Classification of non-stationary Heart Rate Variability using AR-model parameters

Sammanfattning: This thesis explores the connection between the heart rate variability and both stress and age. Two methods are used to classify the heart rate variability data, the autoregressive model and the Markov chain model. The autoregressive model is further expanded to become an autoregressive model with extraneous input using the respiratory signal as input signal. The Markov chain models are compared with their stationary distribu- tion using the Kolmogorov-Smirnov test. Autoregressive parameters are compared using condence intervals. The results indicate statistically sig- nicant deviations between age groups for both the autoregressive models and the Markov model. The stress related results were not as clear as the age related results, however some deviations were obtained for both models, indicating some stress related in uence on the HRV.