Prevalent Approach of Learning Based Cuffless Blood Pressure Measurement System for Continuous Health-care Monitoring

Abstract

For monitoring and controlling hypertension continuous measurement of blood pressure is very much needed, which can be feasible by recent technological advances of wearable devices which replaces traditional methods of blood pressure measurement. Continuous monitoring can provide precious data of individual health conditions. This work focuses on comparative study of previous methods with learning based approach for continuous blood pressure measurement system based on only photoplethysmograph (PPG). We have described the conventional methods of blood pressure measurement with their limitations; learning based feature exploration methods for blood pressure. And we concluded with result and few suggestions as a future work for estimating cuffless BP continuously. This work proposes a novel learning approach with 11 features estimated from 50 patients data collected from MIMIC-II database. The proposed model perceive stable and accurate measurement of blood pressure. Results showed that mean absolute error for systolic blood pressure (SBP) and diastolic blood pressure (DBP) are 12.62, 11.86, 12.70 and 3.78, 3.36, 3.57 in linear regression, support vector machine (SVM) and Gaussian regression respectively and it also shows 6.17, 6.21, 6.08 for mean arterial pressure (MAP).