A novel method of heart rate variability measurement

Shi-Liang Shao; Ting Wang; Chun-He Song; E-Nuo Cui; Hai Zhao; Chen Yao

doi:10.7498/aps.68.20190372

Journals >Acta Physica Sinica >Volume 68 >Issue 17 >Page 178701-1 > Article

Acta Physica Sinica
Vol. 68, Issue 17, 178701-1 (2019)

A novel method of heart rate variability measurement

Shi-Liang Shao^{1、2、3、*}, Ting Wang^2、3, Chun-He Song^2、3, E-Nuo Cui¹, Hai Zhao¹, and Chen Yao^2、3

Author Affiliations

¹School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China

²State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China

³Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences, Shenyang 110169, China

show less

DOI: 10.7498/aps.68.20190372 Cite this Article

Shi-Liang Shao, Ting Wang, Chun-He Song, E-Nuo Cui, Hai Zhao, Chen Yao. A novel method of heart rate variability measurement[J]. Acta Physica Sinica, 2019, 68(17): 178701-1 Copy Citation Text

show less

Fig. 1. Framework of the HRV analysis.HRV分析框架图

Download full size | View in the Article

Fig. 2. Schematic of obtaining HRV signal.HRV信号获得示意图

Download full size | View in the Article

Fig. 3. Implementation process of ICBN analysis method.ICBN分析方法实现过程框图

Download full size | View in the Article

Fig. 4. Mean and variance of the indicators with very significant differences between the two groups of NSR1 and CHF.NSR1和CHF这2组对象具有极显著差异的指标的均值与标准差

Download full size | View in the Article

Fig. 5. Mean and variance of the indicators with very significant differences between the two groups of NSR2 and AF.NSR2和AF这2组对象具有极显著差异的指标的均值与标准差

Download full size | View in the Article

指标	单位	描述与定义
SDNN	${\rm{ms}}$	相邻正常心跳间隔的标准差 ${\rm{SDNN}} = {\sqrt {\dfrac{1}{N}\displaystyle \sum \limits_{i = 1}^N \left( {{\rm{RR}}{s_i} - \dfrac{1}{N}\mathop \sum \limits_{i = 1}^N {\rm{RR}}{s_i}} \right)} ^{}}$
RMSSD	ms	相邻正常心跳间隔差值的平方和均值的均方根 ${\rm{RMSSD}} = \sqrt {\dfrac{1}{{N - 1}}\displaystyle \sum \limits_{i = 1}^{N - 1} {{\left( {{\rm{RR}}{s_{i + 1}} - {\rm{RR}}{s_i}} \right)}^2}} $
pNN50	%	相邻正常心跳间隔差值超过50毫秒的比例 ${\rm{PNN}}50 = \dfrac{{{\rm{num}}\left[ {\left( {{\rm{RR}}{s_{i + 1}} - {\rm{RR}}{s_i}} \right) > 50{\rm{ ms}}} \right]}}{{N - 1}}$
HRVTi	—	相邻正常心跳间隔的总个数除以相邻正常心跳间隔直方图的高度

Table 1.

Statistical features in time domain.

时域分析统计特征

指标	单位	描述与定义	频率范围
Total power	${\rm{m}}{{\rm{s}}^2}$	所有频率范围的功率谱总和	≤ 0.4 Hz
VLF	${\rm{m}}{{\rm{s}}^2}$	甚低频范围内的功率谱	0.003—0.040 Hz
LF	${\rm{m}}{{\rm{s}}^2}$	低频范围内的功率谱	0.04—0.15 Hz
HF	${\rm{m}}{{\rm{s}}^2}$	高频范围内的功率谱	0.15—0.40 Hz
LF/HF	%	LF $\left[ {{\rm{m}}{{\rm{s}}^2}} \right]$ 与HF $\left[ {{\rm{m}}{{\rm{s}}^2}} \right]$的比值	—

Table 2.

Statistical features in frequency domain.

