CUNSTCAMERA (CABINET OF CURIOSITIES) OF CLINICAL SCIENCE
Results of independent metrological testing of programs for calculation of the heart rate variability parameters
7.4.Exactness of estimation of frequencies of spectrum of HRV
Exactness of calculation of frequencies of every accordion of spectral transformation is rendered by substantial influence on repetition of results at the use of the different programs. If the step of diskretization of spectral transformation is too large, then probability of casual hit of certain spectral peak, which is on the border of standard ranges rises, in one or another range. If this spectral peak of has large power, then output these presentations of parameters of spectrum of HRV are substantially distorted.
The use of left-side and right-side tests is foreseen in protocol, that in these test sequences in every spectral range powers (1000 ms2) are set on the counter-clockwise and right scopes of standard ranges.
| The names of files of test sequences are in a format: frequency (f), power (P) | The calculated values of power are in a "left-side" test |
|||||
| CardioLab+ | CardioSpectrum | UM300 | OrtoScience | CARDIO-10DX | МАК-3 | |
| 00032_0041_016_1000.rg | ||||||
| TP (ms2) | 2976 | 5478 | 1465 | 4268 | 3024 | 2924 |
| VLF(ms2) | 566 | 1386 | 126 | 1599 | 1099 | 611 |
| LF(ms2) | 874 | 1721 | 434 | 1404 | 968 | 1024 |
| HF(ms2) | 1013 | 2370 | 498 | 1265 | 957 | 1006 |
In a left-side test from data of table some programs are calculated by the understated powers in area of very low frequencies (VLF). In particular, in the three programs: UM300, CardioLab+, MAK-3 measured VLF taking into account a bias error (for UM300 -50%, see section 7.1 Errors of calculation of spectrum power) 25% makes in relation to the set frequency, 57% and 61% accordingly. In the programs of CardioSpectrum and OrtoScience taking into account their bias error of measuring of VLF the satisfactory measuring is marked on the counter-clockwise edge of this range. Most high-quality results of this test has program CARDIO-10DX.
The graphs of spectrums are below presented in the testable programs on results treatment of this test sequence.
| Кардиоспектр | CARDIO-10DX | UM300 |
 |
 |
 |
| CardioLab+ | OrtoScience | МАК-3 |
 |
 |
 |
The followings results are got in a right-side test:
The names of files of test sequences are in a format: frequency (f), power (P) |
The calculated values of power are in a "right-side" test | |||||
| CardioLab+ | CardioSpectrum | UM300 | OrtoScience | CARDIO-10DX | МАК-3 | |
| 0039_014_039_1000.rg | ||||||
| TP (ms2) | 2907 | 6083 | 1484 | 3474 | 2677 | 2977 |
| VLF(ms2) | 744 | 270 | 270 | 1540 | 3 | 997 |
| LF(ms2) | 1295 | 3211 | 770 | 1807 | 1016 | 987 |
| HF(ms2) | 866 | 2602 | 521 | 127 | 658 | 993 |
More substantial errors are exposed in this test. In the program CardioLab+ part of powers of VLF and HF went out into the test of LF. In the program CardioSpectrum VLF reckoned as LF.
In the program OrtoScience part of HF reckoned as LF. In the program CARDIO-10DX almost all of VLF reckoned as LF.
The graphs of spectrums of implementation of this test in the different programs are presented below.
| CardioSpectrum | CARDIO-10DX | UM300 |
 |
 |
 |
| CardioLab+ | OrtoScience | МАК-3 |
 |
 |
 |
Comment:
the results of this test show existence of high probability of errors of calculation of spectral powers in different ranges at the location of measureable accordions on the scopes of spectral ranges. This methodical problem requires perfection of the programs of calculation of spectrum of HRV (for example, increase of discreteness of spectral transformation). From other side these results reflect a methodological problem which is related to the attempt of hard classification of spectral ranges.
