Abstract This study compares a neural network-based autoregressive exogenous (NNARX) model with a linear autoregressive exogenous (ARX) model in reconstructing central aortic pulse curve from peripheral arterial pulse. Invasive aortic and radial tonometry pressures were recorded in 20 patients in rest condition. A set of 10 patients (learning) was used to estimate the model parameters, the remaining 10 patients (test set) were used for validation. The estimated waveform of aortic pressure obtained by NNARX results more accurate than that estimated by linear ARX model providing a more fine reconstruction of dicrotic notch and systolic flex. Comparison of Augmentation Index measurement computed from NNARX and ARX reconstructed pressure signals with the reference value derived from invasive aortic waveform showed an improvement in accuracy of the NNARX measure.
Non-invasive estimation of central aortic pressure from radial artery tonometry by neural networks
PALOMBO, CARLO
2003-01-01
Abstract
Abstract This study compares a neural network-based autoregressive exogenous (NNARX) model with a linear autoregressive exogenous (ARX) model in reconstructing central aortic pulse curve from peripheral arterial pulse. Invasive aortic and radial tonometry pressures were recorded in 20 patients in rest condition. A set of 10 patients (learning) was used to estimate the model parameters, the remaining 10 patients (test set) were used for validation. The estimated waveform of aortic pressure obtained by NNARX results more accurate than that estimated by linear ARX model providing a more fine reconstruction of dicrotic notch and systolic flex. Comparison of Augmentation Index measurement computed from NNARX and ARX reconstructed pressure signals with the reference value derived from invasive aortic waveform showed an improvement in accuracy of the NNARX measure.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