频域分析统计特征

指标		NSR1(mean $ \pm $SD)	CHF(mean $ \pm $SD)	标准误差差值的95%置信区间		$p$
指标		NSR1(mean $ \pm $SD)	CHF(mean $ \pm $SD)	下限	上限	$p$
注: , , **分别代表 $p < 0.05$$p < 0.01$, $p < 0.001$.
ICBN	WB	21.853 $ \pm $1.479	27.835 $ \pm $7.741	3.050	8.914	0^***
	PW	0.563 $ \pm $0.051	0.455 $ \pm $0.103	–0.151	–0.066	0^***
	TE	0.956 $ \pm $0.019	0.937 $ \pm $0.037	–0.034	–0.005	0.009^**
	CL	1.135 $ \pm $0.133	0.954 $ \pm $0.194	–0.268	–0.094	0^***
	MC	0.684 $ \pm $0.035	0.705 $ \pm $0.026	–0.038	–0.006	0.009^**
时域	SDNN	81.507 $ \pm $38.566	59.535 $ \pm $44.76	–43.951	0.007	0.05
	pNN50	11.476 $ \pm $14.676	10.772 $ \pm $14.110	–8.277	6.870	0.853
	RMSSD	51.172 $ \pm $54.895	60.307 $ \pm $58.497	–20.707	38.976	0.542
	HRVTi	6.886 $ \pm $2.452	4.093 $ \pm $1.494	–3.861	–1.725	0^***
频域	TP	1.809 $ \pm $4.909	1.443 $ \pm $4.052	–2.734	2.001	0.758
	VLF	0.0003 $ \pm $0.448	1.387 $ \pm $6.214	–1.265	3.370	0.367
	LF	0.212 $ \pm $0.333	0.119 $ \pm $0.316	–0.263	0.078	0.281
	HF	1.597 $ \pm $4.608	1.322 $ \pm $3.745	–2.483	1.934	0.804
	LF/HF	0.288 $ \pm $0.184	0.108 $ \pm $0.083	–0.255	–0.105	0^***

Table 3.

Statistical analysis results of HRV index under different analysis methods.

NSR1和CHF患者不同分析方法下的结果

指标	TP	TN	FP	FN	Acc	Sen	Spe	AUC
注: TP, 被判定为CHF病人的数量; TN, 被判定为NSR1对象的数量; FP, NSR1对象被判定为CHF病人的数量; FN, CHF病人被判定为NSR1对象的数量; 正确率 ${\rm{Acc}} = \dfrac{{{\rm{TP + TN}}}}{{{\rm{TP \!+\! FP \!+\! TN \!+\! FN}}}} \times 100\% $; 灵敏度 ${\rm{Sen}} = \dfrac{{T{\rm{P}}}}{{T{\rm{P}} + {\rm{FN}}}} \times 100\% $; 特异度 ${\rm{Spe}} = \dfrac{{{\rm{TN}}}}{{{\rm{FP \!+\! TN}}}} \times 100\% $; ${\rm{AUC}} = \dfrac{1}{2}\left( {\dfrac{{{\rm{TP}}}}{{{\rm{TP + FN}}}}{\rm{ + }}\dfrac{{{\rm{TN}}}}{{{\rm{TN + FP}}}}} \right) \times 100\% $.
WB	19	27	10	2	79.31	90.48	72.97	81.72
PW	19	24	10	5	74.14	79.17	70.59	75.53
CL	20	23	9	6	74.14	76.92	71.88	74.40
HRVTi	23	19	6	10	72.41	69.70	76.00	72.85
LF/HF	25	17	4	12	72.41	75.68	80.95	78.32

Table 4.

Performance comparisons of different indices for CHF recognition

不同特征的CHF识别性能对比

指标	TP	TN	FP	FN	Acc	Sen	Spe	AUC
WB&CL&LF/HF	25	27	4	2	89.66	92.59	87.1	89.85
WB&PW&CL&HRVTi&LF/HF	24	27	5	2	87.93	92.31	84.38	88.35
WB&PW&CL&LF/HF	24	27	5	2	87.93	92.31	84.38	88.35
WB&PW	22	29	7	0	87.93	100	80.56	90.28
WB&CL&HRVTi&LF/HF	25	25	4	4	86.20	86.21	86.21	86.21
WB&PW&HRVTi&LF/HF	24	26	5	3	86.20	88.89	83.87	86.38
WB&CL&HRVTi	25	25	4	4	86.20	86.21	86.21	86.21
WB&PW&HRVTi	24	26	5	3	86.20	88.89	83.87	86.38

Table 5.

