Classification of HRV-signals using Time-Frequency Analysis

Abstract: Heart rate variability (HRV) is a term within cardiology describing the natu- rally occurring variation of the time interval between heartbeats, and high HRV activity has been linked to both cardiovascular and non-cardiovascular health. This bachelor thesis aims to find differences between HRV-signals where short term physical strain (pain) was induced and HRV-signals where no such strain was applied. This was done using time-frequency methods, primarily the spectro- gram, with most of the analysis performed in the so called high frequency band (0.12 to 0.4 Hz). The estimated respiratory frequency was also used as input to limit the analysis to relevant frequencies. Multiple methods examining spectral power in these frequencies was then designed and applied. The most successful methods could classify a participants signal pair correctly 81.4% (p<0.0001) of the time, and a individual signal 72.9% (p<0.0001) of the time, indicating that the induced strain had a statistically significant impact on the HRV-measurements.