ECG Sleep Apnea Detection

Keras implementation for ECG sleep apnea detection

Prerequisites

• Keras: the 'Merge' layer is deprecated in Keras versions newer than 2.0. This code maintains the 'Merge' layer, and therefore needs the older versions of Keras. To install Keras version 2.0 (it still runs the 'Merge' layer):

pip install keras==2.0

• ECG Sleep Apnea Dataset

ECG Sleep Apnea Dataset

• The data in the directory have been contributed by Dr. Thomas Penzel of Phillips-University, Marburg, Germany.
• 35 records (a01 through a20, b01 through b05, and c01 through c10)
• 7 hours to 10 hours of ECG signal, a set of apnea annotations, a set of machine-generated QRS annotations
• .dat files: ECG signal (16 bits per sample, Fs=100Hz)
• .apn files: binary annotation files containing an annotation for each minute of each recording the presence or absence of apnea
• .qrs files: machine generated binary annotation files, made using sqrs125

wget -r -np http://www.physionet.org/physiobank/database/apnea-ecg/

Getting Started

Pre-processing

• RR Interval: extracting the time intervals between consecutive heart beats
• QRS Amplitude: calculates the amplitude of R-peak
• Age and Sex

python pre_proc.py

Train

• Train a model:

python train.py