REF-438 · Vesterinen et al., 2016 + Kiviniemi HRV-Guided Training · Medicine & Science in Sports & Exercise, · Paper
HRV-guided training: individualized periodization based on daily autonomic readiness
Exemplar of wearable-based phenotyping
Key Findings — Vesterinen 2016 Core Finding
n=40 recreational endurance runners, divided into an HRV-guided group (EXP) and a traditional predefined-training group (TRAD). After a 4-week preparation period, an 8-week intensive period: TRAD followed a predefined program with two to three moderate (MOD) and high-intensity (HIT) sessions per week; in EXP the timing of MOD/HIT sessions was decided each morning from HRV.
Decision rule (as in the abstract): a MOD/HIT session was programmed only if that morning's HRV was within the individually determined smallest worthwhile change (SWC) band; otherwise the day was low-intensity training
HRV-guided group performed fewer MOD/HIT sessions: 13.2 ± 6.0 vs 17.7 ± 2.5 (P = 0.021, effect size 0.98); no other training differences between groups
3000-m running performance improved in EXP (2.1% ± 2.0%, P = 0.004) but not significantly in TRAD (1.1% ± 2.7%, P = 0.118); the between-group difference in the change was small (effect size 0.42)
VO₂max improved in both groups (EXP 3.7% ± 4.6%, P = 0.027; TRAD 5.0% ± 5.2%, P = 0.002)
Key: the HRV-guided group reached a similar or slightly better race-distance result with about 25% fewer hard sessions (arithmetic: 13.2 vs 17.7) — the authors conclude resting HRV shows potential for individualising the timing of vigorous sessions.
Kiviniemi 2007/2010 Foundation (not checked against an abstract — no identifier on file)
Earlier studies from the same line of research used daily morning HRV to decide between HIT and easy days and reported equal or better fitness gains with fewer hard sessions than a standardized protocol. Read as background, not as verified numbers.
Phenotype Application
1
HRV-responsive athletes: benefit from HRV-guided autoregulation (inference from the trial's group-level result; the study did not classify individuals)
2
HRV-non-responsive athletes exist in practice (little HRV change despite training load) and may need other monitoring (RPE, performance) — inference, not tested; the study reports no share of such athletes
3
Recovery-sensitive phenotype (Archetype 6): the base's hypothesis for who gains most from HRV guidance — not tested here
Implementation Framework (measurement, rolling average and band are the trial's own method per the full text; the high-value rationale is the base's own practice) (corrected 2026-10-12, full text)
Morning HRV (rMSSD or Ln-rMSSD) measured supine after waking; the trial recorded 4 min of supine R-R intervals each morning with a chest belt (corrected 2026-10-12, full text)
A 7-day rolling average of RMSSD with an individual SWC band of mean ± 0.5 × SD from four weeks of baseline (PREP) values, as in the trial (corrected 2026-10-12, full text)
Decision rule modelled on Vesterinen: HRV within the band → the planned hard session; HRV outside the band (below OR clearly above — an unusually high value can be parasympathetic saturation, not readiness) → easy day; the trial itself used "within the SWC" as the only green light
Practical Application
In 40 recreational runners, deciding each morning from HRV whether to run the planned hard session let the HRV-guided group do fewer hard sessions (13.2 vs 17.7 over 8 weeks) while improving 3000-m time by 2.1% (the fixed-plan group's 1.1% was not significant; between-group effect small, ES 0.42); VO₂max rose in both groups — HRV timing of hard days buys the same fitness for less hard work, not more fitness
The trial's green light was HRV *within* the athlete's normal band — an unusually high reading was treated like a low one (easy day); copy that rule, not "higher is better"
As the trial did (full text): measure supine after waking, compare a 7-day rolling RMSSD to an individual band (mean ± 0.5 × SD of a 4-week baseline), and let the day's plan move (corrected 2026-10-12, full text); the base's own practice adds — hard sessions still get done, only their timing shifts
Caveats
HRV varies day-to-day; trends matter more than absolute values
Different devices vary in HRV accuracy (chest strap > wrist; supine > sitting)
Some athletes show paradoxical HRV (high after a hard block = parasympathetic rebound, not readiness) — a pattern to recognize
HRV does not capture all readiness factors (psychological, GI, etc.)
