Model based Blood Pressure estimation during exercise test using modified fuzzy function

Abstract

Blood Pressure (BP) measurement during exercise test is of great importance. Due to the low accuracy of measuring BP using cuff and barometer during exercise and limitations in the continuous measurements because of the vessel crush, it will be of great advantage to obtain systolic and diastolic BP values in a cuff-less approach. This could be achieved by using extracted features and characteristics of ECG and PPG signals. BP is highly correlated with features such as PTT and HR. However, the correlation is not necessarily linear. It could be nonlinear, multimodal and vague. Therefore, the use of fuzzy function approach with the parameters used in physiological models as its inputs is proposed in this paper. Then, in order to improve the performance of fuzzy function to estimate BP, GK clustering method instead of the FCM and LS-SVM instead of LSE are used in order to produce the antecedent and consequent of the rules respectively. Comparing the results with the BP values which are estimated using NN, and fuzzy systems based on GD training and RLS, indicate better performance of modified fuzzy function with approximately zero mean error and less or almost equal to 8 mmHg as the value of STD in satisfying AAMI standard in systolic and diastolic BP estimation of all stages.

Cuffless Blood Pressure Estimation Algorithms for Continuous Health-Care Monitoring

“…However, such a calibration is reliable for only a short time interval [13]. Although calibration-based methods can not be reliably used as replacements of the conventional BP measurement devices, they are applicable for BP monitoring in short intervals such as exercise tests [14].…”