Title :
Continual pulmonary arterial wedge pressure estimated beat-to-beat by a neural network
Author :
Johnson, Royce W. ; Pellett, Andrew A. ; Morrison, G.G. ; Champagne, Michael S. ; DeBoisblanc, Bennett P. ; Levitzky, Michael G.
Author_Institution :
Dept. of Adv. Technol., Kinetics Concepts Inc., USA
fDate :
30 Oct-2 Nov 1997
Abstract :
Pulmonary arterial wedge pressure has been estimated beat-to-beat by an artificial neural network (ANN). Individual beats were parsed from pulmonary arterial pressure recordings obtained in 13 dogs just prior to measurements of conventional occlusive wedge pressure. The beats were resampled and used as inputs to a back-propagation neural network. The network was trained to estimate the wedge pressures obtained immediately after the beats were recorded. The training was done with 80% of all beats and tested on the remaining 20%. Testing on this 20% showed agreement statistics of bias and imprecision of 0.07±0.70 mmHg. The method clearly demonstrates that it is possible to estimate wedge pressure from individual beats but additional work is needed for practical application
Keywords :
blood pressure measurement; lung; medical signal processing; neural nets; agreement statistics; artificial neural network; backpropagation neural network; beat-to-beat estimation; bias; continual pulmonary arterial wedge pressure estimation; conventional occlusive wedge pressure; dogs; imprecision; Arteries; Artificial neural networks; Blood pressure; Cardiology; Dogs; Kinetic theory; Neural networks; Physiology; Pressure measurement; Testing;
Conference_Titel :
Engineering in Medicine and Biology Society, 1997. Proceedings of the 19th Annual International Conference of the IEEE
Conference_Location :
Chicago, IL
Print_ISBN :
0-7803-4262-3
DOI :
10.1109/IEMBS.1997.756536
