抄録
Many variability indices, which have been proposed to quantitatively represent both short-term variability (=STV) and long-term variability (=LTV), were analyzed mathematically and the following static property was obtained.
All of the approximate expectations for indices developed by Tarlo, Kero, Dalton, Heilbron and Cabal assumed the same formula k√1-ρσ (k: constant, ρ: correlation coefficient between the beat-to-beat interval Ti and the adjacent interval Ti+1, σ: standard deviation of Ti), and were indentical except for the constants. Those values for de Haan's and Yeh's indices were k√1-ρσ/T0 (T0: mean of Ti), while those for Modanlou's, Wade's, and Organ's indices were k√1-ρσ/T02 respectively.
Hence, all of these indices represented the same quantity in essence when the mean beat-to-beat interval was constant. The expected value for de Haan's and Heilbron's LTV indices was approximately k√1+ρσ (k: constant), while those for Yeh's, Organ's, and Cabal's LTV indices essentially showed standard deviation (=σ) of Ti.
From these results, it can be concluded that measuring STV and LTV according to those formulae means evaluating ρ and σ at the same time.
Hence, there may be little significance in measuring them individually if ρ changes little. That is, it may suffice merely to measure the standard deviation (=σ) of Ti as a quantity of variability.
