Validity Test of Wearable Continuous Blood Pressure Measuring System

Purpose of Research For the feasibility test of a wearable blood pressure measuring device, biometric data of patients in surgical intensive care is simultaneously obtained and stored from the device and a patient monitor.

Overview of Research Design This research is simply to obtain and store the data: 1) invasive arterial blood pressure measured with a patient monitor from one wrist, and 2) multi-wavelength photoplethysmography and pressure signal measured with the developed wearable device from the other wrist.

Medical Device for Clinical Trial Patient monitor (IntelliVue MX700, Philips, NED) and wearable blood pressure measuring device

Target Recruiting Number of Participants Considering the expected patient number in intensive care and 10% dropout rate during the research period, the target recurring number of participants is 220.

Criteria of Participant Inclusion

Adult patient in surgical intensive care of Severance Hospital

Voluntarily agreed to participate the research and not included to the exclusion criteria

Criteria of Participant Exclusion

Patient with occlusive peripheral arterial disease

Patient with cardiovascular disease of serious arrhythmia except atrial fibrillation

Patient administered high-dose of pressor agent over 15 mcg/min

Patient who is not available for measurement of invasive arterial blood pressure

Patient who is not available to wear an wearable device on the wrist

Pregnant woman, minor

Evaluation Parameter

Systolic and diastolic blood pressure, which extracted in mmHg unit from the continuous blood pressure output of patient monitor.

Blood pressure, which estimated by the signal analysis of multi-wavelength photoplethysmography and pressure signal measured with the wearable device.

Data Analysis and Statistical Methods

From the multi-wavelength photoplethysmography and multi-channel pressure signal measured, continuous blood pressure will be estimated through the hemodynamic model and machine learning model.

For the machine learning model, the data obtained will be divided into the training data and evaluation data, with the ratio of 80% and 20%.

In order to evaluate the accuracy of the estimated blood pressure, the invasive arterial blood pressure will be considered as the reference blood pressure. Between the estimated blood pressure and the reference blood pressure, the correlation and root mean square error (RMSE) will be calculated.