Title: Change-points in heart rate variability: application in critical care patients
Authors: Ana Paula Rocha - Univ Porto - Fac Ciencias and CMUP (Portugal) [presenting]
Margarida Pereira - Univ Porto - Fac Ciencias (Portugal)
Abstract: The detection of change points in time series analysis and signal processing is a relevant problem in various areas such as economy, finance and biomedical applications. Heart Rate Variability (HRV) time series are complex, namely display non-stationary characteristics, exhibit long range dependence in the mean and conditional heteroscedasticity. Previous results suggest that structural breaks detection can be useful in HRV monitoring. Herein are reviewed several methods for segmenting nonstationary HRV time series, considering the case of multiple change point detection. Some approaches are based on the detection and estimation of abrupt changes in mean, root-mean-square level, standard deviation and slope; alternatively, can be used minimum description length principles and time varying spectral tools. The methods are applied in some typical clinical situations from intensive care patients, where segmentation and modelling of HRV may be used as an auxiliary tool in the assessment of the patients evolution and its management.