New metric for Running (GOVSS) - Initial implementation

[Ale Martinez]

An initial implementation of GOVSS was added to master (this commit, thanks Mark), it Includes 4 new base metrics:

1. LNP - Lactate Normalized Power: similar to NP/xPower for running
   
2. RTP - Running Threshold Power: based on athlete's CV
   
3. IWF - Intensity Weighting Factor (IWF=LNP/RTP), similar to IF/Relative Intensity
   
4. GOVSS - Gravity Ordered Velocity Stress Score, similar to TSS/BikeScore for running
   

plus stress metrics for PMC based on it: GOVSS Long Term Stress, GOVSS Short Term Stress, GOVSS Stress Balance, etc.

The code is based on Dr Skiba document describing the algorithm (http://www.physfarm.com/govss.pdf) with details taken from a spreadsheet implementing it back in 2006 (GOVSSCalculatorV10). I've tested it with a couple of runs, a 10k race and a 5x800x400 VO2max session, and they seem to be the ballpark with the values calculated by http://www.topofusion.com/govss.php

There are some caveats/topics on which I like to open a discussion:

1. GOVSS and PMC integration are the basic metrics, but Intensity Weighting Factor (IWF) it is also an interesting metric to have
   
2. LNP/RTP are exposed as metrics to facilitate debugging, but they could be hidden to avoid metric space pollution if considered not generally useful.
   
3. Athlete's height is estimated from weight using Stillman average formula, but it should be pulled from athlete's data
   
4. Kinetic Energy contribution calculation is disabled, the formula in reference spreadsheet doesn't seem to be right, probably it is outdated.
   
5. Aero contribution calculation is based on reference spreadsheet but it should be verified, probably it is outdated.
   
6. More testing is needed
   

1) and 2) are mostly matter of opinion and 3) seems easy to fix but, for 4) , 5) and 6) some help would be welcomed!

Thanks in advance, Ale.

[Ale Martinez]

A related running metric, xPace was added to master code, it is computed as the constant pace which sustained for the whole duration of the workout on a flat surface would give the same GOVSS. It will be available for user's testing in the next development builds. 

[Mark Liversedge]

Ale,

This is absolutely superb stuff, I cannot thank you enough for all this.

I wonder if it is worth penning a section for the website on GOVSS and Running Metrics ?

I am happy to help with proofreading etc ?

Thanks again,

Mark

[Ale Martinez]

Thanks for your words Mark, I'm happy to contribute to this project!

Yes, I think we should have a Running section summarizing the features available with some examples in the Wiki and a GOVSS chapter in the science section for 3.11 release, I will start to write something now the feature set is mostly complete.

BTW, I would like to add pace zone shading to AllPlot for running activities but I gave up, it proved to much for my nearly-zero QWT knowledge. 

[Ron George]

Hello Ale,

Is Skiba PMC the way to track GOVSS based stress long term and short term?

Does it work for both running and biking? I notice that Ive got only biking workouts populating in the Skiba PMC chart . I thought GOVSS could apply to running as well. What's the way to correct this, thanks.

-Ron

[Ale Martinez]

Hi Ron, for multisport use PMC (TriScore), the TriScore metric is computed as SwimScore for swims, BikeScore for rides and GOVSS for runs, if you want to isolate running you can select the Run filter on the left sidebar.

Cheers, Ale.

[Ron George]

Ale,

With running powermeters now available, it is about time to re-visit the model and items 4), 5) and possibly 6) in your original message.  

Runscribe has publicly stated that they will be using GOVSS power model. 

If the model starts breaking under some defined set of conditions or should not be used outside the domain it was intended for, this would be good to know.

Ron

[Ale Martinez]

El sábado, 11 de noviembre de 2017, 8:30:30 (UTC-3), Ron George escribió:

Ale,

With running powermeters now available, it is about time to re-visit the model and items 4), 5) and possibly 6) in your original message.  

Hi Ron, 4) and 5) were fixed after that message, you can see the resulting code at https://github.com/GoldenCheetah/GoldenCheetah/blob/160eb74cab712bfaa9eab64dd19310e5d42ed193/src/Metrics/GOVSS.cpp and to use the History button to see the changes. 

Runscribe has publicly stated that they will be using GOVSS power model. 

If the model starts breaking under some defined set of conditions or should not be used outside the domain it was intended for, this would be good to know.

I don't have a Running Power "Meter" but, if you want to contribute some files for testing, you are welcome.

[Ron George]

I'm not aware about how to get Runscribe files, save for asking Tim for it. 

Sharing a Stryd file with you from an important race today (see attch'd). There's a middle segment (45 minutes long) with the actual race, from about 00:38:14 to 1:23:54. It would be nice to see what GC's GOVSS implementation throws forth. I'll expect to see the instantaneous and overall average GOVSS power to be higher than Stryd reported values, since GOVSS accounts for the internal cost of running, but interested to see the magnitude of difference. 

Could you please have a look ? 

-Ron

[Ale Martinez]

Sure, which are your weight and height?

[Ron George]

64kg in running attire and shoes (transported weight). 

Height = 1.75m 

[Ale Martinez]

Using those data LNP=309 is way higher than the AP=NP=239 for the segment according to Stryd Data. NB: these value is used only in a relative (to CV power) way in GOVSS.

Removing the Power series and using Edit > Estimate Running Power gives AP=212/NP=212 which is closer and the recommend one to use if you don't have a running power meter.

[Ron George]

Ale,

LNP includes total energy cost of running as I mentioned before, therefore it is higher. I actually like that the magnitude of difference is not obscenely high. This gives a total W/kg ratio = 4.82 which I think correlates more with my effort for this 10K.

I would agree with the second statement, only if you don't have a PM..but a user must know what it is - is it external, internal, total... ?? I would assume it is total, but what is it neglecting for it to be in the low 200's? Not in front of GC now but will have to read the documentation when I get back...

Ron