is there any intent to have pedalling metrics as part of the platform, either in real time or in the final ride results? I know it is not necessarily important for the algorithm (though like HRV, HR decoupling it might be a useful addition?) but it might be useful for training and other situations?
2 Likes
Hi Ron,
No plans for post hoc L/R balance information at this time, since there isn’t much practical training usefulness from knowing this (short of monitoring it after an injury). That being said, we are adding the capability to view L/R balance in real time with the Android app, and may soon be adding it to the iOS app.
1 Like
Power Phase and L/R Balance may not be useful for training analysis, but viewing them in real-time to diagnose power stroke against ride position could be beneficial.
Wouldn’t it be useful for Xert to track out-of-the-saddle time versus seated? .
Since Cycling Dynamics is an open standard any pedal power meter can capture the data.
Shouldn’t a single pedal meter support seated/standing metrics as well?
Standing has distinct effects on power phase, cadence, lactate, HR, and muscles used so it stands to reason 
there’s some value in Xert tracking this data.
There is certainly a value in honing your ability to climb seated versus standing and sprinting OOTS.
Perhaps Smart workouts could incorporate OOTS drills. If you don’t remain standing you don’t get full credit similar to how Smart target watts work now. 
Or how about showing seated and standing seconds on the workout analysis graph?
When you select this metric a color coded line appears (black=seated, red=standing) so you can see the relationship between HR, cadence, and power when you zoom in on an interval.
Besides total time seated versus standing another useful stat would be max seconds you can stand during a work interval at X watts.
Great to hear that pedallling dynamics are coming, I would love to see the activity review page to also have the them as well since at that point it is just a matter of graphing etc. I may be wrong but to be able to see it in review would allow you to see if things are changing over time. Since the data is available it is just at least another pretty graph? or data point?
also when the Android app gets real time pedalling metrics will these be displayed on the workout player on the desktop? or only the Android screen?
Lots of great ideas! There are many second-order information sources to inform the training/tracking engine. Standing vs. sitting, cadence, grade all play a role in affecting MPA beyond just watts. Would be great to look into these relationships more. On my long list of important things to be working on!!
Only on the app initially. As mentioned, there will likely be more comprehensive analysis of how pedaling dynamics impacts power, i.e. would be good to go beyond just displaying the data.
thanks for the response, It is interesting stuff for sure to see how various metrics can inform the outcomes. Since Xert is a multifactorial algorithm it would be interesting how well it might do with other added input variables and to what extent they provide actionable outcomes. I think the key item that interests me in Xert is that at this point it is the only mass market option for multivariate assessment of fitness and feedback. That is its strength and from the point of understanding it is also its weakness because it is hard to understand how multiple items (which may be antagonistic) will impact a single output like TP. People are more used to single variate models.You guys probably have one of the most significant database of user info to work with to help refine your models for sure.
Thanks Ron. Multivariate models are definitely harder to comprehend. We’ve introduced “simplifcations” like Focus Duration, but as you know, that’s a big hurdle on its own.
There can be antagonistic effects, some of which I have been pondering. For example, there could very likely be a trade-off between Peak Power and Threshold Power. At very high TLs, TP can be increased but it appears that PP becomes further compromised (i.e. sees additional declines beyond just not having been trained). It would be great to look for this relationship in the data. Fortunately for us, everything gets expressed as MPA and max efforts would validate the potential antagonistic relationship (sprints become compromised, for example). By how much and what that relationship is could be found.
Our goal has always been to build a revenue model to support further research, eventually employing a team of data scientists to hunt down all this new information in the data. Still working on that but at least enjoying the journey we’ve been on so far.
2 Likes
yes it is interesting, I have some experience in an area called chemometrics where you look at multvariate inputs to predict chemical outputs. I am by no means an expert but it is a lot of fun to see a model work or to look at something and see the right answer come out for a known value. Of course the danger is what is called over fitting, using too many inputs because they add little to the model and sometimes reduce the robustness of the model. I tend to throw everything in and see which 2 or 3 variables describe the majourity of the variability… very interesting stuff. Keep digging and I hope you have success and keep the program moving forward.
1 Like