Modeling fatigue accumulated during a ride

[Diarmuid]

Hi,

I've been using GC for a year or so and find it extremely useful for reviewing rides and planing my training. Recently, based on happenings in both races and long hilly sportives, I am wondering if there is any way in GC to model the accumulation of fatigue during a ride that is not basedl on W' but on the fatigue-inducing effects of riding below Critical Power (on average), such that W'bal depletion is not a factor.

Two examples:

1. A 90-minute race with some intervals above CP, but mostly at or just below CP. I go to sprint at the end but do not have the same energy available as if I was fresh, although my W'bal is fully replenished.

2. A long hilly sportive at a high pace where I climb multiple hills at well above CP and recover on the downhills and flats. At some point my ability to take on another hill at > CP goes out the window and the fatigue in the legs is very evident.

I know the theory is that I should be fully recharged and able to "go again" once W'bal is replenished, but this doesn't happen indefinitely in practice, at least not for me. This is the case even when I make sure to take on sufficient water and food.

This seems a bit like "real-time ATL" in concept, in that the fatigue-inducing effects of riding do not just happen when the ride is over, but will affect you as you go. I'm primarily interested in this so that I can look back on rides where I hit "inability to deliver" and see at what points this fatigue accumulation occurred.

Many thanks in advance for any light that people can shine on this. I''ve searched this forum and elsewhere and found no material that addresses this area.

Cheers,
Diarmuid.

[Mark Liversedge]

On Tuesday, 5 July 2016 15:08:22 UTC+1, Diarmuid wrote:

Two examples:

1. A 90-minute race with some intervals above CP, but mostly at or just below CP. I go to sprint at the end but do not have the same energy available as if I was fresh, although my W'bal is fully replenished.
 

2. A long hilly sportive at a high pace where I climb multiple hills at well above CP and recover on the downhills and flats. At some point my ability to take on another hill at > CP goes out the window and the fatigue in the legs is very evident.

 

Many believe that CP drops off over time as you fatigue. This would cause you to use more W' to just maintain CP over longer durations (>30-40 mins) and possibly slow down recovery.

It would be quite easy to experiment with a 'fatigue' rate .. its something that would be fun to model with user data or R.

There isn't much in the literature about this, although some have looked at how variability can impact fatigue.

Mark

 

[Stefan]

You could approach the issue through old-fashioned post-race analysis (identify matches burned and so). Butit requires experience. I took my until my second and third year of racing with a powermeter before I could make proper use of my data. I mainly race alpine MTB marathons.

Here is a very simple example, same race, one year apart:

Here I killed myself on the first climb. See the result later in the race. A steady decline, I blew up.




The following year I was more cautious, I took it "easy" on the first climb. Furthermore, from other racing and training data I could devise a pacing strategy for all the climbs. 30 min faster.



However, as already alluded to above, it takes some experience. Work with your data, understand your data. Especially the racing data. Therefore, I believe that it is very important that younger, ambitious athletes start as soon as possible to train with power. Simply to build that important database.

[Diarmuid]

Thanks folks for the responses. I agree that there's no substitute for reviewing and understanding one's own data to determine how to change behaviour next time to do better. Still, the engineer in me wants to understand and model what's going on. Can anyone give me some pointers as to how to model a concept such as fatigue rate in GC and apply it to scenarios to predict degraded performance on successive intervals?

As a side note, I'm fascinated that there is little analysis of this topic in the literature - given that it is such a readily observed phenomenon. For example, the long standing idea of having only so many "matches" to be burnt in a race would not be solely down to W'bal depletion, since there are many cases where people have been racing along below CP, but when it comes to the crunch, the earlier efforts find them out in the end.

Diarmuid.

On Tuesday, 5 July 2016 15:08:22 UTC+1, Diarmuid wrote:

[Steve Mansfield]

My own experience is that if I deplete W' on a ride eg giving a hill a nudge, going too fast in a race etc, I'll be unable to maintain anything substantial over CP for the rest of the ride, even though supposedly the W' model shows me replenished after easing off etc. Further I'll probably be toast the day after as well!

[Jean Div]

My uneducated 2c...

