Takahashi Naomi, Kuriyama Akira, Kanazawa Hoshinori, Takahashi Yoshimitsu, Nakayama Takeo
Department of Health Informatics, Kyoto University School of Public Health, Kyoto, Japan.
Department of Mechano-Informatics, Graduate School of Information Science and Technology, The University of Tokyo, Tokyo, Japan.
Pacing Clin Electrophysiol. 2017 Sep;40(9):1004-1009. doi: 10.1111/pace.13138.
To broaden the utility of heart rate variability (HRV) in clinical medicine and mass screening, results based on shorter electrocardiogram (ECG) recordings require validation with those based on standard 5-minute recordings. We investigated the association between HRV variables obtained from 5-minute ECGs with those obtained from ECGs shorter than 5 minutes.
Twenty-two participants aged 20-69 years underwent 5-minute resting ECG recordings in the supine position with natural breathing. Spectral analysis using MemCalc method was performed to calculate high-frequency (HF, which required at least 10 seconds) and low-frequency (LF, which required at least 30 seconds) components. Participants were not strictly preconditioned as in previous experimental studies in order to simulate a setting similar to that of a general health checkup. Associations of each variable between the 5-minute ECG recordings and those for shorter recordings were examined by Pearson's correlation coefficients and Bland-Altman plots.
HF and LF components were log-transformed based on their distributions. Correlation coefficients between 5-minute data and shorter recordings in the supine position with natural breathing ranged from 0.80 to 0.91 (HF by 10-second recording, 0.80; LF by 30-second recording, 0.83, respectively). Bland-Altman plots showed that gaps between the values from both methods slightly increased as the HF and LF component values increased.
Although slight proportional errors were possible, values from standard 5-minute and shorter recordings in the supine position were strongly correlated. Our findings suggest that shorter ECG data without strict preconditioning can be reliably used for spectral analysis.
为了扩大心率变异性（HRV）在临床医学和大规模筛查中的应用，基于较短心电图（ECG）记录的结果需要与基于标准5分钟记录的结果进行验证。我们研究了从5分钟心电图获得的HRV变量与从短于5分钟的心电图获得的HRV变量之间的关联。
22名年龄在20 - 69岁之间的参与者在仰卧位自然呼吸状态下进行了5分钟的静息心电图记录。使用MemCalc方法进行频谱分析，以计算高频（HF，至少需要10秒）和低频（LF，至少需要30秒）成分。为了模拟类似于一般健康检查的环境，参与者没有像之前的实验研究那样进行严格的预处理。通过Pearson相关系数和Bland - Altman图检查5分钟心电图记录与较短记录之间各变量的关联。
根据HF和LF成分的分布进行对数转换。仰卧位自然呼吸状态下，5分钟数据与较短记录之间的相关系数范围为0.80至0.91（10秒记录的HF为0.80；30秒记录的LF为0.83）。Bland - Altman图显示，随着HF和LF成分值的增加，两种方法所得值之间的差距略有增大。
虽然可能存在轻微的比例误差，但标准5分钟记录和仰卧位较短记录所得值之间具有很强的相关性。我们的研究结果表明，未经严格预处理的较短心电图数据可可靠地用于频谱分析。
