Some interesting data mining I've been doing lately
So not too long ago
@avondale and myself were having a discussion about calorie deficit and losing weight. That conversation led me to opening up MyFitnessPal and finding some old weight data that I hadn't been able to figure out how to access in prior attempts. With this extra weight data, it enabled me to combine that information with other information I track to see if I could find any trends in the datasets. So let's dive in.
Weight Loss graph from 2012 to Current
View attachment 433294
With that access to MFP and Garmin, I was able to put together most of all of the recorded weight data I've got.
Garmin VO2max estimate value over time
View attachment 433296
I then went into Garmin and found VO2max data from February 2017 to present. I decided not to include any Garmin VO2max data before Feb 2017 because I used to believe my maxHR was in the 190s (using 220-age). Since maxHR plays a significant role in the Garmin VO2max estimation, all of the data prior to making that change is completely erroneous. All of the data is using a maxHR of either 178 or 175 as I do believe it has dropped ever so slightly in the past few years.
I graphed this over time and blocked out the different training cycles. Arguably my peak performance of my life thus far was the Dopey 2018 races. As you can see at the end of that training cycle my Garmin VO2max peaked at a value of 60. And during the races themselves were in and around 58-59. Conversely, when I ran Lakefront in 2017 (my first sub-3 hr marathon attempt and still PR of 3:14), I peaked at 57 and dropped to about 54 on race day.
You can also see the devastating effect of time off from running on the Garmin VO2max value as the period of the stress fracture and ankle injuries show precipitous drops in the Garmin VO2max value.
Lastly, you can see a nice curve during both of the run+bike+strength training cycles. It still remains to be seen whether this methodology will yield a peak running VO2max similar to Dopey 2018 training, but it's still something interesting to follow.
Something important to note before moving forward. VO2max is not the end all be all. If one person as an estimated VO2max of 59 and a different person a 55, then that won't be the sole decider on who wins on race day. Plenty of other variables. However, with that being said, if the same person has a VO2max of 59 and 55 on different days (or training cycles), then that single person is more likely to be faster when they have a higher value. So comparing against others is not completely translatable. But comparing within one's own self can be.
The other thing to keep in mind is that the Garmin VO2max estimation is based on the relationship between %HR and pace during runs. As we know, T+D plays a role in pace. The higher the T+D the slower you'll run at the same relative HR. So a lower or lowering VO2max does not always mean a loss of fitness, but rather a more "in the moment based on recent runs elevation/T+D profiles" assessment of the relationship between HR and pace.
The relationship of Garmin VO2max estimate value AND resting HR over time
View attachment 433300
So I plotted resting HR (in blue) with the Garmin VO2max estimation (in black) over time and with training cycles broken down. It's something I've always intuitively felt was true, but the data set reinforces the idea that when my resting HR is at its lowest is when my running fitness is at its highest. You can see when I have massive downspikes in resting HR (like 4/25/17 area, Dopey 2018 area, pre-Chicago ankle injury area, and TrainerRoad areas) is when my Garmin VO2max estimation peaks the highest. My off the ball guess is that this is when my heart is becoming more efficient even in my daily life. The stroke volume has increased per beat, and thus my heart needs to beat less. And this carries over into running because to transport the same amount of blood/oxygen/nutrients throughout my body takes less effort (less beats) by my HR and thus enables me to run faster with less effort.
When I suffer an injury (or am run training less) you can see my resting HR goes up. My body detrains and my stroke volume presumably goes down. Thus, requiring my body to beat more often in order to transport the same total volume of blood around the body on a per minute basis. This is also evident during the exclusive training months of only TrainerRoad+80DO where you can see my restingHR is at some of its highest values in the entire time period even though rigorous training was occurring. This can be explained by saying that cycling does not yield the exact same stroke volume efficiency changes that run training yields. But as soon as run training is introduced, the resting HR starts to plummet.
Now this isn't novel information. It's a known fact that run training will increase the size of the HR (I believe the left ventricular) and additionally increase the stroke volume for each individual beat. But to see it in possibly through inference in raw data is pretty cool.
Here's the thing though. Garmin VO2max is great for a measure of performance. Have a higher VO2max and you'll likely be faster. But that doesn't necessarily mean you're as fit/efficient. A component of the VO2max value is your current body weight. So being at a higher body weight requires a more efficient system to be at the same VO2max estimation value. Lower weight and the same efficient system means your Garmin VO2max estimation will go up. To remove body weight from the equation, we move to Absolute VO2max which is a measure of the total amount of oxygen metabolized per minute in the absence of body weight data:
Absolute VO2max (in Liters) = (weight in kg X VO2max) / 1000
Absolute VO2max estimate value over time
View attachment 433316
So over the last 3 years, my body weight has oscillated from as high as 181 to as low as 157. During most of 2016-2018, my body weight was between 157-165 pounds. I raced Lakefront 2017 at around 167 pounds and gave that as a possible reason for my disappointing to me result (although there certainly other factors). So from October 2017 to Jan 2018, I dropped 10 pounds down to 157 and raced really well at Dopey 2018. As from the last Garmin VO2max estimation graph we saw my peak performance value of 60 coincided with that particular training cycle. So the question stands, was it the body weight loss from 167 down to 157 that yielded the higher VO2max value, or was there an increase in efficiency/fitness as well?
