Current Proceedings on Technology
Yazarlar: Necmettin SEZGİN, M. Emin TAĞLUK, Ramazan TEKİN
Konular:-
Anahtar Kelimeler:Sleep apnea syndrome,Snoring signal,Wavelet transform,Logistic regression.
Özet: Sleep Apnea Syndrome (SAS) is a serious health problem. Early recognition and decision on the treatment method is important. In this study a method concerning patient’s snoring signals for estimation of SAS is proposed. 1335 normal and 1463 abnormal snore episodes were employed and processed for their time-frequency (TF) energy profile through wavelet transform. The TF patterns of the signals were segregated into five sub bunds and seven parameters were derived from each sub band. The [episodes x parameters] data matrices were constructed and the Logistic Regression, a statistical model that explain the relationship between decision and condition features while the decision property is binary, is applied to these data matrices. Through this method SAS was 97.50% correctly estimated which is significantly higher than the performance of other strategies we have studied so far.