Transcript Piattaforma per l’integrazione del follow

Introduction
The aim of this work is investigating the differences of Heart Rate Variability
(HRV) features between normal subjects and patients suffering from Congestive
Heart Failure (CHF) at several levels of NYHA scale.
CHF is …
Introduction
Only a few studies have been focused on using HRV measures for diagnosis
purpose in CHF and these studies proposed binary classification for identify
normal and CHF patient without considering NYHA class.
One study* investigate the discrimination power of long term HRV measures;
the other** proposed a classifier based on short term HRV measures but does not
provide any information about NYHA class.
Consequently, we investigate the differences of Heart Rate Variability (HRV)
features between normal subjects and patients suffering from Congestive Heart
Failure (CHF) at several levels of NYHA scale.
*M.H. Asyali, Discrimination power of long-term heart rate variability measures, in: Proceedings of
the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society,
Cancun, September 17–21, 2003.
**Y. Isler, and M. Kuntalp, “Combining classical HRV indices with wavelet entropy measures
improves to performance in diagnosing congestive heart failure,” Computers in Biology and
Medicine, vol. 37, no. 10, pp. 1502-1510, Oct, 2007.
Methods
We performed a retrospective analysis of two public RR interval databases, to
compare values for HRV measure in normal middle-aged subjects and in patients
who suffered from chronic heart failure, with NYHA I, II and III.
We calculated statistical measures from 5-minute RR interval data using standard
methods. Moreover, we estimated short-term frequency domain measures using
the Lomb periodogram.
Measure
SDNN
Description
Unit
Standard deviation of all NN intervals.
ms
RMSSD
The square root of the mean of the sum of the squares of differences
between adjacent NN intervals
ms
AVNN
Average of all NN intervals
ms
pNN50
Percentage of differences between adjacent NN intervals that are > 50
ms
This is one member of the larger pNNx family
%
TOTAL
POWER
Total spectral power of all NN intervals up to 0.04 Hz.
ms2
Total spectral power of all NN intervals between 0 and 0.04 Hz
ms2
LF
Total spectral power of all NN intervals between 0.04 and 0.15 Hz
ms2
HF
Total spectral power of all NN intervals between 0.15 and 0.4 Hz
ms2
VLF
LF/HF
Ratio of low to high frequency power
Methods
For each selected feature and for each NYHA scale, histogram distribution has
been computed.
Finally, we analyzed how the correlation matrix between features of NN series
changes according to the severity of CHF.
Results
We show the mean and standard deviation of each measure according to NYHA
class.
VALUE OF NN MEASURES
Measure
Unit
Normal
NYHAI NYHAII
(mean±SD) (mean±SD) (mean±SD)
SDNN
RMSSD
AVNN
pNN50
TOT. POW.
VLF
LF
HF
LF/HF
ms
ms
ms
%
ms2
ms2
ms2
ms2
-
47.4±24.8 50.4±29.6 24.3±17.6
24.8±14.6 25.5±14.2 14.5±8.0
6.2±10.1 7.5±12.3
1.6±4.9
803±156 797±157
639±87
3089±4474 4160±6740 917±1719
2130±3721 2810±5179 670±1290
671±965 806±1301 158±355
287±494 544±770
89±214
3.7±2.9
2.1±1.6
2.2±2.0
NYHAIII
(mean±SD)
25.8±16.9
17.0±8.1
2.3±4.6
701±103
968±1599
716±1360
149±244
103±139
1.5±1.4
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
We show the histogram distribution of each measure according to NYHA class.
Results
Correlation between NN Measures in Normal Subjects
We show the correlation matrix between HRV measures
CORRELATION BETWEEN NN MEASURES IN NORMAL SUBJECTS
AVN
SDNN RMSSD pNN50 VLF
LF
N
AVNN
1
0.30
0.50
0.47
0.16 0.35
SDNN
1
0.51
0.48
0.84 0.61
RMSSD
1
0.96
pNN50
1
0.23
0.22
VLF
1
LF
HF
CORRELATION BETWEEN NN MEASURES IN NYHA I
0.40
0.53
AVNN
0.85
SDNN
0.83
RMSSD
0.45
0.32
pNN50
1
0.61
VLF
0.53
0.51
HF
1
AVNN
SDNN
RMSSD
LF
HF
1
0.41
0.75
pNN50 VLF
0.67
0.27
0.39
0.55
1
0.63
0.51
0.88
0.78
0.79
1
0.95
0.42
0.65
0.82
1
0.32
0.55
0.74
1
0.70
0.65
1
0.86
LF
HF
CORRELATION BETWEEN NN MEASURES IN NYHA II
AVNN
AVNN
SDNN
RMSSD
pNN50
VLF
LF
HF
1
SDNN RMSSD
pNN50
VLF
CORRELATION BETWEEN NN MEASURES IN NYHA III
LF
HF
AVNN
0.46
0.45
0.42
0.34
0.34
0.42
AVNN
1
0.73
0.62
0.88
0.77
0.69
SDNN
1
1
0.89
0.57
0.74
0.86
RMSSD
1
0.52
0.74
0.93
pNN50
1
0.70
0.63
VLF
0.84
LF
1
HF
1
1
SDNN RMSSD
pNN50
VLF
LF
HF
0,43
0,38
0,24
0,31
0,33 0,34
1
0,59
0,44
0,88
0,76 0,64
1
0,89
0,33
0,53 0,78
1
0,24
0,42 0,73
1
0,62 0,44
1
0,72
1
Discussion
SDNN, RMSSD and TOTAL POWER has higher values in healthy subjects than
in CHF patients.
Moreover, VLF, LF and HF seem to be depressed in CHF patient.
The other parameters, such as AVNN, did not appear to be different between
normal and CHF.
The correlations between some features of NN series increase according to the
severity of CHF.
From the shown results it is possible conclude that there is a variation in HRV
features, according to NYHA classification. We suppose that this variation may
drive the research of a hierarchic classifier in order to distinguish not only
normal versus CHF patient but also mild versus severe CHF.