A Pulse Wave Based Blood Pressure Monitoring and Analysis Algorithm

Pulse wave reflects the heartbeat, pulse rhythm and the dynamic state of angiocarpy, but the feature extraction and auto-identification of hypertension based on pulse wave are difficult and crucial for monitoring and analyzing pulse signal. To address this key problem, an algorithm framework is proposed for auto-identification of hypertension, which named HHT-ELM analysis and recognition algorithm. The proposed algorithm framework includes preprocessing module, feature extraction module and signal recognition module. The first module adopts the low-pass filter and the morphological filter to denoise the original signal from the sensor. The Hilbert-Huang transform (HHT) is utilized to extract characteristics, and the empirical mode decomposition (EMD) is the main component of HHT, which is used to disassemble the pulse wave to acquire multiple intrinsic modal functions (IMF). Then the Hilbert transformation of each IMF yields the Hilbert spectra and the marginal spectra to extract the feature in the second module. Finally, the extreme learning machine (ELM) is employed to classify features, so the automatic recognition can be achieved in the last module. The study results show that the algorithm can diagnose hypertension with an accuracy rate of over 93%. Therefore, it provides a novel method of automatic pulse signal process and analysis for the clinical diagnosis and portable monitoring of hypertension.