Biomedical Engineering Institute, Kaunas University of Technology, Kaunas, Lithuania.
Physiol Meas. 2018 May 31;39(5):055007. doi: 10.1088/1361-6579/aac24a.
The growing interest to integrate consumer smart wristbands in eHealth applications spawns the need for novel approaches of data parametrization which account for the technology-specific constraints. The present study aims to investigate the feasibility of a consumer smart wristband to be used for computing pulse rate parameters during free-living activities.
The feasibility of computing pulse rate variability (PRV) as well as pulse rate and physical activity-related parameters using the smart wristband was investigated, having an electrocardiogram as a reference. The parameters were studied on the pulse rate and step data from 54 participants, diagnosed with various cardiovascular diseases. The data were acquired during free-living activities with no user lifestyle intervention.
The comparison results show that the smart wristband is well-suited for computing the mean interbeat interval and the standard deviation of the averaged interbeat intervals. However, it is less reliable when estimating frequency domain and nonlinear parameters. Heart recovery time, estimated by fitting an exponential model to the events, satisfying the conditions of the 3 min step test, showed satisfactory agreement (relative error <20%) with the reference ECG in one-third of all cases. On the other hand, the heart's adaptation to physical workload, expressed as the slope of the linear regression curve, was underestimated in most cases.
The present study demonstrates that pulse rate parametrization using a consumer smart wristband is in principle feasible. The results show that the smart wristband is well suited for computing basic PRV parameters which have been reported to be associated with poorer health outcomes. In addition, the study introduces a methodology for the estimation of post-exercise heart recovery time and the heart's adaptation to physical workload during free-living activities.
将消费者智能腕带集成到电子健康应用中的兴趣日益浓厚，这就需要新的方法来对数据进行参数化，以考虑到技术的特定限制。本研究旨在调查在自由活动期间使用消费者智能腕带来计算脉搏率参数的可行性。
使用智能腕带作为参考，研究了计算脉搏率变异性（PRV）以及脉搏率和与身体活动相关的参数的可行性。研究对象为 54 名患有各种心血管疾病的参与者的脉搏率和步频数据。数据是在没有用户生活方式干预的自由活动中采集的。
比较结果表明，智能腕带非常适合计算平均心动间隔和平均心动间隔的标准差。然而，在估计频域和非线性参数时，它的可靠性较低。通过对事件拟合指数模型来估计心脏恢复时间，满足 3 分钟步测试验的条件，在三分之一的情况下与参考 ECG 具有令人满意的一致性（相对误差 <20%）。另一方面，心脏对体力工作负荷的适应能力，表现为线性回归曲线的斜率，在大多数情况下都被低估了。
本研究表明，使用消费者智能腕带对脉搏率进行参数化在原则上是可行的。结果表明，智能腕带非常适合计算基本的 PRV 参数，这些参数已被报道与较差的健康结果相关。此外，该研究介绍了一种在自由活动期间估计运动后心脏恢复时间和心脏对体力工作负荷适应能力的方法。