Limitations
Moderate sample (n=40, two groups); allocation method not stated in the abstract
The between-group difference in 3000-m change was small (ES 0.42); VO₂max gains did not differ in favour of HRV guidance
Outcome metrics were VO₂max and a 3000-m run, not real-world racing
Does not address how to handle athletes who do not respond to HRV guidance itself
Daily measurement may not be sustainable for amateurs
How this entry was checked and changed
Why this level
regraded 2026-09-27 from A: two-arm controlled trial, n=40, allocation method not stated in the abstract; between-group effect small (ES 0.42)
Content correction 2026-09-27
"4-week intervention" -> 4-week preparation period followed by an 8-week intensive period. "+2.1% vs +1.1% (p<0.05)" -> the within-group P values were 0.004 (EXP) and 0.118 (TRAD); the between-group difference was small (ES 0.42), not a significant group contrast.
Decision rule "rising/stable = HIT; falling = easy" and "HRV > baseline + SWC → HIT" -> the abstract programmed MOD/HIT only when HRV was WITHIN the SWC; otherwise low intensity. Corrected.
Added the VO₂max result (both groups improved: 3.7% vs 5.0%). "~10-15% HRV-non-responsive" -> not in the abstract; removed. "7-day rolling average", "±0.5×SD / ±1.0×SD", "SWC = 0.5 × CV" -> the "±1.0×SD" and "0.5 × CV" figures are not from this paper, but the 7-day rolling average and the SWC of mean ± 0.5 × SD are the trial's own method (corrected 2026-10-12, full text).
Level A -> B+ (two-arm trial, n=40, allocation not stated); added Tier/Sport lines.
2026-09-28: added Journal line from the Europe PMC and OpenAlex records (DOI 10.1249/MSS.0000000000000910, PMID 26909534). Note: the Representative citations line says 48(11):2196-2205, which does not match the DOI's record — left unchanged, flagged for the owner.
2026-09-28: Vesterinen 2016 citation volume/pages corrected from the DOI's Crossref record (48(11):2196-2205 -> 48(7):1347-1354).
2026-10-01: cross-reference to SYNTH-13 no longer names an archetype — that model was retired as an unvalidated composite.
Full text (checked 2026-10-12)
Participants were 20 women and 20 men with at least 2 years of running background; women were 34 ± 8 years with VO2max 49 ± 4 mL·kg-1 min-1 and men 35 ± 7 years with 56 ± 5 mL·kg-1 min-1 (full text).
Allocation: runners were matched into pairs by sex, age, training background, VO2max, 3000-m time and HRV, then randomized within each pair to EXP or TRAD (full text).
Correction: the entry labels the rolling average and individual SWC band "Not from this paper", but the trial measured 4 min of supine R-R intervals each morning (Omegawave chest belt), used a 7-day rolling average of RMSSD, and set the SWC as mean ± 0.5 * SD from four weeks of PREP values (full text).
When RMSSD7day fell outside the SWC, subjects trained only at low intensities or rested until RMSSD7day returned to the mean level of the SWC; the SWC was updated after the first four weeks of INT, and TRAD did ~50% of sessions below LT1 and ~50% MOD/HIT in a 3:1 week pattern (full text).
31 subjects completed (13 in EXP, 18 in TRAD); nine dropped out due to injuries (n = 2), sicknesses (n = 2) and insufficient compliance (n = 5) (full text).
3000-m time improved by −14.3 ± 14.1 s in EXP and −7.8 ± 19.8 s in TRAD; individual changes ranged from -1 to 6% in EXP and -4 to 8% in TRAD, and MOD/HIT session counts from 5 to 24 (EXP) versus 11 to 21 (TRAD) (full text).
Running speed at LT1 improved only in EXP (moderate between-group difference), no sex differences were found, and the number of MOD/HIT sessions did not correlate with the change in RS3000m (r = 0.42, P = 0.17) (full text).