Back

Exercise Intensity Modulates the Validity of Non-Linear Heart Rate Time Series Analysis Window Length: Implications for DFAa1 Monitoring

De Maeseneer, J.; Olieslagers, A.; Gronwald, T.; de Beukelaar, T.

2026-02-04 physiology
10.64898/2026.02.02.703217 bioRxiv
Show abstract

PurposeDetrended fluctuation analysis alpha-1 (DFAa1) has emerged as a promising non-invasive biomarker for exercise intensity assessment. However, the standard 2-min analysis window lacks temporal resolution necessary for real-time training applications. This study systematically investigated the validity of shortened DFAa1 windows (30s and 1min) versus the 2-min reference across different intensities. MethodsPhysically active males completed three continuous cycling protocols: low-intensity training at the first lactate threshold (LOW, n=19), moderate-intensity training at the second lactate threshold (MOD, n=19), and a 30-min self-paced time trial (TT30, n=18). DFAa1 was calculated using 30-s, 1-min, and 2-min moving windows, advancing in 1s increments. Validity was assessed using intraclass correlation coefficients (ICC), Bland-Altman analysis, and standard error of measurement (SEM). ResultsDuring LOW, both shortened windows showed poor agreement with the 2-min reference (30s: ICC=0.02, mean bias of -0.05; 1min: ICC=0.37, -0.02). During MOD, the 30-s window remained unreliable (ICC=0.32, -0.01), while the 1-min window achieved moderate reliability (ICC=0.63, 0.00). During TT30, both shortened windows substantially improved performance (30s: ICC=0.78, -0.02; 1min: ICC=0.95, -0.01), with the 1-min window achieving excellent reliability. ConclusionDFAa1 analysis window validity is intensity-dependent, with shortened windows showing progressively improved agreement as exercise intensity and heart rate increases. While the 2-min window remains essential for low-intensity monitoring, 1-min or 30-s windows provide appropriate validity during high-intensity exercise, enabling more-responsive real-time feedback. These results support adaptive windowing strategies that dynamically adjust window length based on exercise intensity and the number of included data points, to optimize the analytical validity-temporal responsiveness trade-off.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

1
PLOS ONE
4510 papers in training set
Top 14%
14.2%
2
Scientific Reports
3102 papers in training set
Top 7%
10.0%
3
European Journal of Applied Physiology
12 papers in training set
Top 0.1%
10.0%
4
Frontiers in Physiology
93 papers in training set
Top 0.3%
9.1%
5
Physiological Reports
35 papers in training set
Top 0.1%
7.1%
50% of probability mass above
6
American Journal of Physiology-Heart and Circulatory Physiology
32 papers in training set
Top 0.1%
7.1%
7
Journal of Applied Physiology
29 papers in training set
Top 0.1%
6.3%
8
Frontiers in Sports and Active Living
10 papers in training set
Top 0.1%
3.6%
9
American Journal of Physiology-Endocrinology and Metabolism
34 papers in training set
Top 0.1%
3.6%
10
Sensors
39 papers in training set
Top 0.6%
2.7%
11
PeerJ
261 papers in training set
Top 6%
1.9%
12
Medicine & Science in Sports & Exercise
15 papers in training set
Top 0.2%
1.7%
13
eLife
5422 papers in training set
Top 47%
1.3%
14
Journal of The Royal Society Interface
189 papers in training set
Top 3%
1.2%
15
Experimental Physiology
19 papers in training set
Top 0.3%
1.1%
16
The FASEB Journal
175 papers in training set
Top 2%
0.9%
17
Ultrasound in Medicine & Biology
10 papers in training set
Top 0.4%
0.9%
18
GeroScience
97 papers in training set
Top 1%
0.9%
19
Journal of the American Heart Association
119 papers in training set
Top 4%
0.8%
20
Frontiers in Bioengineering and Biotechnology
88 papers in training set
Top 3%
0.7%
21
International Journal of Environmental Research and Public Health
124 papers in training set
Top 7%
0.7%
22
The Journal of Physiology
134 papers in training set
Top 2%
0.7%
23
Cognitive, Affective, & Behavioral Neuroscience
25 papers in training set
Top 0.3%
0.6%
24
Journal of Experimental Biology
249 papers in training set
Top 3%
0.6%
25
PLOS Digital Health
91 papers in training set
Top 3%
0.6%