Athlete summary
On a long easy ride, heart rate often drifts up at the same watts. That drift is not just “fitness noise”: Rothschild et al. (Eur J Appl Physiol 2025) showed it tracks the drop in power at VT1 — the moderate→heavy boundary coaches use for Z2 / polarized easy.
When Koyaro knows your baseline VT1 (LT1) and can measure HR at that power early vs late, it can predict a late-ride VT1. Respiratory frequency (FR) from a strap or lab file is optional and improves the published full model. V̇O2peak (the logged vo2max measurement) unlocks the mid and full models.
If HR or baseline VT1 is missing, Koyaro does not invent a fatigued VT1 from FTP. Whole-ride Pw:Hr % decoupling and kJ/kg MMP retention stay as separate signals.
This does not replace:
Engine formulas (Table 2)
Cite-locked GEE coefficients from Rothschild, Gallo, Hamilton, Stevenson, Dudley-Rode, Charoensap, Plews, Kilding & Maunder, Eur J Appl Physiol 2025;125:2911–2920 (doi 10.1007/s00421-025-05815-0).: copy published coefficients — do not invent.
Decoupling here is Δ at a fixed power near baseline VT1 (paper: last 1 min of the VT1 step of a 5- or 6-step test), not Pw:Hr percent:
Field band: | P − VT1| ≤ 12.5 W (half of the paper’s 25 W five-step increment).
Model selection (richest published subset)
| Inputs | Model | Apparent MAE | R2 |
|---|---|---|---|
| VT1 + ΔHR + ΔFR + time + V̇O2peak | Full | 7.2 W | 0.95 |
| VT1 + ΔHR + V̇O2peak | HR + V̇O2peak | 7.6 W | 0.94 |
| VT1 + ΔHR | HR-only | 8.3 W | 0.94 |
| Missing HR or VT1 | Unknown | — | — |
Bootstrap-cv MAE for the full model is 8.3 W (paper). VE decoupling is not used (rrm = −0.25, ns).
Full (Table 2 row 1)
V̇O2peak is relative (mL·kg⁻¹·min⁻¹). Time is exercise minutes at the late assessment. FR is breaths·min⁻¹.
HR + V̇O2peak (Table 2 row 2)
HR-only (Table 2 row 3)
Literature goldens
| Source | Constraint |
|---|---|
| ΔVT1 vs ΔHR | rrm = −0.76 |
| ΔVT1 vs ΔFR | rrm = −0.40 |
| ΔVT1 vs ΔV̇E | rrm = −0.25 (ns — unused) |
| Study A means | VT1 217 W; V̇O2peak 59.9 mL·kg⁻¹·min⁻¹ |
| Observed VT1 drop after ~2.5 h | 1–45 W (pooled prior studies) |
Limits (encode honesty)
- Protocols were primarily ~90% of baseline VT1 for ~2.5 h — not race surges or intervals.
- Pre- and peri-exercise feeding was controlled in the source studies.
- Do not invent baseline VT1 from FTP×0.75 or V̇O2peak from watts.
- Do not invent FR = 0 to unlock the full model.
- Complementary to kJ/kg and fatigued CP — do not overwrite those.
Literature
- Rothschild JA, Gallo G, Hamilton K, Stevenson JD, Dudley-Rode H, Charoensap T, Plews DJ, Kilding AE, Maunder E. Durability of the moderate-to-heavy intensity transition can be predicted using readily available markers of physiological decoupling. Eur J Appl Physiol. 2025;125:2911–2920. doi 10.1007/s00421-025-05815-0.