A Novel Wearable Multi-Channel Pulse Monitor Combined with Machine Learning for Screening Hypertension Population

Hypertension is a chronic disease that seriously threatens the health of residents. However, the existing studies have difficulty meeting the needs of daily monitoring and screening hypertension. Continuous monitoring of pulse wave provides a novel perspective for screening hypertension groups. In this work, we proposed a novel wearable multi-channel pulse monitor, which can not only simultaneously collect pulse waves from three positions of the human wrist, but also distinguish between hypertension groups and healthy groups. The pulse wave data collected by the monitor undergoes feature extraction in the time domain and preprocessing. Based on mutual information score, we found out and selected top 12 differentiating features between the hypertensive group and the healthy group, and proposed a classification model based on support vector machine, whose accuracy is 96.28% and the F1 score is 0.9744. The experimental results demonstrate the application prospect of pulse wave in hypertension screening and clinical diagnosis, and the monitor has significant research prospects and commercial value.