You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
shaoxiong 234061a438 initial commit 2 weeks ago
layers initial commit 2 weeks ago
utils initial commit 2 weeks ago
README.md initial commit 2 weeks ago
evaluate.py initial commit 2 weeks ago
main.py initial commit 2 weeks ago
models.py initial commit 2 weeks ago
requirements.txt initial commit 2 weeks ago

README.md

Clinical Outcome Prediction

Data

Request the MIMIC dataset via PhysioNet

Preprocess

Follow the clinicalBERT to extract the datasets. File system expected:

-data
  -discharge
    -train.csv
    -val.csv
    -test.csv
  -early
    -train.csv
    -val.csv
    -test.csv

Data file is expected to have column "TEXT", "ID" and "Label" (Note chunks, Admission ID, Label of readmission).

Labels: ADMISSION, ADMISSION_TYPE, DIAGNOSIS, LOS

Run

python main.py \
    --loader_workers 4 \
    --data_dir ./data/discharge \
    --out_dir ./log \
    --dataset discharge \
    --task ADMISSION,DIAGNOSIS,LOS \
    --encoder_model ./log/models/pretraining \
    --batch_size 32 \
    --accumulate_grad_batches 4 \
    --encoder_learning_rate 2e-5 \
    --learning_rate 2e-5 \
    --nr_frozen_epochs 2 \
    --monitor val_loss \
    --metric_mode min