Key result
Combined PPG features cut systolic blood pressure estimation error by ~7 mmHg using SVR.
Why the study?
Traditional cuff-based blood pressure measurement is uncomfortable for continuous monitoring, motivating the development of a cuffless photoplethysmography estimator that incorporates second derivative features related to aortic compliance and stiffness.
Does a Support Vector Regression blood pressure estimator combining 14 second derivative PPG features and 21 time-scale PPG features improve blood pressure estimation accuracy compared to a conventional 21-feature neural network estimator in MIMIC II PPG pulse cycles?
Does a Support Vector Regression blood pressure estimator combining 14 second derivative PPG features and 21 time-scale PPG features improve blood pressure estimation accuracy compared to a conventional 21-feature neural network estimator in MIMIC II PPG pulse cycles?
Incorporating second derivative photoplethysmography (SDPPG) vascular aging features into a support vector regression model improves cuffless, noninvasive blood pressure estimation accuracy by up to 40% over conventional time-domain PPG neural network models.
Answers are grounded in this paper and cite their sources.
Supports cuffless BP estimation accuracy in retrospective data; leaves open prospective validation before clinical adoption.
In personal healthcare, blood pressure (BP) is an important vital sign to be monitored frequently.However, traditional BP measurement devices require cuff's inflation and deflation that is very uncomfortable for many users.Cuffless noninvasive BP estimation methods are very attractive especially on using Photoplethysmography (PPG) approach for achieving continuous BP monitoring and minimal user's inconvenience.From recent studies on the second derivative of PPG (SDPPG) for vascular aging, SDPPG contains the information about aortic compliance and stiffness, which is highly related to blood pressure.To making use of this new finding, 14 new SDPPG based features are proposed in this paper.They are combined with conventional 21 time-scale PPG features to develop a Support Vector Regression based BP estimator.Experimental results demonstrated that the combined features based BP estimator could improve accuracy of the conventional time-scale PPG based BP estimation by 40%.Index Terms-Blood pressure, photoplethysmography (PPG), second derivative wave, support vector regression. I. INTRODUCTIONBlood pressure (BP) is one of the four vital signs of human body besides heart rate, respiratory rate, and body temperature.Recently, the risk of having hypertension is extending from older to younger groups of the population.This trend making the number of people who need daily monitoring of BP is significantly increased.In addition, the demand for public resources in the hospital is becoming more and more serious and the type of healthcare is tending to individual-centered rather than hospital-centered.The need of self-monitoring BP devices dramatically increases in the recent decades.However, most of the commercial BP measuring devices require to inflate and deflate a cuff for determining the BP values.This measurement has a high demand for the medical practitioner"s operation skills.Self-monitoring BP devices using cuffless and noninvasive detective methods are very attractive for personal healthcare applications.It was reported that
No takes yet. Share an insight, caveat, or question.
Liu et al. (2017) studied Blood pressure monitoring (normal, hypertension, and hypotension) (n=910). Support Vector Regression (SVR) blood pressure estimator using 35 combined features (21 time-scale PPG features and 14 second derivative PPG features) vs. Conventional Neural Network (NN) blood pressure estimator using 21 time-scale PPG features was evaluated on Mean Absolute Error (MAE) of estimated systolic blood pressure (SBP, mmHg) (p=<0.01). Combining 14 second derivative PPG features with 21 time-scale features in an SVR estimator improved BP accuracy by up to 40%, reducing SBP MAE from 15.1 to 8.54 (p<0.01).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: