Cardiovascular Age and Central Blood Pressure assessment based on Peripheral Blood Pressure Morphology and Artificial Neural Networks

Cardiovascular health can be assessed from central blood pressure waveform and arterial stiffness, highly associated with Arterial Age (AA). However, the acquirement of these parameters is challenging. This paper proposes the estimation of Systolic Central Blood Pressure (SBPc) and classification of Chronological Age (CHA) by groups, as a substitute of AA, by means of non-invasive cuff-pressure Arterial Pulse Waveform (APW) acquisition and further pulse analysis, using regression analysis for SBPc and classification for CHA. A set of Artificial Neural Network (ANN) for each respective outcome was trained and validated with an in-silico database (n=4374) from a One-Dimensional (1-D) model. As a result, a Root Mean Squared Error (RMSE) of 0.39 mmHg and Mean Absolute Percentage Error (MAPE) of 0.61% was obtained, while an accuracy of 97.8% was achieved for classification. Following this, an in-vivo dataset (n=32) was used to evaluate the performance of both ANN obtaining an RMSE of 5.85mmHg and MAPE of 4.3%, while the accuracy decreased to 68.9%. The proposed methodology could have the potential to determine the AA of a subject using only a single peripheral APW. Furthermore, a populated in-vivo evaluation remains to be conducted.