IDEARS - Integrated Disease Explanation and Associations Risk Scoring

Applies to the UKB datasetes, UKB dementia, AD and PD classification and SHAP

Overview

This is the codebase for IDEARs - Integrated Disease Explanation and Associations Risk Scoring. Its overall architecture is shown below.

How to Run

To ease the configuation, please install Anaconda and set this up in a virtual environment.

1. Install Anaconda:

https://www.anaconda.com/products/individual

2. Create the environment:

conda env create -f .\conda-env.yml

3. Acticate the environment:

conda activate conda-env

Then on Windows, run startlocal_woDocker.bat and on Linux, run startlocal_woDocker.sh

Codebase Structure

• data_gen.py is used to perform ETL on the data and to create the model datasets
• data_proc.py is used for extra data processing including the creation of normalised datasets
• ml.py is used to run the models including logistic regression, XGBoost and for model interpretability using SHAP
• analysis.py is used to create charts, perform extra statistical tests including paired t tests

The jupyter notebooks used for AD are:

• AD_ml_part_1.ipynb
• Master_ml.ipynb

Overview

Import modules etc.

Directory Tree and Explanations

This folder shows the implementation of the IDEARs platform.

Enquiries

Michael Allwright - [email protected]