REST

Elite Female Football Athletes Dataset and Analysis Tools

This repository contains various scripts and tools for analyzing a novel dataset of 21 elite female football athletes. The dataset comprises 17 days of actigraphy, well-being, caffeine consumption, screen time, and daily hand strength test data. The aim is to provide a comprehensive understanding of the interplay between lifestyle factors, sleep, and athletic performance.

Repository Structure

• algorithms: The folder contains several sleep detection algorithms and non wear algorithms. Additional, it provides sleep statistic functions and a base class to load in the actigraphy files.

• data_preprocessing: Script for preprocessing and loading in of the original gt3x files. Addditional, a list of visualisations.

• generate_reports.py: Example script how to read in actigraphy files and generate sleep statistics, sleep annotations and plots.

• technical_validation.py: Script for performing technical validation of data or models. The script reproduces the figures and results of the paper.

• annonymisation.py: Script for anonymizing sensitive data. It applies techniques to remove or obfuscate personally which we used for the annonymisation of the data.

• transfer_learning_inference.py: Script for performing inference using transfer learning models. It applies a LSTM model which has been trained on the MESA sleep study to REST for sleep prediction.

Installation

1. Clone the repository:

   git clone https://github.com/simula/REST.git
   cd REST

2. Create and activate a virtual environment:

   conda env create -f environment.yml

Usage