This will be a LOT harder than modelling a smaller and somewhat more finite W' fatigue - 

1. Nutrition and level of athlete carb-dependence - the aerobic engine of a fat-adapted athlete is nearly infinite for the purposes of 1 ride below threshold, the anaerobic engine of all athletes is a lot more finite and predictable. Further complicating, the aerobic engine of a carb-dependant athlete is going to last around 2-3 hours before they need some kind of sugar to keep going or performance will fall. How do you isolate this? Assume 'perfect nutrition' either way?

2. Over the durations in question, environmental factors will come into play in a big way. You would probably need to account for this.

3. Small differences between individuals over long periods will be compounded into much bigger numbers - 5% variance on an all out 10m W' effort = 30s. 5% difference on a 14 hour aerobic effort is a much bigger number.

..or am I way off?

[Nathan Townsend]

We're considering running a project on this stuff.  I also met Jerzy Zoladz in Krakow last week for beer and a pizza (see last author on the recent review below) and we discussed the idea further or doing some heavy mechanstic CP based research in the next couple of years (eg: using P31 NMR to look at in vivo changes in [AMP], [ADP], [PCr], [Pi] and [H+] during long duration sub-CP intensity exercise.

Skeletal muscle fatigue and decreased efficiency: two sides of the same coin?

http://www.ncbi.nlm.nih.gov/pubmed/25688762

The long and short of it, is that the assumption that neither CP or W' do not change when long duration sub-CP exercise occurs must be challenged.  Jerzy says there is no doubt that peripheral fatigue is occurring, but there is a severe lack of data on exactly what is going on, and for example the relationship between glycogen depletion and metabolic stability (see below abstract for a correct definition of "metabolic stability")...

Slow VO₂ kinetics during moderate-intensity exercise as markers of lower metabolic stability and lower exercise tolerance

http://www.ncbi.nlm.nih.gov/pubmed/20821336

[Nathan Townsend]

On Tuesday, 19 July 2016 13:51:23 UTC+3, Nathan Townsend wrote:

The long and short of it, is that the assumption that neither CP or W' do not change when long duration sub-CP exercise occurs must be challenged.  

Just a little extra on this bit..... everyone who uses GC regularly (and has done some truly max efforts out to 60-90min duration) knows that the CP + W' predicted power is higher than what can really be sustained.  So it's clear that one or the other, or both of those parameters must be declining.  The challenge is to figure out which one, or what the relative contribution is.  This is important for work-balance modeling of course because both parameters affect the calculation of W'bal at any moment.

 

[mickebergma...@gmail.com]

Last weekend I did a 24 hour race in a 4 guy team.
In praxis we did about 1 hour biking and then changed rider so about 3 hours rest.
The track (Nürburgring) is very hilly with about 500 hm per 26 km lap.

I knew the uphills will be tough and did a lot of thinking about W'. I didn't get to analyze the data before the end of the race but now I did.

The feeling of fatique correlates quite nicely with W' the first 3 laps. Then obviously something changes radically as W' is way off.
Threre's a ~20 min climb and the feeling was every lap that I could not have done it faster.
First lap W' got down to -12.2
Next -8.0
#3 -7.2
#4 -2.6
#5 +7.1
#6 +11.9
#7 +9.7

NP and heart rate also go down during the race. First lap NP 272 and HR 158. Last lap NP 229 and HR 140.
The change could be due to clycogen depletion or overall fatique. Don't know but I find it very interesting. Maybe FTP should be changed over the laps.
Can I change the parameters for a lap and get it recalculated?

[mickebergma...@gmail.com]

To make it more clear I made a spreadsheet of the 7 laps. Let's hope the formatting looks OK after the "send click"

Lap	Dist	Elev Gain	Avg Speed	Max Speed	Avg HR	Max HR	Avg Power	NP	Calories	Training Effect		W	W'	W' min
1	26.1	467	29.9	92.5	158	170	229	273	716	5		713	142	-12.3
2	25.9	482	28.7	96.4	154	169	210	257	681	4.6		680	111	-8.2
3	26.0	484	29.4	87.2	151	166	209	264	664	4.2		666	117	-7.2
4	25.9	489	27.3	84.7	142	160	202	250	687	3.3		687	96	-2.6
5	25.9	493	27.2	89.5	140	154	200	244	682	3.4		667	81	7.1
6	26.1	497	26.7	92.2	139	158	179	230	628	3.2		624	71	11.9
7	23.0	455	25.7	88.6	140	158	189	229	603	3.3		603	60	9.6

[Mark Liversedge]

It is interesting to consider W' as accumulated fatigue and possibly impacting CP over time.

