A New Blood Pressure Estimation Approach Using PPG Sensors: Subject Specific Evaluation over a Long-term Period

In this paper, a new approach for predicting the blood pressure (BP) from the photoplethysmogram (PPG) signal is proposed related with a new original public dataset. The originality of the dataset is based on the fact that subjects are periodically monitored over weeks, while public datasets consider short acquisition periods. The proposed BP estimation approach uses key frequencies in the spectrum of the PPG signal isolated using the LASSO algorithm, then a predictive model is constructed as a patient-specific BP estimation model. The efficiency of the proposed methodology is evaluated on experimental data recorded over a long time period. Moreover, an evaluation of the various temporal markers of the PPG signal that have been proposed in the literature is conducted on the same data set. It is showed that only few of these temporal markers are useful for the estimation of the systolic and diastolic blood pressures. The results highlight that better blood pressure estimations are obtained when using the spectrum of the PPG signal rather than optimally selected temporal markers.