Detecting vascular age using the analysis of peripheral pulse

Vascular ageing is known to be accompanied by arterial stiffening and vascular endothelial dysfunction, and represents an independent factor contributing to the development of cardiovascular disease. The microvascular pulse is affected by the biomechanical alterations of the circulatory system, and has been the focus of studies aiming at the development of noninvasive methods able to extract physiologically relevant features. Objective: proposing an approach for the assessment of vascular ageing based on a support vector machine (SVM) learning from features of the pulse contour. Methods: the supervised classifier was trained and validated over 20935 models of pulse wave, obtained with a multi-Gaussian decomposition algorithm, applied to laser Doppler flowmetry signals of 54 healthy, non-smoker subjects. Results: the multi-Gaussian model showed a mean R2 of 0.98 and an average normalized root mean square error of 0.90, demonstrating the ability to reconstruct the pulse shape. Over 30 training and validation experiments, the SVM showed a mean Pearson’s r of 0.808 between the rate of waves classified as old and the age of the subjects, along with an average area under the ROC curve of 0.953. Conclusion: the SVM showed the capability to discriminate differently aged individuals. Significance: the proposed method might detect the ageing-related modifications of the vascular tree; furthermore, since diabetes promotes vascular alterations comparable to ageing, this approach may be also suitable for the screening of diabetic angiopathy