One of the most important concepts I have learned during my 73 years on Earth is that "Best is the Enemy of Good." That is, I should not skip doing something good because there might be something else that is even better but which I would not actually do. In this post, I am going to describe how I am thinking about using the average heart rate I measure for a ride to better estimate the Intensity of that ride even knowing that is not a perfect measure.
Let me start by redefining some terms I use often on this blog: Volume, Intensity, and Load. Volume is how long I ride measured in minutes. Intensity is how hard I ride. A ride might be harder because I am going uphill or riding fast, for example. One way of measuring Intensity is by measuring my heart rate. When I am riding hard, my heart beats faster. Load puts those two things together, Load = Volume x Intensity. Load increases both Fitness and Fatigue, terms which I won’t define because I won’t be using them much in this post and because they mean what you probably think they mean. My final point is that although Intensity can be measured by heart rate, Intensity and heart rate are not linearly related. That is, doubling heart rate more than doubles Intensity. I have blogged about this before so won’t belabor it here, but will just mention that at present the relationship between heart rate and Intensity I believe applies to me is that Intensity = 0.00065 x e(0.06 x heart-rate); Intensity increases exponentially with heart rate.
In July of 2022 I began routinely using my TranyaGo heart rate monitor to track my rides. The data I have available from this device is not as useful as the data that was available from the Garmin heart rate monitor that I used from 2012 through 2017, but realistically, I was not going to replace that Garmin heart rate monitor after it failed but I did purchase and have been using the TranyaGo, so what can I do with the data that it provides? What I want to do with the TranyaGo data is to improve how I track the Load generated by my various rides and thus the Fatigue generated. This would allow me to avoid both undertraining and overtraining and thereby reach a level of Fitness which approaches optimal. Realistically, it turns out the only two pieces of information TranyaGo provides that I can use for that purpose is my average heart rate of the ride and a graph of my heart rate over the course of the ride. That graph is useful to get a rough, subjective, visual impression of how the ride went, but to date I have not been able to use that graph quantitatively, meaning that the only information I can use to estimate Intensity and thus Load is my average heart rate.
Nobody I am aware of has tried to use average heart rate to calculate Load so I had to come up with a way of doing that on my own. The usual way of estimating Load is to divide up Intensity into Zones, associate each Zone with an Intensity, measure how much time (Volume) during a ride is in each Zone, and multiply each of those Intensities times each of those times and sum them together to get the Load of a ride. The way I have used average heart rate to calculate Load is to not use Zones at all but to assign an Intensity to each individual heart rate. For example, an average heart rate of 100 beat per minute corresponds to an Intensity of 0.26, 101 to 0.28, 140 to 2.89, and so forth. I then multiply that Intensity times the total number of minutes in the ride to get an estimate of Load. How does my way of estimating Load compare to the traditional way? Under some circumstances they are reasonably close while under others they differ enormously.
Let’s start with the case where my estimate of Load is way off. Consider the paper Gillen et al., a paper I have blogged about repeatedly. That paper claims that a workout consisting of three 20 second intervals ridden as fast as possible has the same benefits as a workout consisting of 45 minutes cycling at a heart rate 70% of one’s maximum heart rate (Hughes Zone 2.) Both workouts include 2 minutes of warmup and 3 minutes of cool down ridden at an easy pace (Hughes Zone 1) and the three intervals are separated by 2 minute recovery periods ridden at an easy pace. I am currently estimating the Intensities of the seven Hughes Heart Rate Zones as 0.25 for Zone 1, 1 for Zone 2, 3.5 for Zone 3, 7 for Zone 4, 13 for Zone 5, 24 for Zone 6, and 32 for Zone 7. Using those numbers, the Gillen et al. interval workout has an calculated average Heart Rate of 99. From the equation above, that corresponds to an Intensity of 0.25. The total length of the ride is 10 minutes, giving a Load of 2.5. This is very different from the value of 45 given in Gillen et al. However, calculating this the usual way, minute by minute using Zones, leads to an estimated Load of 34, both a value much closer than that claimed by Gillen et al. and very different that the estimate from average heart rate. Clearly, my average heart rate method is not useful for measuring the Load of an interval session. As it happens, heart rate produces inaccurate estimates for interval sessions no matter how you do the calculations and so coaches recommend calculating the Intensity of an interval session based on theory rather than heart rate, which is what I did above.
Now consider a case where my estimate and the traditional estimate produce similar values. Shortly after I moved to California, I used my Garmin Heart Rate monitor to measure one last ride. As a result, I have the data needed to compare the Load estimated by average heart rate to the traditional approach. Using the traditional approach, the total Load of that ride would be calculated as 170. Using my average heart rate approach gives a total Load of 215, certainly not exactly the same, but in the same ballpark. Moreover, what is important is not absolute Intensity (there is no such thing) but relative Intensity, comparing one ride to another. Because the error introduced by my average heart rate approach is similar for similar rides, this will have little effect on that all important relative value. Besides, the alternative is to ignore Intensity altogether when calculating Load, what I am effectively doing when I only track how long a ride is in minutes. Thus, I believe that despite the limitation of only having average heart rate data, using that to track Load is an improvement over what I have been doing up until now. Fortunately, the type of ride for which my method works fairly well turns out to be the type of ride I do most frequently.
So at last we can turn to the figure at the top of this post. This graph shows the minutes (Volume) and Load of 34 recent rides ridden over 45 days. During this period of time, I did only two kinds of rides, a 30 minute ride on my trainer or an approximately 100 minute ride on my Cañada route. Thus, there are only 3 values for Volume; 0 minutes, 30 minutes, and approximately 100 minutes. Load gives a different picture. I have circled some illustrations of that. The first pair of circles (going left to right) shows three rides that are almost identical in Volume but significantly different in Load. That is because they were ridden at different speeds. Of course these different speeds affected the Volume of the rides, a fast ride might take less than 100 minutes while a slow ride would take more, and if you look carefully at the graph, that is apparent, but both because Heart Rate goes up faster than speed, and because Load is exponentially related to heart rate, that small different in Volume translates to a much larger difference in Load.
The second pair of circles illustrates the impact of varying how I do my trainer ride. Usually I use trainer rides as recovery rides, riding them in Zone 1. In the circled example, I decided to ride a version of the Gillen et al. training session. Rather than doing the three sprints used by Gillen et al. I did a total of six 20 second sprints each separated from the next by two minutes in Zone 1. With warmup and cool down, this was still a 30 minute ride and thus does not stand out at all on the graph of Volume. As noted above, I could not use average heart rate data to calculate Load so I did a theoretical estimate assuming 2 minutes in Zone 7 and 28 minutes in Zone 1. When I did that, the Load was much higher than for a recovery ride on the trainer, which is as expected.
The third set of circles illustrates how little Load is produced by my recovery rides. A recovery ride is visually very different from no ride at all on the Volume graph but almost undetectable on the Load graph.
Back in 2017 when I had first moved to California, I blogged that I was going to switch from tracking how many miles I rode to tracking how many minutes I rode because I believed minutes was a better estimate of how much exercise I was getting in the hills of California. I think it is clear from the data presented in this post that Load would be an even better estimate. That said, my attempt to use average heart rate to estimate Intensity is very much a work in progress. Although I am using my TranyaGo to routinely track heart rate on my rides, I have yet to integrate that with my routine training data in a way I can use it to manage my training. I will continue to blog about this as I figure out how best to do that.
