Evaluation of Model Performance for Estimating Resting Blood Pressure Using Independent Components of Facial Thermal Images

Abstract

A significant number of patients suffer from hypertension. Blood pressure monitoring using a non-contact measurement technique is expected to detect and control this medical condition. The technique has been studied for non-contact blood pressure measurements using facial thermal images obtained by a thermography device. A model for resting blood pressure estimation using facial thermal images has been constructed in previous studies. The model performance has not been evaluated using unknown data for practical use. Additionally, other modeling methods, such as non-linear regression methods, need to be investigated for improving estimation accuracy. The objective of this study is the construction of the models for estimating resting blood pressure using several modeling methods and to evaluate the performance of models using unknown data. As a result, the skin temperature in the nasal region is suggested to be related to the resting blood pressure. The model could estimate resting blood pressure at a 5.73 mmHg root mean square error between the measured and estimated values.