From ventilation to muscle: reading performance as a system

From ventilation to muscle: reading performance as a system

Why profiling a WorldTour rider demands reading ventilation as an ecosystem of cross-referenced indices — and where diagnosis stops being a measurement and becomes a line of reasoning.

Download:

https://doi.org/10.5281/zenodo.20664961

A power meter never tells you why

A power meter measures one thing: the output. The watts at the pedal are the terminal result of an entire chain, not its explanation.

Two riders can hold exactly the same power at threshold with opposite physiological architectures — one capped upstream by ventilatory mechanics,

the other by the capacity to extract oxygen at the periphery. The power meter superimposes them. Profiling has to separate them.

This is the very logic of an automotive diagnostic scanner. You do not read a single sensor: you interrogate every subsystem of the vehicle, and the tool returns not a list of values but the fault code to address first.

Profiling an athlete applies that logic to the oxygen chain — from ventilated air to the quadriceps mitochondrion that consumes it. At every link, one question: is this where it caps?

And if so, is it remediable?

Ventilation is one of those links. It is also the one that stayed unreadable the longest — not for want of sensors, but for want of a grid.

Ventilation, the last system without a grid

We have long recorded minute ventilation (VE), respiratory frequency (Rf) and tidal volume (Tv).

The problem was never to measure, but to interpret. Faced with a Tv of 2.4 L at VT1, or an Rf that runs away as the threshold is crossed, the question stays unanswered until you have a reference: is it good? for whom? at what intensity?

The literature offers general physiological constants (Wasserman et al. 2012), but no reading grid specific to the elite cyclist — whose ventilatory function operates in ranges that clinical norms do not cover.

Yet ventilation is not a passive consequence of effort. The way one ventilates — the Rf/Tv couple, the moment it tips — is a trainable lever. Spontaneously, the rider drifts toward a costly regime: high Rf, low Tv, a pattern one may call anarchic because it maximises ventilatory cost for mediocre useful ventilation.

VST (Ventilatory Strategies Training) restructures this pattern toward an economical regime: lower Rf, higher Tv, sustainable at the target intensity — sustainability being the constraint that separates a trained pattern from a mere mechanical maximum. But you only train what you can read.

The trap, then, would be to look for the index that would sum up ventilatory function. It does not exist, any more than a single sensor sums up the state of an engine.

Ventilation is a chain: inspiratory pump → volume balance → pattern efficiency → conversion into muscular extraction → cardiac coupling. The limiter can sit at any of these links — and, decisively, the same figure does not carry the same meaning depending on the link against which it is read.

The reading principle: never an index alone

This is where the ecosystem logic carries its full weight. The FIV index (Forced Inspiratory Volume) is its archetype, and is worth dwelling on, because it teaches a grammar that holds for all the others.

A raw FIV — a forced inspiratory volume in litres — is worth nothing on its own. The same number is read in cascade, against several successive references, each illuminating a different facet:

•       against the speed of realisation: a volume mobilised in under 0.80 s is not the same asset as a volume merely reached, slowly;

•       against FEV1, which yields the ICIF and reveals the balance between inspiratory capacity and forced expiratory capacity;

•       against FEV6, which yields the IMD and flags a possible mobilisation ceiling.

These partial readings then converge in a four-quadrant synthesis: a large inspiratory volume is favourable only if it is realised fast, balanced against expiratory capacity, and close to FEV6 with no mobilisation deficit.

An impressive FIV can thus mask a limitation: it inflates the ICIF artificially while it is in fact FEV1 and FEV6 that are low — expiratory values weak in normative terms but missed by general-population thresholds (an FEV1 too low for an elite athlete, low against the cohort, and tracked against VST at 6, 12 and 24 months).

This architecture — raw value → multiple cross-checking → synthesis — is not peculiar to FIV. It is the grammar of the whole ecosystem.

Each index is a reading layer; none concludes alone; the diagnosis is born of their convergence or their contradiction. An index that contradicts its neighbours is as informative as one that confirms them — it points to a singularity to understand, not an error to correct.

The chain, link by link

Ventilatory function (D1) — the upstream mechanics

The first discriminant, D1, interrogates the ventilatory mechanics themselves. It splits into two questions that must never be merged.

Is the supply capped (D1.a)?

Does the inspiratory pump have the strength and the excursion to follow demand? The S-Index measures maximal dynamic inspiratory pressure (cmH₂O, best of eight manoeuvres).

It captures both power and volume excursion at once, and correlates with MIP without being equivalent to it (r = 0.74, Areias et al. 2020) — a functional measure validated for sport (Kowalski et al. 2024).

