Estimation of continuous blood pressure from pressure pulse wave morphology

Hypertension is the significant cause of death and many disabilities. Blood pressure is an index for knowing the cardiovascular status. More accurate continuous blood pressure can be obtained by arterial cannulation technique. Drawback with this method is that it causes potential risk to patient due to its invasive nature. Currently existing cuff based and oscillometric methods provide only short time data and it cannot be used for continuous monitoring. The purpose of this study is to estimate the continuous blood pressure using the pressure pulse wave morphology. Radial artery pressure pulse wave is used for the estimation. Initial blood pressure obtained with the help of regression model is added with the pressure difference between the systolic and diastolic feature points to evaluate the continuous blood pressure. About eleven informative features were extracted from pressure pulse wave and two regression models were constructed for comparing the performance. From mean absolute error calculated for the two models, multivariable linear regression model showed better accuracy than deep learning regression model. Initial blood pressure measured by multi-variable linear regression model is added with pressure difference to estimation continuous systolic and diastolic blood pressure. The results showed the reliability of pressure pulse wave morphology for non invasive blood pressure estimation.