The development of a technique for measuring the sensitivity of baroreceptors is described. This technique exploits the principles of parametric identification using ARMAX models: the input–output behaviour (systolic pressure–heart rate) of the baroreceptor system is identified, and thus its sensitivity. The cost function for choosing the model is defined by generalising the classic sum of square residuals, i.e., introducing cross-correlation between input and residuals and auto-correlation of residuals. The resulting algorithm is compared with the ‘modulus’ technique (Robbe et al., 1987) by means of simulated data with added noise. In the case of modulus, the systematic error increases considerably with noise. This tendency towards overestimation is quantified to 118% (percentage error divided by ratio of noise to signal root-mean-square values), whereas that of the proposed technique is only 12.5%. Experimental data was used to study the correlation between the measurements obtained using the two techniques (Pearson’s correlation index was 0.81). This analysis confirmed the tendency to overestimation of modulus.
Measurement of human baroreceptor reflex sensitivity by means of parametric identification
ANGRILLI, ALESSANDRO;
1998
Abstract
The development of a technique for measuring the sensitivity of baroreceptors is described. This technique exploits the principles of parametric identification using ARMAX models: the input–output behaviour (systolic pressure–heart rate) of the baroreceptor system is identified, and thus its sensitivity. The cost function for choosing the model is defined by generalising the classic sum of square residuals, i.e., introducing cross-correlation between input and residuals and auto-correlation of residuals. The resulting algorithm is compared with the ‘modulus’ technique (Robbe et al., 1987) by means of simulated data with added noise. In the case of modulus, the systematic error increases considerably with noise. This tendency towards overestimation is quantified to 118% (percentage error divided by ratio of noise to signal root-mean-square values), whereas that of the proposed technique is only 12.5%. Experimental data was used to study the correlation between the measurements obtained using the two techniques (Pearson’s correlation index was 0.81). This analysis confirmed the tendency to overestimation of modulus.File | Dimensione | Formato | |
---|---|---|---|
DeCecco.pdf
accesso aperto
Tipologia:
Published (publisher's version)
Licenza:
Accesso gratuito
Dimensione
188.05 kB
Formato
Adobe PDF
|
188.05 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.