iSLEEPS

This repository has been made to help searcher to have a quick start on analysing the iSLEEPS dataset available here in iHUB-data: iSLEEPS dataset.

Introduction 📝

1. Dataset: The dataset consists of polysomnography (PSG) patient recordings during sleep studies of ischemic stroke patients, capturing brainwave activity across different sleep stages. It includes multiple channels of raw PSG data, annotated with sleep stages and additional annotations such as sleep apnea events, providing a comprehensive resource for sleep research and machine learning model development.

2. Purpose: The purpose of the preprocessing code is to take the raw EDF data and corresponding annotation .xlsx files, match the 30-second epochs, and transform them into .npz format suitable for model training. Additionally, it provides baseline scripts to help users get started with training machine learning models on the processed data.

Repository Structure 📂

• preprocess: Script to preprocess raw data, matching 30-second epochs and converting them into .npz format for model training.
• explore.ipynb: Basic exploratory data analysis (EDA) script to help understand the dataset.
• models: Contains baseline model training scripts to get started with training deep learning models on the processed data.
• utils/: Directory containing utility functions for data loading and processing.
• requirements.txt: List of Python dependencies needed to run the scripts.

Code Structure ⚙️

├── README.md
├── requirements.txt
├── preprocess
│   ├── __init__.py
│   ├── channel_mapping.py
│   ├── config.py
│   ├── main.py
│   |── numpy_subjects.py
│   |── staging_preprocess.py
├── exploratory_data_analysis
│   |── 
│   ├── 
├── images
├── 
└── src

🚀 Getting Started

1. Clone the Repository

First, clone this repository to your local machine:

# Clone the repository
git clone https://github.com/suvadeepmaiti/iSLEEPS

# Navigate into the repository directory
cd iSLEEPS

2. Install Dependencies

# Install dependencies
pip install -r requirements.txt

3. Usage

Define the path where you downloaded the dataset in the config.py file.

raw_file_path = '/scratch/sleep/nimhans_raw_data/' #redefine this 

Define the path to the output directory where preprocessed data will be saved

output_data_path = '/scratch/sleep/nimhans_preprocessed_all/' #redefine this 

4. Preprocess the Data

python main.py # Preprocess the data

🖼️ Figures

Here are some of the figures used in this project:

Figure 1: Age Distribution

Figure 2: AHI Distribution

Figure 3: Subject-wise Sleep Stages Distribution

Figure 3: Annotated vs Predicted Hypnogram