Affiliation:
1. Bhilai Institute of Technology, India
2. Chhatrapatishivaji Institute of Technology, India
Abstract
Photoplethysmograph signal carries very useful cardiac information such heart rate, oxygen saturation level, blood pressure, and diabetic condition. Blood pressure is one such cardiac information that can be estimated by extracting features of PPG signal. Cuff-less blood pressure measurement using photoplethysmograph (PPG) signal is one of non-invasive methods. It allows continuous monitoring of blood pressure in simple, rapid, and low-cost mode. This chapter segregates PPG features and re-investigates their effectiveness in terms of BP measurement. Machine learning algorithm based on K-nearest neighbour is applied for classification of samples. MIMIC II multi-parameter database of ECG and finger PPG is applied on the KNN classifiers. Classification accuracy comes to 92%, and correlation between predicted and observed SBP and DSB are 0.89 and 0.85, respectively.