Entropy Estimation Methods in HRV Analysis of Patients with Myocardial Infarction

by    S. Lau, J. Haueisen, E. G. Schukat-Talamazzini, A. Voss, M. Goernig, U. Leder, H.-R. Figulla

Preprint series: 06-02, Reports on Computer Science

The paper is published: Conference of the European Study Group on Cardiovascular Oscillations 2006

CR: 62P10,92C55,94A17

Abstract: Heart rate variability (HRV) is a marker for autonomous activity in the heart.
A key application of HRV measures is the stratification of mortality risk
after myocardial infarction.
The information entropy is a promising measure of HRV.
Our hypothesis is that the information entropy of HRV, a non-linear approach,
is a suitable measure for this. As a first step,
we aimed at evaluating the effect of myocardial infarction on the entropy.

Our method was to compare the entropy to standard HRV parameters.
Essentially, one multivariate classification rule was generated based on
existing HRV measures and one based on existing and new entropy measures.
The gain in classification accuracy was then an evaluation criterion.
The classification rules were expressed as decision trees.
The simplicity and parameter choice of the augmented tree
was the second criterion. Additionally, five entropy estimation techniques
were compared in terms of estimation accuracy and discrimination strength.

A key finding is that the entropy is reduced in patients
with myocardial infarction with very high significance.
Additionally, a simple threshold of the meanNN-normalised entropy
outperforms the best multivariate standards-based infarct classifier by 5-10%.
The statistical and compression-based entropy estimations
are with a correlation of >94% highly consistent and thus reliable.
The entropy based on Burrows-Wheeler compression, implemented in Bzip2,
yields the best entropy estimation for infarct analysis purposes.

Keywords: Heart rate variability (HRV), myocardial infarction,entropy, entropy estimation, Burrows-Wheeler compression, Bzip2, Ngram, LZ77, Gzip, adaptive linear regression, classification, decision tree

Upload: 2006-01-11

Update: 2006-01-11

The author(s) agree, that this abstract may be stored as full text and distributed as such by abstracting services.