MuLT aims at employing unsupervised, self-supervised, and supervised learning algorithms to create different data representation in a try to improve classification performance.
This repository contains a broad set of experiments using The Multiple Myeloma Research Foundation (MMRF) CoMMpass data set.
Our experiments cover the following topics:
- Data cleaning and preparation;
- Modelling of FISH markers from gene expression levels;
- Compute of accuracy gain by adding genes to clinical predictors;
- Definition of Treatment Sensitivity (TS) outcome from treatment response;
- Modelling of TS from genes, clinical markers, and treatments;
- Structure analyses of genes and clinical markers; and
- Simulation of optimal treatments from TS predictors.
Note: All *.ipynb files correspond to the topics number above. To reproduce experiments just run *.ipynb files following numeric order.
lucasvenez at gmail dot com