Multiple Task Variational Autoencoder for survival analysis

This repos contains source code for paper:

Vo, Thanh-Hung, Guee-Sang Lee, Hyung-Jeong Yang, In-Jae Oh, Soo-Hyung Kim, and Sae-Ryung Kang. 2021. "Survival Prediction of Lung Cancer Using Small-Size Clinical Data with a Multiple Task Variational Autoencoder" Electronics 10, no. 12: 1396. https://doi.org/10.3390/electronics10121396

Data

• The form of data mostly come from CSV/XLSX
• Predefined "fold" column, e.g., randomly 1-5, or 0-4, default fold 1 is for the test (can customizable in configure file, test_fold)

Configure:

The important step to configure is in *.json, where most configuring data, model, etc., are placed. There are some important notes:

• meta_df: where XLSX data file
• cat_names, cont_names, and y_names should be defined
• model, loss_func: model and loss function definition
• process_pipeline_1: you may need custom cut points if y different

For *.sh, the bash shell to run experiments. The command-line to run is "python -m prlab.cli run" (in a bash script, *.sh), change to "python -m prlab.cli k_fold" if want to run k-fold cross-validation (k should be defined, k_start default is 0)

Run

• setup environment (virtualenv is recommend, "pip install --upgrade pip wheel setuptools" may be needed)
• pip install -r requirements.txt
• setup and login for wandb to save logs
• custom run.sh to the path of configuring and some other configure
• ./run.sh
• The report will be in models//reports.txt.

Source code is provided 'as-is' WITHOUT any WARRANTY or SUPPORT. Using this script is at YOUR OWN RISK.