This is where I think Absolute VO2max can step in and help answer that question. Since we've now removed weight from the equation. The body weight is limited during the Lakefront 2017 training cycle (primarily because I was avoiding stepping on the scale and was using the mindset of "if I feel fast/healthy, then I'm at the right weight"). But right in the middle of the training cycle, you can see an Absolute VO2max value of 4.19 and at the end a value of 4.24. Looking back at the Garmin VO2max graph shows that the 4.19 occurred during peak VO2max values during that training cycle. But when the end of the cycle came the VO2max value fell, but the Absolute VO2max value actually rose. So more fit, but not as fast.
Compare that to the Dopey 2018 training cycle where I was consistently trying to lose weight (and did drop from 167 to 157 over the course of 2.5 months). Is the increase in VO2max estimation solely a response of dropping weight? To a degree yes, but it's also evident that there were fitness/efficiency gains made during that same time. The drop in body weight is not the sole reason for the dramatic rise in VO2max as evidence by several values found higher than 4.24 during the Dopey 2018 training cycle. In fact, the highest recorded value occurred when the Garmin VO2max hit 60, which also coincided with the lowest body weight of 157, but also showed the highest absolute VO2max value of 4.42.
Again, the stress fracture yielded a precipitous drop.
Now here's the second reason I wanted to put this graph together. During when I was primarily running, my body weight was between 157-165 pounds. But during these TrainerRoad training cycles, my body weight has been anywhere between 170-181 pounds. So the question became, am I as fit/efficient as I was in my heavy running days, but my weight is the reason that my Garmin VO2max is lower right now? Or said another way, is my current run+bike+strength training yielding similar results to my primarily run training in past years? The answer appears to be yes. During the last TrainerRoad training cycle, you can see that my absolute VO2max was staying all but higher than almost everything during Dopey 2018 training. And during this latest training cycle the absolute VO2max value has been steadily climbing. So I'm as (if not more) efficient/fit than I have ever been using this current training methodology.
Now this clearly begs a question, would I be just as efficient/fit using this current methodology if I were to lose more weight? Or is my current weight the reason I'm able to maintain this current training volume and efficiency/fitness values? I don't think I've got the answer to this one. But I think I'll continue to try and live a healthy lifestyle and see where the next 18 weeks takes me.
Next, I wanted to see if there was a relationship between how much running duration I do in a week and the Garmin VO2max estimations or absolute VO2max values.
Running Duration (in min per week) over time
View attachment 433328
So this graph just simply represents the amount of run training I did in a single week. It's the cumulative duration in minutes per week. It shows time off from injuries, tapering, reintroduction of running, essentially running the whole gambit of a training cycle. Goes to show that during most of my run training cycles I was between 400-525 minutes per week on a regular basis (6.6-8.75 hours). Yet these days, I've been doing about 150-250 min of running per week (2.5-4.1 hours).
The relationship of Garmin VO2max estimate value AND Running Duration over time
View attachment 433332
So I wouldn't necessarily say that the running duration and Garmin VO2max value peaks follow any certain trend. In the case of the end of Daniels 2017 training and Dopey 2018 training it would certainly seem like when the run duration dropped late in the training the Garmin VO2max value peaked upwards. But I'm not seeing any type of equivalency that says run X hours per week yields Y Garmin VO2max value.
The area I was more interested in with this graph was the more recent training. Did I ever have similar Garmin VO2max values on such low run training? Granted, there is other training currently occurring, but the idea was to try and see if that cycling+strength was showing up in this graph. Let's use a Garmin VO2max value of 56. During primarily run training cycles, it was taking around 450 min (7.5 hours) of run training to usually at least have a Garmin VO2max value of 56. But during the three periods of time where I was doing structured bike training + running (last three weeks of Stress Fracture during 2018 and the two TrainerRoad cycles) you can see my running duration was in the 250-270 min (4.1 hours) of run training to usually at least have a Garmin VO2max value of 56. So what does that say to me, a reduced run training volume supplemented with structured bike/strength training can yield a similar Garmin VO2max estimated value. It would appear that a reduction of about 40% run volume was around the same value. Granted during the TrainerRoad training cycle, the replacement has not been 1:1. Rather my normal run training volume (which was all the training back then) dropped from 6.6-8.75 hours, but the total time spent training has increased during TrainerRoad to 12-16 hours. So a reduction of 40% run volume needs to be accounted for by a near doubling of overall time spent training to yield a similar Garmin VO2max estimation.
