A linear regression model with dynamic pulse transit time features for noninvasive blood pressure prediction

Cardiovascular diseases (CVDs) have become the leading cause of death globally (as reported by the World Health Organization, June 2016). An effective method of preventing CVDs is to measure and monitor blood pressure (BP), which serves as a physiological indicator for cardiovascular systems. A previous research has proposed the use of pulse transit time (PTT) information to compute the BP measure. We propose herein a novel method based on a linear regression model that incorporates static and dynamic PTT features to predict BP measures more accurately. Moreover, the proposed model considers the estimated systolic blood pressure (SBP) when estimating the diastolic blood pressure (DBP). Experimental results first show that the proposed method can predict the BP more accurately than conventional methods, with notably higher correlation scores and lower mean square errors. These results confirm the effectiveness of incorporating dynamic PTT features for accurate BP predictions. Next, our experimental results demonstrate that the proposed method attains a better DBP prediction capability, showing that the estimated SBP provides useful information for the DBP prediction.