Analysis of aging effects on the arterial pulse contour using an artificial neural network

Age correlates both with the prevalence of cardiovascular diseases and with arterial pulse contour changes in healthy adults. If the effect of normal age-related changes can be better elucidated many of the increasingly popular methods of noninvasive pulse contour analysis could be made more reliable. In this paper, the predictability of age in 200 healthy adults is assessed using a radial basis function (RBF) network. The inputs are the complex coefficients of the Fourier transform of a single normalized pulse together with beat length, pulse and mean pressures, the harmonic at peak energy, and gender. Age is the output. Age predictability using different combinations of these components reveals that the first 27 harmonics contain aging information. Use of pulse pressure and harmonic of peak energy improved prediction, while mean pressure, beat length, and gender did not. These results suggest that age-related pulse contour changes in healthy adults are due to pulse pressure rather than absolute pressure, and energy shifts in the frequency distribution rather than changes in heart rate.

[1]  S. Daniels,et al.  Age-related changes in total arterial capacitance from birth to maturity in a normotensive population. , 1997, Hypertension.

[2]  S M Finkelstein,et al.  Vascular abnormalities associated with long-term cigarette smoking identified by arterial waveform analysis. , 1997, The American journal of medicine.

[3]  J A Reggia,et al.  Feature discovery and classification of Doppler umbilical artery blood flow velocity waveforms , 1996, Comput. Biol. Medicine.

[4]  S. Finkelstein,et al.  Effects of aging per se on arterial stiffness: systemic and regional compliance in beagles. , 1996, American heart journal.

[5]  P. Snow,et al.  Artificial neural networks: current status in cardiovascular medicine. , 1996, Journal of the American College of Cardiology.

[6]  A Murray,et al.  Comparison of three arterial pulse waveform classification techniques. , 1996, Journal of medical engineering & technology.

[7]  R. Miranda,et al.  Circular Nodes in Neural Networks , 1996, Neural Computation.

[8]  J. Cohn,et al.  Noninvasive pulse wave analysis for the early detection of vascular disease. , 1995, Hypertension.

[9]  G. D. Johnston,et al.  Fish oil improves arterial compliance in non-insulin-dependent diabetes mellitus. , 1994, Arteriosclerosis and thrombosis : a journal of vascular biology.

[10]  J. Cohn,et al.  Vascular abnormalities in non-insulin-dependent diabetes mellitus identified by arterial waveform analysis. , 1993, The American journal of medicine.

[11]  J R Jansen,et al.  Computation of aortic flow from pressure in humans using a nonlinear, three-element model. , 1993, Journal of applied physiology.

[12]  T Pike,et al.  Automated recognition of corrupted arterial waveforms using neural network techniques. , 1992, Computers in biology and medicine.

[13]  J. Carroll,et al.  Determination of pulse wave velocities with computerized algorithms. , 1991, American heart journal.

[14]  S M Finkelstein,et al.  Reduced vascular compliance as a marker for essential hypertension. , 1991, American journal of hypertension.

[15]  K P Brin,et al.  Estimation of total arterial compliance: an improved method and evaluation of current methods. , 1986, The American journal of physiology.

[16]  N. Westerhof,et al.  Aortic Input Impedance in Normal Man: Relationship to Pressure Wave Forms , 1980, Circulation.