Performance comparisons of different indices for CHF recognition.

不同特征组合的CHF识别性能对比

指标		NSR2(mean $ \pm $SD)	AF(mean $ \pm $SD)	标准误差差值的95%置信区间		$p$
指标		NSR2(mean $ \pm $SD)	AF(mean $ \pm $SD)	下限	上限	$p$
注: , , **分别代表 $p < 0.05$, $p < 0.01$, $p < 0.001$.
ICBN	WB	21.483 $ \pm $1.367	24.243 $ \pm $3.105	1.731	3.789	0^***
	PW	0.567 $ \pm $0.074	0.454 $ \pm $0.090	–0.148	–0.077	0^***
	TE	0.941 $ \pm $0.050	0.960 $ \pm $0.010	0.004	0.035	0.013^*
	CL	1.113 $ \pm $0.146	0.999 $ \pm $0.170	–0.183	–0.047	0.001^**
	MC	0.687 $ \pm $0.043	0.664 $ \pm $0.032	–0.024	–0.008	0.04^*
时域	SDNN	74.698 $ \pm $26.193	139.016 $ \pm $62.480	10.331	43.773	0^***
	pNN50	10.123 $ \pm $9.610	45.495 $ \pm $30.687	4.904	25.620	0^***
	RMSSD	36.402 $ \pm $19.003	170.926 $ \pm $97.980	15.220	104.256	0^***
	HRVTi	7.735 $ \pm $3.210	6.049 $ \pm $2.488	–2.918	–0.455	0.008^**
频域	TP	0.615 $ \pm $0.612	13.493 $ \pm $19.369	7.000	18.754	0.001^**
	VLF	0.0002 $ \pm $0.0003	0.017 $ \pm $0.063	–2.256	36.147	0.083
	LF	0.157 $ \pm $0.154	1.204 $ \pm $1.620	0.553	1.540	0.002^**
	HF	0.458 $ \pm $0.515	12.272 $ \pm $17.908	6.381	17.247	0.001^**
	LF/HF	0.515 $ \pm $0.419	0.126 $ \pm $0.059	–0.519	–0.259	0^***

Table 6.

Statistical analysis results of HRV index under different analysis methods

NSR2和AF患者在不同分析方法下的结果

指标	TP	TN	FP	FN	Acc	Sen	Spe	AUC
WB	30	37	13	6	77.91	83.33	74.00	78.67
PW	30	42	13	1	83.72	96.77	76.36	86.57
SDNN	31	34	12	9	75.58	77.50	73.91	75.71
pNN50	28	39	15	4	77.91	87.50	72.22	79.86
RMSSD	29	33	14	10	72.09	74.36	70.21	72.29
LFHF	42	24	1	19	76.74	68.85	96.00	82.43

Table 7.

Performance comparisons of different indices for AF recognition.

不同特征的AF识别性能对比

指标	TP	TN	FP	FN	Acc	Sen	Spe	AUC
WB&PW&pNN50&RMSSD&LFHF	38	41	5	2	91.86	95.00	89.13	92.07
WB&PW&SDNN&LFHF	38	40	5	3	90.70	92.68	88.89	90.79
WB&PW&RMSSD&LFHF	38	40	5	3	90.70	92.68	88.89	90.79
WB&PW&SDNN&RMSSD	37	41	6	2	90.70	94.87	87.23	91.05
WB&PW&SDNN&pNN50&RMSSD	36	42	7	1	90.70	97.30	85.71	91.51

Table 8.

Performance comparisons of different indices for AF recognition.

不同特征的AF识别性能对比

Shi-Liang Shao, Ting Wang, Chun-He Song, E-Nuo Cui, Hai Zhao, Chen Yao. A novel method of heart rate variability measurement[J]. Acta Physica Sinica, 2019, 68(17): 178701-1

Download Citation

Tools

Save the article for my favorites

Paper Information