Sex-stratified metabolic signatures of adiposity indices and their associations with clinical biomarkers in the UK Biobank
📦 This repository contains code associated with the:
- Machine Learning (ML) development
- Parallelization across CPU cores to reduce runtime
In this research I identified Metabolic Signatures of eight adiposity indices using a multi-step estimation algorithm built upon adaptive elastic-net regularization (doi) with stability selection.
This ML workflow reduces potential false positives and improves the stability of feature selection, while maintaining the estimation accuracy.
🧠 This research advances the understanding of the pathophysiology of regional and overall obesity and provide insights into the molecular mechanisms through which adiposity impacts health, offering potential directions for future pharmacological or lifestyle interventions.
This research has been conducted using the UK Biobank Resource under specific Application Number.
Access to the UK Biobank is available through: http://www.ukbiobank.ac.uk
The script includes simulated data to demonstrate and reproduce the full machine learning workflow.
- Before running the ML script, run the requirements script to install all necessary libraries.
- R version: 4.4.1 (2024-06-14 ucrt)
📄 More details about the research can be found on its published form, at eBioMedicine
© 2025 Christos Papagiannopoulos — Licensed under GPL-3.0 — Attribution required.