Date of Award
Master of Science in Biomedical Engineering - (M.S.)
Stanley S. Reisman
Ronald H. Rockland
David S. Kristol
The following study was conducted to design and implement computer programs to derive five time domain and two frequency domain Heart Rate Variability (HRV) measures from 24 hour Holter monitor recordings and to generate activity plots that display the variation of three of the five time domain measures and two frequency domain measures over a period of 24 hours.
Working with known pre and post experimental ECG data taken from a cyclical exercise program, standard parameters in both the time and frequency domains were derived using programs designed in Labview 7.0 engineering software package. Five parameters in the time domain namely SDNN, SDNN index, SDANN, rMSSD, pNN5O, and two in the frequency domain, namely the low and high frequency band areas. The designed programs were then tested on sample data files. Results of these tests were then validated by verifying them with results calculated on statistical tools available on Matlab and MS-Excel and also against standard ranges of values for the implemented measures. The calculated results were found to lie well within the standardized ranges obtained from literature.  A time domain analysis was then performed on a data set that consisted of 22 files of 24 hour Holter monitor recordings. These 22 files were part of a cyclical exercise study in which 11 healthy women with ages ranging from 32 to 58 participated as subjects. Eleven files were recorded prior to the cyclical exercise program and another 11 files were recorded upon completion of the program. Each recording contained the beat to beat intervals of a normal day in the subject's life. The generated results were then tested for significance using standard t-tests and Analysis of Variance (ANOVA).
In summary, no significant changes among pre and post experimental HRV measure values were revealed for parameters in the time and frequency domain. But the designed programs were found to reiterate a previously performed analyses in the frequency domain on the same data set, i.e., no significant changes. The implemented programs were then modified to plot 24 hour changes with pre-selected time and frequency domain parameters by dividing the 24 hour recording into 5 minute segments and generating the HRV measures from each interval. By performing such a calculation an array of HRV measures was formed. This array contained the HRV measure for each 5 minute interval over course of 24 hours. The array was then plotted using Labview 7.0 or Matlab and the peculiarities of the plots were then compared. Trends in change between the two domains that were studied were noted and suggestions made on how to derive a better understanding of the plots. The modified programs were designed with an objective to study the changes in the time domain parameters during a 24 hour period. Due to time constraints, the study of the time plot could not be completed, but the programs to generate 24 hour changes in 3 time domain and 2 frequency domain parameters and their corresponding plots were derived successfully. The programs now await to further testing and validation. Once all applications complete testing and validation, they will comprise a valuable "toolbox" for Heart Rate Variability Analysis.
Mathews, Binoy John, "Design and implementation of heart rate variability measures in the time domain" (2004). Theses. 556.