USER'S GUIDE - Biosignal Analysis and Medical Imaging Group
USER'S GUIDE - Biosignal Analysis and Medical Imaging Group USER'S GUIDE - Biosignal Analysis and Medical Imaging Group
3.4. Time-varying methods 30where N l is the number of length l lines. The determinism of the time series is measured bythe variable∑ lmaxl=lDET =minlN l∑ N−m+1j,k=1RP(j, k) . (3.27)Finally, the Shannon information entropy of the line length distribution is defined asShanEn = −l∑maxl=l minn l ln n l (3.28)where n l is the number of length l lines divided by the total number of lines, that isn l =3.4 Time-varying methodsN l∑ lmax. (3.29)l ′ =l minN l ′The time-varying methods of the software include the trends of the time-domain measuresRR, SDNN, HR, SD of HR, RMSSD, NN50, and pNN50. For frequency-domain measuresthe trends are obtained for VLF, LF, and HF peak frequencies, VLF, LF, and HF bandpowers, and LF/HF ratio. In addition, trends are calculated for the nonlinear measuresApEn and SampEn. The trends for the time-domain and nonlinear measures are obtainedby using a moving window the length and shift of which can be changed.The frequency-domain measures trends are, instead, obtained from a time-varying spectrumestimate. The time-varying spectrum is estimated either by using the moving windowFFT, which is also known as the spectrogram method, or with the Kalman smootheralgorithm. The Kalman smoother algorithm is an iterative algorithm for estimating theparameters of a time-varying model. In the software, a time-varying AR model is used tomodel the HRV signal. The adaptation of the Kalman smoother algorithm affecting on theresolution of the spectrum can also be altered.3.5 Summary of HRV parametersThe presented time-domain, frequency-domain, nonlinear, and time-varying measures ofHRV calculated by the software are summarized in Table 3.1. For each measure, preferredunits and a short description is given. In addition, a reference to the equation in which thespecific measure is defined is given when possible and related references are given for someof the measures.Kubios HRV Analysisversion 2.0 betaBiosignal Analysis and Medical Imaging GroupDepartment of PhysicsUniversity of Kuopio, FINLAND
3.5. Summary of HRV parameters 31Table 3.1: Summary of the HRV measures calculated by the softwareTime-VaryingNonlinearFrequency-Domain Time-DomainMeasure Units Description ReferencesRR [ms] The mean of RR intervalsSTD RR (SDNN) [ms] Standard deviation of RR intervals [Eq. (3.1)]HR [1/min] The mean heart rateSTD HR [1/min] Standard deviation of intantaneous heart rate valuesRMSSD [ms] Square root of the mean squared differences between successive RRintervals [Eq. (3.3)]NN50Number of successive RR interval pairs that differ more than 50 mspNN50 [%] NN50 divided by the total number of RR intervals [Eq. (3.4)]HRV triangularThe integral of the RR interval histogram divided by the height ofindexthe histogram [44]TINN [ms] Baseline width of the RR interval histogram [44]Peak frequency [Hz] VLF, LF, and HF band peak frequenciesAbsolute power [ms 2 ] Absolute powers of VLF, LF, and HF bandsRelative power [%] Relative powers of VLF, LF, and HF bandsVLF [%] = VLF [ms 2 ]/total power [ms 2 ] × 100%LF [%] = LF [ms 2 ]/total power [ms 2 ] × 100%HF [%] = HF [ms 2 ]/total power [ms 2 ] × 100%Normalized power [n.u.] Powers of LF and HF bands in normalized unitsLF [n.u.] =LF[ms 2 ]/(total power [ms 2 ] − VLF [ms 2 ])HF [n.u.] =HF[ms 2 ]/(total power [ms 2 ] − VLF [ms 2 ])LF/HFRatio between LF and HF band powersSD1, SD2 [ms] The standard deviation of the Poincaré plot perpendicular to (SD1)and along (SD2) the line-of-identity [5, 6]ApEn Approximate entropy [Eq. (3.11)] [40, 12]SampEn Sample entropy [Eq. (3.14)] [40]D 2 Correlation dimension [Eq. (3.21)] [15, 17]DFA Detrended fluctuation analysis: [36, 37]α 1Short term fluctuation slopeα 2Long term fluctuation slopeRPA Recurrence plot analysis: [47, 7, 49]Lmean [beats] Mean line length [Eq. (3.26)]Lmax [beats] Maximum line lengthREC [%] Recurrence rate [Eq. (3.24)]DET [%] Determinism [Eq. (3.27)]ShanEn Shannon entropy [Eq. (3.28)]Time-domain measures:RR, SDNN, HR, STD HR, RMSSD, NN50, and pNN50Frequency-domain measures:Peak frequencies, absolute powers, relative powers, normalized powers, and LF/HF ratio(Note: Time-varying spectrum is estimated using the spectrogram or Kalman smoother method)Nonlinear measures:ApEn and SampEnKubios HRV Analysisversion 2.0 betaBiosignal Analysis and Medical Imaging GroupDepartment of PhysicsUniversity of Kuopio, FINLAND
- Page 3 and 4: 4.3.2 Report sheet . . . . . . . .
