Hi David-
Would it be possible to pimp the activity tags beyond the fields that Strava offers?
For example I am thinking of something like hydration/nutrition tracking.
I did this in the summer and I was really surprised how much fluid is needed in different conditions.
The data could even be displayed in one of those good looking charts
Temperature is already in the FIT file. Weight before/after and fluid uptake would be a user input.
The chart could plot average fluid loss per hour vs temperature (or even more advanced wet bulb temperature to account for humidity).
What do you think?
I would need to add custom fields to an activity for that. Its quite an esoteric use case but is now on the todo list.
I am surprised you find the use case esoteric I would argue that fuelling=hydration+food. For longer events dialling the two in is probably even essential to get to the end or perform well.
I would add another suggestion: to track and even predict substrate demand:
Total energy expenditure already comes from the powermeter. The percentage of carbohydrates (roughly) is about 50% at fatmax, or lets say middle of CogganZone 2 and curve linear rises to 100% at FTP or above.
With, say the already computed 30sec averaging, intervalls.icu could easily compute the estimated carbs/fat burned for a session. This could even be part of the workout builder as a prediction or datapoint how to correctly fuel for workouts or events.
For people who have more exact numbers from metbolic testing the â50%-pointâ on the MMP could be made user adjustable.
I think this would be really handy for a lot of triathletes or 2hrs+ cycling events
I have had quite a few requests for calculating carbs vs fat burned in activities. The question is how much variation exists between athletes i.e. how well does the â50% carbs at middle of Z2, linear to 100% carbs at FTPâ relationship hold up? I wouldnât want to implement this if the results are mostly bogus due to athlete variations.
Donât go down that rabbit hole. Others have tried (Xert for instance) and it incites more discussion.
Fuelling strategy is not generic, beyond âyou need to eat and drinkâ, whereas even that leads to discussion, i.e. not for sub one hour workouts.
I would argue that this method is as accurate as training zones, and definitely a lot more accurate compared to what sport watches/trackers compute. By adjusting the â50%-sliderâ you could make it very accurate (for people who have done a metabolic profile). The relationship VO2/fat/carbs is not exactly linear, but imho thatâs close enough. You could of course use more elaborate models (i have a paper on how to do that if you are interested). To simulate you could i.e. use the equations of A E Jeukendrup ( ie. âMeasurement of substrate oxidation during exercise by means of gas exchange measurementsâ) and simply enter different RE/RQ values.
As @Cyclopaat hints, if you dig deeper it becomes a rabbit hole. I volunteered a lot for lab work and metabolic testing in the human performance lab at my sports university. If you look at the raw data the numbers bump around a lot (and need to be averaged) and can even be different on different days, depending on prior exercise, diet, time of day⌠and so on. Thatâs why the HPL guys usually have a feeding protocol that calls for âno more carbs in the last 3 hrs leading into the testâ, but that is basically the exact opposite of what athletes do in ie. a race⌠Thatâs also why INSCYD goes with calculated values derived solely from Vo2max/Vlamax⌠and to sell their remote test
In a nut-shell: If the zones match up well with the underlying physiology, the calculated substrate utilisation will too. At the very least you would have a best-case/worst-case data point, which is a lot better than nothing and. AFAIK non of the other platforms have anything like that atm⌠And this project could really shine once more with innovation
Tx Could you please post a link to the âmore elaborateâ models paper. I have added this to the todo list. Happy to implement if I can back the implementation with a reference paper.
This one has the âbest validatedâ equations (Jeukendrup):
This is one paper that validates above equations:
This one ist for math nerds, that want an equation for the âmost exactâ curve fit of their personal data: