DFA α1

DFA α1 is the slope of a log–log line that says how the wobble of the heartbeat series grows with the length of the window you look at it through, measured over windows of 4 to 16 beats. It has no unit.

Why this measure exists

Peng, Havlin, Stanley and Goldberger introduced detrended fluctuation analysis in 1995 to ask a question the standard deviation cannot answer: do the beats remember each other? In pure noise each beat is independent, and the running total of the series wanders like a drunk’s walk; in a series with long-range correlations the wandering grows faster with scale. The exponent $\alpha$ captures the difference: 0.5 for white noise, 1.0 for 1/f noise, 1.5 for a random walk. Healthy heartbeat series showed long-range correlations that broke down in severe heart failure, and the same paper found a crossover: the slope over short windows (α1) differs from the slope over long ones (α2), with the break at about 16 beats. Sassi and colleagues’ 2015 position statement puts DFA among the fractal methods that have improved understanding of the signal, while finding their clinical use so far rather limited.

The formula

$$y(k) = \sum_{i=1}^{k}\left(\mathrm{RR}_i - \mathrm{RR}_{\mathrm{mean}}\right),\qquad F(n) = \sqrt{\frac{1}{N}\sum_{k=1}^{N}\left(y(k) - y_n(k)\right)^2},\qquad F(n) \propto n^{\alpha}$$

Integrate the series: $y(k)$ is the running sum of the intervals’ deviations from the mean. Cut $y$ into boxes of $n$ beats, fit a straight line $y_n$ in each box, and take the root-mean-square of what is left, $F(n)$. Repeat for every box size and plot $\log F(n)$ against $\log n$. The slope over 4 to 16 beats is α1 and over 16 to 64 beats is α2 (Kubios uses 4 to 12 and 13 to 64 by default, and Sassi and colleagues note that many clinical studies take α1 over about 4 to 11 beats).

Worked example

Does the rhythm remember?

Five minutes of the teaching recording

Fluctuation F(n) against box size in beats, log–log α1 α2

4 16 64

α1 (4–16 beats)

0.49

α2 (16–64 beats)

0.65

The rested recording is synthetic: a clean breathing wave every four beats with random noise between beats, so over 4 to 16 beats it looks almost like noise (α1 0.49), and the slower waves give 16 to 64 beats some memory (α2 0.65). Real recordings at rest usually give an α1 of about 1 or higher.

Rested White noise 1/f noise Ectopic beats

White noise mixed in none

The controls work when JavaScript is on.

A log–log chart of fluctuation against box size from 4 to 64 beats, with dots for each box size and two fitted lines: one over 4 to 16 beats for α1, one over 16 to 64 beats for α2. Readouts give α1 and α2, followed by a sentence explaining them. Four buttons choose the rested recording, white noise, 1/f noise or the recording with ectopic beats; a slider mixes white noise into the chosen series.

Five minutes of the teaching recording give α1 = 0.49 and α2 = 0.65. That α1 is low on purpose: the recording is synthetic, a clean breathing wave about once every four beats with random noise between beats, so over short windows it looks almost random, while the slower blood-pressure wave and drift give the longer windows some memory. Real recordings at rest usually give an α1 of about 1 or higher. Switch to white noise and both slopes sit near 0.5 (α1 0.48, α2 0.50); switch to 1/f noise and both sit near 1 (α1 1.04, α2 1.03). Mix noise into any series and α1 slides towards 0.5.

What moves it

A strong, regular breathing swing at rest can pull α1 down, as the teaching recording shows. Exercise intensity moves it from about 1 at easy effort towards 0.5 at hard effort: Rogers and colleagues found in 2021, in 15 runners, that the point where α1 crossed 0.75 matched the first ventilatory threshold, which is the basis of the DFA α1 threshold method some apps offer. Age and heart disease change the values too. Shaffer and Ginsberg note that DFA was designed for recordings spanning several hours, and list 120 seconds as a proposed ultra-short minimum for α1.

When it misleads

Premature beats and artifacts artifacts Artifact An error in a heartbeat recording, such as a missed, extra or mistimed beat, often caused by movement or poor sensor contact. Artifacts can badly distort HRV numbers. More about Artifact add memory-less jumps and pull α1 towards noise, so the series must be cleaned first. The box ranges differ between programs, so Kubios’s α1 and Peng’s are not identical. Short recordings give α2 from too few boxes to trust. And a value near 1 is not “better”: it is a pattern, and in different settings both higher and lower values have gone with worse outcomes.

Where you will see it

Kubios HRV prints DFA α1 and α2 under its nonlinear measures. Some training apps show α1 live during exercise to find the aerobic threshold. The Task Force does not cover it; Sassi and colleagues’ statement reviews the evidence. What are nonlinear HRV measures and do I need them? weighs up whether any of this matters for a reader with a wearable.