Home > Publications > Doctoral dissertation & Master thesis
Domestic Conference
Title Methodology for Hypertension Diagnosis from PPG only using PCA for Preventive Healthcare
Degree MS
Author Byunghak Kong
Advisor Kiseon Kim
Graduation Date 2016.02.25 File
    Date 2017-03-06 11:02

Cardiovascular diseases (CVD) have seriously affected the leading causes of death in the world. Hypertension is a very fatal cause of CVD such as stroke, heart attack and kidney failure. Because of this, the daily monitoring of blood pressure is required in a continuous and non-invasive manner. In recent years, seminal studies on the estimation of blood pressure using photoplethysmography (PPG) have been reported. PPG enables us to monitor blood pressure condition in daily life and to prevent unexpected consequences. The related works have proposed many features of PPG signal to estimate blood pressure. However, there exists a limitation of extracting features from PPG signal of all people due to its inherent non-stationary characteristics.

In this thesis, we propose new features by applying the principal component analysis (PCA) technique to extract features. This approach is designed to overcome the non-stationary characteristic of PPG signal and to extract information to estimate blood pressure. Moreover, a novel methodology for hypertension diagnosis using PPG alone is introduced. This methodology is to classify the blood pressure condition of the subject, whether in danger or normal range, instead of exact blood pressure value. This is focused on preventive healthcare and tested based on the MIMIC database that is freely available to public by Physionet.

광주과학기술원 한·러 MT-IT 융합기술연구센터 광주과학기술원정보통신공학부