Estimating heart rate variation during walking with smartphone
Estimating heart rate variation during walking with smartphone
复制标题
DOI：
10.1145/2493432.2493491
复制
发表时间：
2013-09
期刊：
影响因子：
--
通讯作者：
M. Sumida;Teruhiro Mizumoto;K. Yasumoto
中科院分区：
文献类型：
--
作者：
M. Sumida;Teruhiro Mizumoto;K. Yasumoto
为了实现的应用程序，支持用户享受步行与适当的物理负载，我们提出了一种方法来估计物理负载及其变化，在步行过程中只有可用的功能的智能手机。由于身体负荷与心率呈线性关系，我们的目的是用智能手机估计心率。为此，我们构建心率预测模型，通过机器学习从包括加速度和步行速度在内的步行数据中预测心率变化。为了跟踪身体负荷的意外变化，我们把注意力集中在具有类似于心率的特性的摄氧量上，并设计了一种新的技术来从加速度和GPS数据估计摄氧量，以便将其用作模型的输入。此外，为了适应个体之间的心率变化的差异，我们设计了技术来优化参数为每个配置文件为基础的类别的用户和归一化心率吸收个体差异。我们将所提出的方法应用于不同人在各种路线上的实际步行数据，并证实该方法估计心率变化的平均误差小于7次/分钟。
Aiming to realize the application which supports users to enjoy walking with an appropriate physical load, we propose a method to estimate physical load and its variation during walking only with available functions of a smartphone. Since physical load has a linear relationship with heart rate, our purpose is to estimate heart rate with a smartphone. To this end, we build heart rate prediction models which predict heart rate variation from walking data including acceleration and walking speed by machine learning. In order to track unexpected change of physical load, we focus attention on oxygen uptake which has a similar property to heart rate and devise a novel technique to estimate the oxygen uptake from acceleration and GPS data so that it is used as an input of the model. Moreover, to adapt to difference of heart rate variation among individuals, we devise techniques to optimize parameters for each profile-based category of users and to normalize heart rate to absorb individual difference. We applied the proposed method to actual walking data on various routes by different persons and confirmed that the method estimates heart rate variation with the mean error of less than 7 beat per minute.