In long-term HRV analysis, it is common choice to study the difference signal IRR(i) = RR(i+1) - RR(i). In this work we first verified the fitting of a Lévy stable distribution on the signals IRR obtained from four databases, available on Physionet. They included normal subjects (N) but also individuals suffering from congestive heart failure (CHF) or showing ST segment changes (ST). The study showed that a L´evy stable distribution was generally more appropriate on the series than a Gaussian one (N: 1:70+/-0:19; CHF: 1:74+/-0:18; ST: 1:66+/-0:22). The differences between the populations were not significant (p > 5%). Based on the value of RMSSD on local short intervals, we built a simple Gaussian mixture density for each IRR series. Such mixture densities were able to properly describe the histograms in the databases under analysis. This explanation, which also avoids the necessity of invariant densities with not-finite second moments, might be closer to the physiological situation at hand.
Characterizing histograms of heartbeat interval differences with Gaussian mixture densities / R. Sassi - In: Computers in Cardiology, 2009 / [a cura di] A. Murray. - [s.l] : IEEE, 2009. - ISBN 9781424472819. - pp. 157-160 (( Intervento presentato al 36. convegno CinC tenutosi a Park City nel 2009.
Characterizing histograms of heartbeat interval differences with Gaussian mixture densities
R. SassiPrimo
2009
Abstract
In long-term HRV analysis, it is common choice to study the difference signal IRR(i) = RR(i+1) - RR(i). In this work we first verified the fitting of a Lévy stable distribution on the signals IRR obtained from four databases, available on Physionet. They included normal subjects (N) but also individuals suffering from congestive heart failure (CHF) or showing ST segment changes (ST). The study showed that a L´evy stable distribution was generally more appropriate on the series than a Gaussian one (N: 1:70+/-0:19; CHF: 1:74+/-0:18; ST: 1:66+/-0:22). The differences between the populations were not significant (p > 5%). Based on the value of RMSSD on local short intervals, we built a simple Gaussian mixture density for each IRR series. Such mixture densities were able to properly describe the histograms in the databases under analysis. This explanation, which also avoids the necessity of invariant densities with not-finite second moments, might be closer to the physiological situation at hand.File | Dimensione | Formato | |
---|---|---|---|
[22]_CinC_2009_ParkCity.pdf
accesso aperto
Tipologia:
Publisher's version/PDF
Dimensione
122.91 kB
Formato
Adobe PDF
|
122.91 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.