It might be fun to try and model this post-hoc and fit to workouts, where we know there is an all out effort towards the end of a ride. For example, on my favorite 2hr loop there is a hill as I approach home that I generally attack hard before coasting home.

Will play with a chart to try and do this.

Mark

[mickebergma...@gmail.com]

I made a (very quick 'n' dirty) model for glycogen depletion.
It still doesn't make any sense but I must have some numbers wrong.

I know the calorie consumption per stint. (about 700). Convert to carbs when I know I'm using almost 100% carbs at that intensity. 179 g. (one stint rounded to one hour for simplicity)

Add the intake of 60 g carbs per hour.

Estimate basic consumption during resting hous to 1800 kcal/day or 19 g carbs per hour. Estimate 35% of that being from carbs.

With this the glycogen is increasing and not decreasing. Any idea whre I'm thinging wrong?
Already from the siplified calc usage 180 g/per stint and intake 4x60 g we see something's not adding up.

[Christian Wiedmann]

The most obvious issue is that carbohydrate intake is not the same as glycogen replenishment. There's a limit to how quickly your body can turn carbohydrates back into glycogen. Here's the first article I found after a quick google search: http://thesportjournal.org/article/glycogen-replenishment-after-exhaustive-exercise/

[Jean Div]

Really interesting!

I did one of these earlier this year, although our laps were a bit shorter and therefor more intense at just over 30m. That also meant that we had only 1hr30 rest, but we doubled up once after midnight so that we could all get a small catnap. Was curious about the effect of sleep deprivation on the sets after the sun rose again.. cos we were all REALLY shattered. Another example of something that I don't think that we'll be able to account for in a general and predictable way on a fatigue index as described above.

[Ciaran O'Grady]

Good questions. I have studied the decline in gross efficiency during sub-maximal prolonged cycling and found data to support that the increase in O2 consumption is likely due to a mixture of increased ATP cost of power production and increased O2 cost of ATP resynthesis. There are a lot more factors that you'd need to account for as well, as mentioned above; environmental factors, nutritional status and upkeep etc. Daniel Green has some really interesting work on this.

Above paper:

http://onlinelibrary.wiley.com/doi/10.1111/sms.12673/full (researchgate link: http://bit.ly/2aIDCXh)

Ciaran.

[mickebergma...@gmail.com]

Thanks for the input.

I know the body can use at best 60-90 g of carbs per hour so I happened to use 50 g which is exactly the number in the report Christian linked.
Maybe the issue isn't glycogen but overall fatigue.

My 52-58 min stints were not exactly sub maximal. IF started at 1.065 and ended at 0.89. From feeling they were all as hard.

[mickebergma...@gmail.com]

One thing I didn't consider is that during maximum effort the carb intake is not that high.
Lowering it to 20 g for the biking time already makes the clycogen level go down.
Not enough to make sense though.

[Nathan Townsend]

Very nice work Ciaran!

We're planning a study on something similar at the moment but more focused directly on changes to CP+W' with prolonged exercise below CP.  I can definitely use these results as part of my physiological rationale for our hypothesis.  The Work-balance model at present doesn't account for these changes.  Daniel Keir's work suggests that it is the VO2 (as opposed to power output) that is more tightly linked to a given "threshold" intensity.  So clearly if VO2 can be maintained, then in the face of declining efficiency, power must decrease. 

We plan on doing VO2 kinetics of course and some EMG measurements using a method published by Anna Coehlo (Harry Rossiter's group) last year, but I'm particularly interested in the NIRS based mVO2 technique. Will have to read up about it. Can I email you offline if I have any questions?

[Nathan Townsend]

just a quick note, here, those CHO rates refer to exogenous utilization ie: that is about the max rate that consumed CHO can be oxidised.  It doesn't refer to glycogen utilization which occurs at higher rates per kg of muscle.

