Abstract
We highlight some of the pitfalls involved in the analysis of electrocardiogram data (ECG) when the correlation dimension (CD) estimates are used to discriminate between various states of the human heart or to distinguish healthy dynamics from different types of arrhythmia. It is argued that due to the PQRST complex, high autocorrelation and/or short data sets, the CD estimates would be distorted when continuously sampled ECG voltage data are used instead of beat-to-beat (RR) interval series.
Original language | English |
---|---|
Pages (from-to) | 359-369 |
Journal | Nonlinear Dynamics, Psychology, and Life Sciences |
Volume | 12 |
Issue number | 4 |
State | Published - 2008 |