I’ve had a lot of trouble recently trying to nail down some of the ideas grinding around in my brain. As demonstrated by the growing clutter of abandoned topics in my Drafts folder over the past few months.
So I thought an easier place to start would be first laying out some questions and smaller ideas, instead of trying to perfect a grand unified theory of exercise physiology, and end up not writing anything at all..
Muscle Oxygenation vs Heart Rate for Internal Training Load Monitoring
A recent small, but compelling finding in our research with Moxy muscle oxygenation was looking at the reliability & repeatability of muscle oxygen saturation (SmO2) compared to HR for a typical high intensity interval workout.
We had a group of well-trained male subjects (VO2max > 60 ml/kg/min) perform a series of 4x4min high intensity workouts with multiple Moxy sensors. One of the ways we looked at the SmO2 data was to calculate a ‘deoxygenation area’ for each interval. Where the difference between peak SmO2 and measured SmO2 during each interval were summed to arrive at a total deoxygenation area for the workout. This ends up looking like an ‘area above the curve’, as illustrated below.
This calculation is somewhat analogous to measuring external work in joules (J or kJ) for a workout, where the power is summed during each interval to calculate the total workload (W*sec = J).
The process using SmO2 gives us something like a measure of total internal workload at the muscle under investigation (arb units: SmO2*sec). We compared this metric to time above 90% HRmax as a measurement of total internal training load.
‘Deoxygenation area’ (blue) and time >90% HRmax (red) calculated for a representative workout, calculated in WKO5
We aren’t saying these are interchangeable measurements, nor are we making any claims about what this ‘deoxygenation area’ represents mechanistically. The promising finding is that the relationship between internal and external workload appears to be more reliable for deoxygenation area than for time >90% HRmax. I’ll show the stats below, but the concepts are worth considering first.
The chart above illustrates how the deoxygenation area (blue) is more consistently proportional to power (yellow), whereas HR shows a typical drift through the workout. Time >90% HRmax likely underestimates both the internal and external training load of this workout.
I’ve argued in the past that time >90% HRmax is a decent, but far from perfect proxy for time >90% VO2max. However my opinion has gradually been drifting away from accepting this relationship as a useful comparison.
In this case we did not measure VO2, but the workouts were prescribed at 110% of power at each athlete’s second ventilatory threshold (VT2 or RCP). This intensity should be similar to a typical 4x4min ‘VO2max’ workout.
As a brief aside into my current thinking, HR is simply too variable to be a reliable metric for internal training load. Both the strength and weakness of HR is that as a systemic internal measurement it is sensitive to many internal and external factors.
Power is exclusively a measurement of external work and doesn’t tell us how the body is producing that power. There are many helpful models that can estimate internal intensity from external workload (a distinction for another time), but I’ve written previously about some issues I have with over-extrapolating metabolic information from power.
I still believe RPE – relative perceived exertion – is a better metric for simple tracking of internal workload. Hard workouts should be hard, and easy workouts should be easy. Your brain knows how to take care of the details.
Everyone loves looking at images of numbers, so let’s look at some Excel charts full of numbers! I’ll try to be brief going over the relevant stats, but I think the take-home is straightforward enough here.
The interesting numbers to look at here are the coefficients of variation (CV). This tells us how much each measurement varies between workouts, as a percent of
the mean value (top row).
This is a single representative subject. ie, the subject with the best numbers
to show the trend we’re seeing 🙂
The massive difference between the variability of Time >90% HRmax (22%) and the deoxygenation area factors for both the right and left VL (6-7%) suggest that whatever this NIRS-derived metric is telling us, it’s more closely related to external workload across workouts (4%) for this subject.
These are the same stats run on the full group (9 subjects). Again I think the most relevant comparison here is how closely the variability of the deoxygenation factors follows that of the external workload, compared to time >90% HRmax.
Remember, this is a moderately heterogenous group so we would expect some variability in work (15%) based on differences in individual power targets. The variability of the oxygenation factors (16-23%) is no greater than what might be expected between individuals, whereas time >90% HRmax (46%) has nearly 3x the variability!
Disclaimer: Stats are not my strong suit. I barely know what I’m talking about here, which is why I wanted to show the data themselves. Feel free to call me out if I’ve interpreted these data inappropriately.
EDIT: A good question posted below got me to look at how power, RPE, HR, and oxygenation area were related. More detail below, but briefly:
- The subjects reported almost universally that the field trial was lower RPE than the lab trials, despite power actually being a few percent higher on average.
- Time above 90% HRmax tended to be lower in the field, although not significantly different than in the lab (the variability was high overall, as above).
- Oxygenation area at the VL tended to be slightly higher in the lab, meaning the work-oxygenation factor, which reflects the power output per unit of deoxygenation, was significantly higher (more “efficient?”) in the field. Where power was slightly higher and RPE was lower!
Further Implications of a Work:Oxygenation Factor
This is another promising application for using Moxy and other portable NIRS sensors for tracking internal training load. I’ve encountered some other papers looking at similar ‘area under the curve’ calculations for NIRS measurements, sometimes with mixed results. And similar findings for reliability of NIRS over HR. Such as in trail running where NIRS measurements were found to give a better indication of running intensity than HR across varied terrain (Born et al, 2017). I would imagine this finding would translate well to cycling races where intensity is even more highly variable.
This work-oxygenation factor gives us a potential tool for comparing how much external work has been performed for a given level of local deoxygenation. Importantly this can be calculated at various workloads and over various durations, starting to build a more detailed power-duration-oxygenation profile for an individual athlete.
The final units of this measurement can be derived independent of time as W/SmO2 (or W/HHb for different NIRS devices) and may reveal a proxy of local metabolic efficiency. In plain language the work-oxygenation factor describes “how many watts are produced per unit of oxygen de-saturated from a set baseline, at the target muscle under investigation.”
I think this metric will be most useful in tracking changes within an individual athlete over time, rather than comparing efficiency between individuals. Looking at how this value may change over time with training might reveal something more about how physiology has responded to that training.
This is actually completely different NIRS data, just to illustrate what tracking work-oxygenation factor might look like. And to raise the question of what these data might suggest if they were real, about how the athlete has adapted to training over the previous months.
For example, a greater work-deoxygenation factor would be found in the case of either higher power output for the same level of deoxygenation, or less deoxygenation for the same power output. Higher isn’t necessarily better, so what could this change be telling us?
Both situations might indicate a greater relative O2 delivery capacity to the working muscle (and if this statement sounds like it’s coming out of nowhere, wait for a forthcoming article going into a bit more detail on what information can be interpreted from NIRS). The former case suggests that the muscle was able to raise O2 extraction to match the increase in delivery, resulting in greater power output.
The latter case might suggest that while O2 delivery has improved, the local muscle has not – was not able to, or has not needed to – increase extraction to match delivery, resulting in power remaining the same. This could indicate that the athlete might be extraction-limited at the local tissue, or that the net power output is coming from somewhere else.
This has implications for assessing physiological limiters to performance and what training prescription this athlete might respond better to.
Questions to Ponder
What kinds of training could a coach prescribe an athlete if she suspected muscle oxygen extraction was relatively more limited than oxygen delivery?
What about oxidative capacity of other muscle groups contributing to power output? Not all watts come from vastus lateralis. How can we integrate muscle oxygenation measurements from multiple sites to gain better information on metabolic recruitment patterns and contribution to power output?
That’s a topic that raises many questions. Which I will hopefully write about soon(ish).