Evaluation and development of methods for time-frequency analysis of heart rate variability

Abstract: Non-stationary signals are very common in nature, consider for example speech, music or heart rate. Using the concept of time-frequency analysis this thesis studies the performance of different time-frequency distributions of both simulated and real non-stationary signals. The signals studied are linear and non-linear frequency modulated (FM) signals. Two methods are studied to increase performance of the signals' time-frequency distributions. Since lag-independent kernels perform well with slow varying frequency modulated signals both methods use these. One method uses filtering with compact support lag-independent kernels and the other uses a penalty function with multitapers corresponding to lag-independent kernels. These methods are then evaluated using two performance measures and the results are used to improve the time-frequency distributions of heart rate variability signals. The thesis suggests that both of these methods improve the time-frequency distribution of such signals.