Abstract
In this study, we focused on blood pressure estimation with less-invasive using photoplethysmography. Blood pressure estimation in previous researches were realized from a change of pulse waveform especially by autonomic nerve. However, some researches are suggested that the change is different that of after mealtime. Because this difference significantly influences the estimation accuracy of blood pressure, we need to identify whether the change of after mealtime or the change of autonomic nerve. Therefore we tried to detect mealtime using features of pulse waveform. First, we conducted experiment to measure photoplethysmography of 5 subjects hourly for 5 days. Next, we extracted features from pulse wave, and calculated variation every one hour by the finite difference between each measurement. As a result, there were some features where pulse waveforms change significantly due to mealtime. It was possible to separate “eating” or “not eating” in time between the measurements by Adaboost with high accuracy based on those features. The result suggested that detecting mealtime is possible using photoplethysmography.
