Non-Invasive Blood Pressure Estimation using Poincaré Plot Indices of Photoplethysmography Signals

Abstract

Blood pressure as a valid vital biological signal plays an essential role in prognosis of cardiovascular diseases. Considering the importance of preventative strategies, non-invasive, continuous monitoring of blood pressure has special importance in personal health care. In this paper, we explored the possibility of dynamical behavior of PPG signal in blood pressure estimation. We use Poincaré’s section-based features of Photoplethysmography signal for blood pressure estimation that has the advantages of decreasing the need for various sensors to be attached to the human body and has the superiority of requiring minimal computation and calculation. The proposed method is developed and validated using a subset of Multi-parameter Intelligent Monitoring in Intensive Care (MIMIC) II database, containing arterial blood pressure and Photoplethysmography Signals. In this study, the extracted features from Poincaré map of the PPG signal are achieved the mean absolute error of 2.1 mmHg for systolic pressure and 1.4 mmHg for diastolic pressure prediction. The results meet the AAMI and BHS standards, which confirms the proficiency of Poincaré’s section-based indices in BP estimation.