JMIR Publications

ABSTRACT

Background:

Radial artery pulse wave is a widely used physiological signal for disease diagnosis and personal health monitoring as it contains a lot of important information concerning the heart and blood vessels. Periodic radial artery pulse signal is needed to be decomposed into single pulse wave periods (segments) for physiological parameter evaluations. It is inevitable to get abnormal periods in this process because of the quality of pulse wave signals, external interference and imperfections of segmentation methods.

Objective:

The objective of this paper is to develop a machine learning model to detect abnormal pulse periods from real clinical data.

Methods:

Various machine learning models such as k-Nearest Neighbor (KNN), logistic regression (LR) and support vector machines (SVM) are applied to classify the normal and abnormal periods in 8561 segments extracted from radical pulse wave of 390 outpatients. The recursive feature elimination (RFE) method is used to simplify the classifier.

Results:

It was found that a logistic regression model with only 4 input features can achieve a satisfactory result. The area under curve (AUC) of receiver operating characteristic (ROC) curve from the test set is 0.9920. In addition, these classifiers can be easily interpreted.

Conclusions:

It is expected that the model can be used in smart sport watch/band applications to accurately evaluate human health status.