As part of the UNSW DataSoc Hackathon, we had to predict strokes from a dataset of medical records. They included health and personal records, as well as the treatments they are on.
Predictive_Modelling has some errors on the XGBoost at the bottom from when I tried to clean this up for GitHub on jupyter. I couldn't get LightGBM to work because I'm using 32bit python, but the code worked flawlessly on colab