JMIR Publications

ABSTRACT

Background:

Continuous telemonitoring of vital signs in the clinical or home setting may lead to improved knowledge of patients’ baseline vital signs, earlier detection of patient deterioration and facilitate migration of care towards home. Little is known about the performance of available wearable sensors, especially during daily life activities, while accurate technology is essential when used for clinical decision making.

Objective:

The aim of this study was to assess the data availability, accuracy and concurrent validity of vital sign data measured with wearable sensors in volunteers during various daily life activities in a simulated free-living environment.

Methods:

Volunteers were equipped with four wearable sensors (Everion at the left and right arm, VitalPatch, and Fitbit Charge 3) and two reference devices (Oxycon Mobile and iButton) to obtain continuous measurements of heart rate (HR), respiration rate (RR), oxygen saturation (SpO2), and skin temperature (T). Participants performed standardized activities including resting, walking, metronome breathing, chores, stationary cycling and recovery afterwards. Data availability was measured in percentage missing data. Accuracy was evaluated by median absolute percentage error (MAPE) and concurrent validity using Bland-Altman bias and 95% limits of agreement (LoA).

Results:

Twenty volunteers (median age [range] of 64 [20-74] years) were included. Data availability was high for all vital signs measured by VitalPatch, and for HR and T measured by Everion. Data availability for HR was lowest for Fitbit (35.7% missing data). Everion showed up to 100% missing data for SpO2. Accuracy of HR was overall high for all wearable sensors. For RR, VitalPatch had an overall MAPE of 8.6% and Everion of 18.9%, where the MAPE was higher during physical activity (up to 27.1%) for both sensors. Accuracy of T was high for VitalPatch (MAPE up to 1.7%) and decreased during the experiment for Everion (MAPE from 6.3% to 9.0%). Bland-Altman analyses showed small biases of VitalPatch for HR (0.1 beats/minute), RR (-0.2 breaths/minute) and T (0.5 °C). Everion and Fitbit underestimated HR up to 5.3 (LoA of -39.0 to 28.3) beats/minute and 11.4 (LoA of -53.8 to 30.9) beats/minute respectively. Everion had small bias with large LoA (-10.8 to10.4 breaths/minute) for RR, underestimated SpO2 (over 1.0%) and overestimated T up to 2.9°C.

Conclusions:

Data availability, accuracy and concurrent validity of the studied wearable sensors varied and differed per activity. In this study, physical activity decreased accuracy and should be taken into account when using VitalPatch or Everion for patient monitoring.