This repo showcases two generative adversarial net (GAN) approaches to precision oncology. In each folder the scripts and notebooks to train a model and to apply a trained model are provided.
To install the necessary libraries run:
pip install -r requirements.txt
With a dataset of co-occurring gene pairs, train a GAN to learn the pair distribution in order to generate and discriminate co-occurring gene pairs.
With two dataset consisting of a patients disease, demographics and genetic information and the corresponding treatments train a conditional GAN to produce suitable treatment suggestions based on the patient information.
This work is based on the medGAN approach introduced in the following paper:
Generating Multi-label Discrete Patient Records using Generative Adversarial Networks
Edward Choi, Siddharth Biswal, Bradley Malin, Jon Duke, Walter F. Stewart, Jimeng Sun
Machine Learning for Healthcare (MLHC) 2017