An S-Index below the elite standard immediately points to the inspiratory muscles as the likely limiter, and therefore to inspiratory muscle training as a first-rank lever.

Is the supply efficient (D1.c)?

Is the mobilised volume converted into useful gas exchange, or wasted in dead-space ventilation and ventilation/perfusion mismatch?

This is an entirely different question from the previous one: an athlete may have a powerful pump (D1.a intact) and ventilate like a leaking bag (D1.c degraded).

Here come the ICIF (inspiratory/expiratory volume balance), the IMD (mobilisation deficit relative to expiratory capacity), the VEI, which relates the square of Tv to the ventilatory cost VE × Rf — that is, how much useful volume for how much ventilatory effort — and the EqO₂, the ventilatory equivalent for oxygen, where a lower figure marks better efficiency.

[Mechanistic hypothesis] An Rf-high / Tv-low pattern increases the dead-space fraction per cycle (each breath first fills the conducting airways before reaching the alveolus) and raises the ventilatory cost at equal VE. The VST restructuring would shift work toward volume, reducing cost for equivalent useful alveolar ventilation. Evidence level: INTERMEDIATE on the HNS cohort, consistent with the CPET efficiency framework (Wasserman et al. 2012).

There is a third category,

D1.b, which must be named for the honesty of the framework: structural limitation, not remediable in the short term by training (thoracic geometry, fixed pulmonary mechanics). Distinguishing it from remediable D1.a is an essential part of the work — promising a gain where the structure does not allow it would be a fault.

Tidal volume utilisation

Having the capacity is not enough: it must be used. The TVU (Tidal Volume Utilisation) expresses the percentage of the ventilatory window actually mobilised at a given frequency — the reference tightening as Rf rises, because a high frequency mechanically compresses the volume available per cycle. The TVE (Tidal Volume Efficiency) relates Tv to body size (L/kg), filling a gap: the literature provides an absolute reference for peak Tv (Neder et al. 1999: ~2.70 ± 0.48 L in men) but no body-weight-normalised grid, even though the same volume does not have the same value for a 58 kg climber and a 78 kg rouleur.

These two indices bridge to the central VST signature: the gap between the spontaneous Rf/Tv/VE pattern and the restructured target, zone by zone of intensity. It is this gap that defines the object of training — not a number, a trajectory.

Muscular proxy NIRS (D1.c / D2) — leaving ventilation for extraction

This is the link where the chain changes nature: we move from air to muscle. Near-infrared spectroscopy (NIRS) opens a local window onto extraction and perfusion (Vogiatzis 2008/2009; Legrand et al. 2007).

The VEC (Ventilation–Extraction Coupling) relates the amplitude of O₂ extraction (ΔFeO₂) to the ventilatory cost (EqO₂): it judges whether ventilation actually translates into extraction, or spins idle. The ICP SmO₂ compares the oxygenation of a non-locomotor compartment (arm) with that of the locomotor compartment (quadriceps) — and one must be precise about what it actually measures, because the naive reading misleads.

The arm is not an inert reference. It does no locomotor work, yet its SmO₂ desaturates too, inevitably, especially above VT2: when sympathetic drive surges, delivery there falls faster than demand.

The arm is therefore a systemic probe — it reads the global state of redistribution and vasoconstriction — while the quadriceps reads a local demand that floors near maximum.

The ICP is not preferential locomotor extraction: it is the ratio between local demand and systemic state, a proxy for the shift in the local↔systemic balance. And because the quadriceps saturates its information near max, it is the kinetics of the ratio — its slope, its inflection, the rate of arm desaturation — that carries the signal, far more than its instantaneous value.

[Mechanistic hypothesis] An acceleration of arm desaturation, or a terminal inversion of the ratio, time-locked to the escalation of ventilatory work, would evoke a signature of sympathetic redistribution of respiratory origin — the work of the respiratory muscles stealing flow from the locomotors (Dempsey et al. 2006; Sheel & Romer 2012). Evidence level: the mechanism is STRONG in the literature; its individual expression remains INTERMEDIATE, and WEAK on a single marker.

This is the second discriminant, D2: is the peripheral oxidative capacity actually mobilised, or does an under-extraction persist (elevated residual quadriceps SmO₂) that caps the output independently of everything happening upstream?

Cardio-peripheral coupling

Between ventilation and extraction, the heart delivers. The EF (Efficiency Factor) reads the power-to-weight delivered per heartbeat in the steady state — a proxy for cardio-metabolic efficiency. The O₂ Pulse (VO₂ / HR) approximates stroke volume (to within the arteriovenous O₂ difference): its progressive rise to VO₂max is expected; a plateau or an early drop signals a cardiac or extraction limit, to be read jointly with the ICP SmO₂ — an O₂ Pulse that plateaus while the quadriceps are still desaturating does not say the same thing as one that plateaus when extraction is already maximal.