- Page 5 and 6: 1.1. System requirements 51.1 Syste
- Page 7: 1.2. Installation 74. Next, select
- Page 10 and 11: 1.2. Installation 1010. The MATLAB
- Page 12 and 13: 1.2. Installation 1214. When the in
- Page 14 and 15: 1.4. Software home page 14• Delet
- Page 16 and 17: Chapter 2Heart rate variabilityHear
- Page 18: 2.2. Derivation of HRV time series
- Page 21 and 22: 2.3. Preprocessing of HRV time seri
- Page 23 and 24: 3.2. Frequency-domain methods 23In
- Page 25 and 26: 3.3. Nonlinear methods 253.3.2 Appr
- Page 27 and 28: 3.3. Nonlinear methods 27−0.6−0
- Page 29: 3.3. Nonlinear methods 29987Time (m
- Page 33 and 34: 4.1. Input data formats 33Figure 4.
- Page 35 and 36: 4.2. The user interface 35Figure 4.
- Page 37 and 38: 4.2. The user interface 37In additi
- Page 39 and 40: 4.2. The user interface 39Figure 4.
- Page 41 and 42: 4.2. The user interface 41Figure 4.
- Page 43 and 44: 4.3. Saving the results 431. Softwa
- Page 45 and 46: 4.3. Saving the results 45Figure 4.
- Page 47 and 48: 4.4. Setting up the preferences 47F
- Page 49 and 50: 4.4. Setting up the preferences 49F
- Page 51 and 52: Chapter 5Sample runsIn this chapter
- Page 53 and 54: 5.1. Sample run 1: General analysis
- Page 55 and 56: 5.1. Sample run 1: General analysis
- Page 57 and 58: 5.2. Sample run 2: Time-varying ana
- Page 59 and 60: Appendix AFrequently asked question
- Page 61 and 62: 61 Why do the power values of Kubio
- Page 63 and 64: References[1] V.X. Afonso. ECG QRS
- Page 65 and 66: References 65[28] J. Mateo and P. L
3.4. Time-varying methods 30where N l is the number of length l lines. The determinism of the time series is measured bythe variable∑ lmaxl=lDET =minlN l∑ N−m+1j,k=1RP(j, k) . (3.27)Finally, the Shannon information entropy of the line length distribution is defined asShanEn = −l∑maxl=l minn l ln n l (3.28)where n l is the number of length l lines divided by the total number of lines, that isn l =3.4 Time-varying methodsN l∑ lmax. (3.29)l ′ =l minN l ′The time-varying methods of the software include the trends of the time-domain measuresRR, SDNN, HR, SD of HR, RMSSD, NN50, <strong>and</strong> pNN50. For frequency-domain measuresthe trends are obtained for VLF, LF, <strong>and</strong> HF peak frequencies, VLF, LF, <strong>and</strong> HF b<strong>and</strong>powers, <strong>and</strong> LF/HF ratio. In addition, trends are calculated for the nonlinear measuresApEn <strong>and</strong> SampEn. The trends for the time-domain <strong>and</strong> nonlinear measures are obtainedby using a moving window the length <strong>and</strong> shift of which can be changed.The frequency-domain measures trends are, instead, obtained from a time-varying spectrumestimate. The time-varying spectrum is estimated either by using the moving windowFFT, which is also known as the spectrogram method, or with the Kalman smootheralgorithm. The Kalman smoother algorithm is an iterative algorithm for estimating theparameters of a time-varying model. In the software, a time-varying AR model is used tomodel the HRV signal. The adaptation of the Kalman smoother algorithm affecting on theresolution of the spectrum can also be altered.3.5 Summary of HRV parametersThe presented time-domain, frequency-domain, nonlinear, <strong>and</strong> time-varying measures ofHRV calculated by the software are summarized in Table 3.1. For each measure, preferredunits <strong>and</strong> a short description is given. In addition, a reference to the equation in which thespecific measure is defined is given when possible <strong>and</strong> related references are given for someof the measures.Kubios HRV <strong>Analysis</strong>version 2.0 beta<strong>Biosignal</strong> <strong>Analysis</strong> <strong>and</strong> <strong>Medical</strong> <strong>Imaging</strong> <strong>Group</strong>Department of PhysicsUniversity of Kuopio, FINLAND