Application of support vector machines and Gaussian mixture models for the detection of obstructive sleep apnoea based on the RR series

  • Authors:

  • A. G. Ravelo;C. M. Travieso;F. D. Lorenzo;J. L. Navarro;S. Martín;J. B. Alonso;M. A. Ferrer

  • Affiliations:

  • Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain;Departamento de Señales y Comunicaciones, Universidad de Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain

  • Venue:
  • MATH'05 Proceedings of the 8th WSEAS International Conference on Applied Mathematics
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper we present the performances of two automatic statistical methods for the classification of the obstructive sleep apnoea syndrome based on the RR series obtained from the Electrocardiogram (ECG). We study the effect of working with Support Vector Machines (SVM) and compare its performance with a reference detector based on Gaussian Mixture Models (GMM). These classifications methods require two previous stages: preprocessing and feature extraction. Firstly, we apply a preprocessing over the ECG for estimating the R instants which is previous to feature extraction. Secondly, a power-ratio-based coefficient (PRC) and a Linear Frequency Cepstral Coefficients (LFCC) parameterization over the RR signal is applied to extract the relevant characteristics. We fix the set of features for both classification methods.