Deep Cortical Arousal Detection (DeepCAD)

Introduction

A cortical arousal detection algorithm that uses a single-lead ECG to detect arousal during sleep.

Description

This repo contains training and inference code for DeepCAD in PyTorch. Training is done on Multi-Ethnic Study of Atherosclerosis (MESA) dataset: https://sleepdata.org/datasets/mesa

Requirements

Python 3.5 or later

• torch >= 1.0.0

Training

python train.py project_path dataset_path model_folder n_epochs n_batch lr weight_decay

Inference

python detect.py detected_arousal deepcad.pth.tar model example_psg.hdf 0.40

Example: