.
python3 main.py
Configs available at options.py.
Requirements:
- python 3.7
- pytorch 1.5.0
@inproceedings{ji2020dilated,
title={Dilated Convolutional Attention Network for Medical Code Assignment from Clinical Text},
author={Ji, Shaoxiong and Cambria, Erik and Marttinen, Pekka},
booktitle={3rd Clinical Natural Language Processing Workshop at EMNLP},
year={2020}
}