Regardless, your statement Maybe the issue isn't glycogen but overall fatigue. is on the money IMO.  Glycogen depletion is certainly important, but it isn't everything.  There are lots of other things happening. I'm moving more towards the opinion that CNS regulation is a key player, not Noakes magical feedforward central governor calculator, but rather a feedback mechanism in which afferent signals from all over the body (and likely some within the CNS itself) act to downregulate central motor drive.  Therefore, it requires a higher level of conscious effort to recruit the same amount of muscle following prolonged fatiguing activity compared with the baseline unfatigued state. For submax activity this can be achieved at greater perceived effort, but for maximal sprint efforts we just cannot generate as much voluntary muscle activation, so peak power decreases (btw I'm not making this up, I'll be presenting some data which shows this effect later in the year at a conference in Canada).

[Steve Mansfield]

So then the question becomes "what is overall fatigue"? We would need to look for another mechanism. How easy / hard is it to more accurately track things like electrolyte balance?

http://www.ncbi.nlm.nih.gov/pubmed/7854827

[Nathan Townsend]

On Wednesday, 10 August 2016 23:47:32 UTC+3, Steve Mansfield wrote:

So then the question becomes "what is overall fatigue"? We would need to look for another mechanism. How easy / hard is it to more accurately track things like electrolyte balance?

http://www.ncbi.nlm.nih.gov/pubmed/7854827

Indeed that is the question, but I don't think the answer lies in trying to add a parameter for every mechanistic variable (eg: gradual changes in electrolyte balance) which makes for an ever increasing level of complexity model.  Fatigue and performance are essentially two sides of the same coin, and these are outcome variables, not explanatory variables.  This is what W'bal is trying to acheive though, not necessarily explain the precise mechanism. For example, I don't think it is necessary (for the purposes of work-balance modeling) to attempt to model glycogen depletion, because it isn't the only thing that leads to long duration fatigue.  Just try to model the fatigue/performance itself under some "standard" or optimal condition (which would need to be defined... eg: temp, humidity, CHO and fluid intake). Deviation from the model could be the result of many variables.  So rather than try to model ALL of the possible variables that could affect performance (which is unknowable anyway) so that your model stays perfectly accurate all the time, you just understand that something (without knowing exactly what) is causing deviation.  

Something that I\ve found when comparing the differential W'bal model to the integral model (using LMM stats) is that there are minimal deviations between models for computed W'bal when minimal fatigue exists (eg: W'bal stays above 60-70%), but when W'bal starts dropping and exercise bout, either constant load or interval, approaches a maximal effort, then the model's differ significantly. Anyone using W'bal modeling at the moment in GC would note this already.  So my gut feeling is that we might expect something similar when we compute W'bal with an "optimised" model that takes into account slow fatigue.  ie: if something causes deviation from the model like illness, it won't be very apparent in the low fatigue W'bal zone of 75-100% unless you also tag your RPE thoughout a session at very submaximal levels. Where the extraneous effect will become very obvious though is whenever a maximal effort is being produced.  Then you know for sure, your lack of CHO, sleep, recovery from previous training, illness, the ambient temp (its hot today) or whatever, is making a serious impact on your ability to drop W'bal to zero later on in some highly random workout.  

[lesmce...@yahoo.com]

Just a quick sidebar question...any idea why the amount of Work is trending down like other performance related variables as the ride progresses? Given that the distance and the route are both constant, should not the amount of Work each lap be consistent? 713 to 603 is quite a variance. The only way I can think of to get this result would be to "pedal less / coast more" each successive lap (except lap 4, the anomaly).

[mickebergma...@gmail.com]

Forgot this subject for a while being busy working.
Checking the coasting per lap is easy.
9 min - 11 - 10.4 - 11 - 10 - 11.8 - 9.5

The last lap is shorter (ends at the finish line and not our pit) so this makes a difference in the energy. Rest are all the same.

[Daniele Marrama Saccente]

It is not trivial if it is stated that: Fatigue is Multifactorial
I propose useful slide prof. D'Andrea Gabriele (Professor of Biochemistry at the University of L'Aquila) presented the Endurance Conference 2016 which was held in L'Aquila - www.ec2016.it

Fatigue: Causes and Mechanisms
http://www.ec2016.it/wp-content/uploads/2016/05/Gabriele-DAndrea.pdf

Daniele

[mickebergma...@gmail.com]

Thanks Daniele. Interesting reading.

As I got curious I opened a file from this summer. 300 km in 9:52.
In this I don't see any pattern of fatique. Or not strong anyway.
Most likely the main difference is lower effort (average 170 W, NP 205) which was more or less sutainable over the entire race.