The relationship of Absolute VO2max estimate value AND Running Duration over time
View attachment 433335
For the comparison of Running duration on a per week basis compared to Absolute VO2max I see little relationship during the primary run training cycles. The volume was roughly the same from 2016 through Dopey 2018 and yet a general increase in fitness/efficiency was seen. So running volume does not seem to be the primary reason for any increases here. However, when running stops or is greatly reduced in the absence of other training, you can see the Absolute VO2max goes down. Additionally, during the 19 week off period from running when I was doing exclusively bike+strength training, the absolute VO2max value fell roughly to the equivalent of running levels around right before Daniels 5l/10k 2017 training started (right after Dopey 2017). But as soon as some run training was reintroduced during the latest TrainerRoad training cycles the absolute VO2max value rose quite sharply again. Again, even though the total volume of running is down, the absolute VO2max value suggests I'm in a really good place from a running fitness/efficiency place. Maybe not as much performance (as seen by Garmin VO2max), but definitely from a fitness place.
Lastly, I tried to view training from a "activity calories" aspect to see if there was any relationship between the Garmin VO2max or absolute VO2max changes and changes in activity calories.
Activity Calories (per week) over time
View attachment 433336
About the only thing I glean from this is, not all exercises are created equal when it comes to my Garmin's estimation of calories burned on a per discipline basis (strength, run, or bike). Because I'm putting in about double the top end time spent training from my running days, but have only seen a roughly 50% increase in activity calories (6000 vs 9000 calories even though 8 hrs vs 16 hrs of training).
The relationship of Garmin VO2max estimate value AND Activity Calories over time
View attachment 433337
Not really seeing anything of significance. The reason the increase in Garmin VO2max occurs during TrainerRoad is probably more of a function of actually running then it is a volume thing. Do you see anything here?
The relationship of Absolute VO2max estimate value AND Activity Calories over time
View attachment 433338
Again, nothing really jumping off the page at me on this one either. Any trends you see?
In summary
So what did I conclude from this latest data mining exercise:
1) When you train well, Garmin VO2max goes up. When you don't train, Garmin VO2max goes down.
2) There appears to be a relationship between my resting HR value and Garmin VO2max value. When my resting HR is at its lowest is usually when my Garmin VO2max is at its highest. This is probably explained by the adaptation seen from run training where there is an increase in blood stroke volume increasing the efficiency of each beat to move the same volume of blood around the body with less effort/beats. An equal (and even increased) amount of cycling training does not yield the same body adaptation for increasing stroke volume (or reduction in resting HR).
3) Weight loss is not the sole reason why Dopey 2018 went so well and there was a really high recorded Garmin VO2max estimation. There is evidence in the absolute VO2max values (under the absence of weight data) that there was an increase in fitness/efficiency seen as well. The combination of weight loss and gains in efficiency/fitness yielded the positive outcome at that race.
4) While I'm on average about 5-10 pounds heavier during these TrainerRoad cycles than previous run training cycles, the absolute VO2max values definitely show that the training is working from a fitness/efficiency evaluation standpoint. It is not clear whether dropping additional weight would yield a better outcome because it stands to reason that my current weight is keeping me healthy.
5) I may not be as run fast as I've ever been, but the absolute VO2max data suggests I might be more efficient/fit than I have ever been.
6) During normal run training cycles I was usually around 6-8.5 hours per week. During TrainerRoad training cycles (thus far) I've been around 2.5-4.1 hours of run training and 12-16 hours of total training.
7) There does not appear to be a X value of run duration that equals a Y value of Garmin VO2max estimation.
8) However, it did appear that more often than not to have a Garmin VO2max value above 56 usually coincided with run training volume of about 7.5 hours per week. But with the addition of run+bike+strength, the total volume used to gain a similar VO2max value as dropped down to around 4 hours. So a reduction of 40% of run volume (from 7.5 down to 4) yields a similar Garmin VO2max value when the total amount of training using run+bike+strength nearly doubles (from 7.5 hours to 15 hours). So to yield a similar running performance when substituting with bike+strength is not a 1:1 relationship and rather something closer to 1:2 (for every 1 hour dropped in training you need to do an additional 2 hours of total training). For example, drop 6 hours of run training down to 4 hours (-2 hours running) needs to have 12 total hours of training to yield similar result (4 hour run + 8 hour bike/strength [likely more bike than strength]). And additionally, I'd venture to guess the bike/strength training need to be structured and progressive.
9) Nothing unique was seen when evaluating using Activity Calories.
That was a fun data mining adventure! So you've got any thoughts? Anything I missed or misinterpreted in your opinion? Any other combinations of data sets to throw together? Again, obviously all of this was done using a single person's data (me) so that's to say it may not be translatable to everyone. But I think some of the concepts seen here could easily be useful outside of just my data set.