Return to search

Classification analysis of ECG and study on the ratio of low frequency spectral powers to high frequency spectral powers of RR intervals

In the thesis, we study the data collected from VICU of Kaohsiung Veterans General Hospital.The data consists of the following two
kinds. The data of the first kind is ten minute ECG data with sampling frequency 500Hz. The data of the second kind is long term one minute average heart rate. There are two main topics
studies in this thesis. In the first part, we consider the standard deviations which be taked log and medians of RR intervals, PR intervals and RT intervals, respectively derived from first kind data. By considering the odds ratio¡Bsensitivity and specificity, and we select the heart classification vectors. The object is to classify among normal and ventricular heart disease (VSD), normal and ventricular heart
disease with congestive heart failure (VSD+CHF). We proposed a method of both variables and two-step classification procedure. Furthermore, we built the logistic regression models. In the second part, we derive the distribution of the spectrum power ratio of low frequency to high frequency. Shewhart control chart and Tabular cusum chart are constructed to monitor RG. Also, we discover the periodicity of RG. Finally,
there exists positive correlation between RA¡]or RG¡^ and autocorrelation of lag one of RR interval.

Identiferoai:union.ndltd.org:NSYSU/oai:NSYSU:etd-0627100-181353
Date27 June 2000
CreatorsChen, Chih-Yuan
ContributorsMong-Na Lo Huang, Meihui Guo, Kai-Sheng Hsieh
PublisherNSYSU
Source SetsNSYSU Electronic Thesis and Dissertation Archive
LanguageCholon
Detected LanguageEnglish
Typetext
Formatapplication/pdf
Sourcehttp://etd.lib.nsysu.edu.tw/ETD-db/ETD-search/view_etd?URN=etd-0627100-181353
Rightsoff_campus_withheld, Copyright information available at source archive

Page generated in 0.0018 seconds