Detection of Central Sleep Apnea Based on a Single-Lead ECG

Abstract

Central sleep apnea (CSA) is a sleep-related disorder in which breathing is either diminished or absent, typically for 10 to 30 seconds, intermittently or in cycles. CSA is usually due to an instability in the body's feedback mechanisms that control respiration. Central sleep apnea can also be an indicator of Arnold-Chiari malformation. Therefore, various attempts have been made to produce a monitoring system for automatic Central sleep apnea scoring to reduce clinical efforts. This paper describes a system that can identify Central sleep apnea by means of a single-lead ECG and a Multilayer Perceptron network (MLP). Results show that a minute-by-minute classification accuracy of over 83% is achievable.