In this paper we apply several feature selection algorithms on features extracted from heartbeat timeseries recorded from young and elderly healthy subjects. We investi gate the results in an attempt to see which of those features are selected from the majority of the algorithms and start a discus sion on the results. All commonly used HRV analysis methods are investigated starting from simple statistical methods to non-linear ones. The results show that for the specific dataset most algorithms agree on the features they consider more significant.
(EN)
University of Ioannina
