The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Impact of missed beats on heart rate variability using time-frequency analysis
Author:	Hailu, Kilbrom T.
View Online:	njit-etd2006-005 (xiii, 124 pages ~ 12.0 MB pdf)
Department:	Department of Biomedical Engineering
Degree:	Master of Science
Program:	Biomedical Engineering
Document Type:	Thesis
Advisory Committee:	Rockland, Ronald H. (Committee chair) Reisman, Stanley S. (Committee member) Schesser, Joel (Committee member)
Date:	2006-01
Keywords:	Heart rate variability Time-frequency analysis
Availability:	Unrestricted
Abstract:	ECG recordings, especially during an exercise stress test, are exposed to artifacts. The source of artifacts can be due to technical problems, biological events and errors in the automatic detection. As a result missed beats, extra beats and ectopic beats can occur. In this study analysis of missed beats is performed using time-frequency analysis. Some of the patient's data had missed beats. Fourier spectral analysis was first applied to the missed beats during the resting state. The impact of missed beats using Fourier spectral analysis is significant, especially, when the missed beats are in a row. The Fourier spectral analysis does not appropriately represent heart rate variability (HRV) during and after exercise because it assumes a signal to be stationary. In addition, Fourier analysis provides information of the heart rate only in the frequency domain. Therefore, there was a need to apply time-frequency analysis to examine the effect of missed beats. For the purpose of analysis, beats were eliminated in a row and arbitrary. Then time-frequency analysis was applied to the missed beats and the original data by superimposing on each other. Their differences were plotted as error functions. The sizes of the error functions were measured. The results show that the errors due to missed beats do not widely spread throughout the spectrum. That is, missed beats do not affect the entire time-frequency signals. Their effect is local. The errors due to missed beats are reduced when using interpolation.