The point where the system becomes a line of reasoning

Everything so far is reading. Here is where profiling stops reading and starts reasoning — and where the ecosystem justifies its existence.

Consider a real and difficult phenomenon: a rider who loses control of respiratory frequency above VT1 and cannot regain it on the descent — a ventilatory hysteresis.

The Rf runs away, the Tv collapses, and the pattern stays uncoupled even as power comes back down.

The power meter sees nothing. VO₂ alone, nothing either. Two mechanisms, however, produce this same picture and call for opposite responses:

•       a respiratory-muscle metaboreflex: excessive respiratory work fatigues the diaphragm and intercostals, their metabolic afferents trigger sympathetic vasoconstriction, and the loop self-sustains;

•       a CO₂ intolerance / chemoreflex drive: ventilation is driven by CO₂ and hydrogen ions, and the runaway Rf follows the acid–base chemistry.

The first calls for respiratory muscle work and a restructuring of the pattern. The second calls for work on tolerance and on the kinetics of CO₂ clearance. Getting the mechanism wrong means training beside the point for weeks.

How to decide, with field, non-invasive means?

By coupling three axes that, separately, do not conclude: the trajectory of transcutaneous CO₂ (the only direct access to the acid–base side when the gas analyser measures O₂ only), the shift in the local↔systemic balance read by the kinetics of the ICP SmO₂, and the Rf/Tv/VE ventilatory pattern that expresses respiratory work itself.

Each of these signals carries a fragment of the answer; it is their mutual temporal alignment — which cause precedes the other, which persists on the descent — that designates the mechanism.

And here too, the defensive posture holds: none of these signals decides alone, transcutaneous CO₂ has its diffusion lag that must be accounted for, and some configurations remain ambiguous (a drive linked to the hydrogen ions of acidosis can mimic other causes).

The general mechanism of the metaboreflex is firmly established (Dempsey 2006; Sheel & Romer 2012); its attribution to this athlete, in this session, becomes robust only through convergence.

The operational protocol that formalises this coupled reading — the reference trajectories, the case grid, the decision tree — is HNS's internal method. What matters, in the frame of this article, is not the recipe: it is that a question of this finesse — which mechanism drives this rider's ventilatory runaway? — can be posed and investigated with field sensors.

That is precisely what an ecosystem permits and an isolated measurement forbids.

Synthesis: from diagnosis to priority

Read separately, these indices are only a panel of instruments. Their value is born of convergence — and the clearest example is the triple lock at VO₂max.

When three markers converge simultaneously toward their elite targets — a systemic extraction coefficient of the order of 80–85 % (Calbet et al. 2006; Lundby 2017), a sufficient fall in alveolar O₂ fraction, and a deep quadriceps desaturation — then the VO₂max output is truly expressed: the system consumes the oxygen it delivers, and the limiter sits elsewhere than in terminal extraction.

Conversely, if one of these locks fails to close, the diagnosis immediately reorients: capped ventilatory supply (D1.a), degraded efficiency (D1.c), or muscular under-extraction (D2).

The same average VO₂max can thus hide a system saturating its extraction and a system capping for want of supply — two athletes, two opposite training priorities, that only the convergence of markers distinguishes.

This is the whole point of an ecosystem rather than a list: it lets you measure each link, identify the limiter by cross-checking, and rank the training priority.

The cascade does not merely display values — like a real automotive diagnostic, it reads the whole vehicle and returns the fault code to address first. A low S-Index, isolated, proves nothing; but a low S-Index that converges with a low TVE, an inefficient VEI and a mobilisation deficit at VT1 draws a coherent upstream ventilatory limiter — and a clear, defensible, actionable priority.

Posture

Everything above rests on a strict distinction between established physiology — sourced, dated — and the mechanistic hypotheses specific to the HNS framework, flagged as such and graded by evidence level (STRONG / INTERMEDIATE / WEAK).

The cohort reference ranges used here are built on a proprietary sample (n = 80, six years of field data), unpublished, paper in preparation 2026; they are never presented as external consensus. And the whole is a functional reading, with no medical aetiological diagnosis.

This is not caution for caution's sake. It is the condition for a staff to trust a figure: knowing where it comes from, against what it is read, and what it does not say. An index you cannot place within its uncertainty is not a tool — it is an opinion in disguise.

Follow the oxygen from ventilation to the muscle, know which link to train, and know why.

That is the whole object of the VST ecosystem — not to pile up numbers, but to turn a physiological chain into a chain of decisions.

Don't believe me, just let me show.