Wednesday, December 16, 2015

How much easier is running on an AlterG? Developing equal-intensity curves for anti-gravity treadmill running

Have you ever run on an AlterG? Once firmly in the realm of space-age gadgetry only available to professional athletes, AlterG anti-gravity treadmills seem to be cropping up everywhere nowadays.  College athletic departments, physical therapy offices, and even the occasional high school are purchasing AlterGs for their widely lauded ability to allow runners and other athletes to continue to train pain-free even with significant injuries.  By reducing your effective body weight, the AlterG allows you to run at normal training speeds with drastically reduced impact and active forces.  With careful modulation of the body weight settings, you can often maintain running fitness even during the rehab period of formerly season-ending injuries like a stress fracture. 

The AlterG achieves its anti-gravity effects using a pressurized "bubble" that encapsulates the runner's lower body.  Special compression shorts with an airtight skirt zip securely into a thick vinyl bubble that surrounds a standard running treadmill.  The heart of the AlterG, a computer-controlled air pump, inflates the bubble to above atmospheric pressure, applying an evenly-distributed force from air pressure to the runner which counters the force of gravity.  By adjusting the air pressure inside the bubble, the AlterG can adjust your effective body weight. The AlterG uses a force plate to correlate changes in the bubble's internal air pressure and your effective weight while standing on the treadmill.

In the past few months, I've been fortunate enough to have access to an AlterG.  I've also been fortunate to not have to use it for any injuries (knock on wood...), so I used the opportunity to look into a question that I've been wondering since learning about the AlterG: How much easier is running on an anti-gravity treadmill compared with running on land?

Because the majority of the metabolic cost of running comes from absorbing impact and accelerating your body weight against the force of gravity to propel yourself forward, it's axiomatic that reducing your effective body weight while maintaining the same running speed will reduce the energetic cost of running.

Notably, this is not the same situation that occurs when a runner loses weight normally—if a 150 lb runner decreases his weight to 140 lbs by restricting his caloric intake, muscle loss is inevitable (this is part of the problem with the idea of "racing weight").  Though he now weighs less, and thus the energetic cost of running a given speed is decreased, he has also lost some muscle, so his ability to produce energy is reduced as well.

There's no good bio-energetic equation to predict the metabolic cost of running; the only way to get a good answer would be with an experiment, which I set out to conduct. 


Now, this isn't a real scientific study with a large number of participants and advanced statistical models.  The only subject I had access to was myself, so this was an experiment of one.  There is almost surely some degree of individual variation in the metabolic cost of running at reduced body weight, and I didn't have access to a high-tech physiology lab either.  Neither of these are a big issue, though, since all we're really looking for is a rough picture of the overall trend.

To get a full picture of how the metabolic cost of running changes over a range of effective body weights, I needed to collect data at different speeds and different body weights.  To estimate metabolic cost, I would have to use heart rate monitoring, the only objective measurement available to me.  With the help of a friend who is studying exercise physiology at the University of Minnesota, I planned out a progressive treadmill test that would allow me to record my average heart rate at paces ranging from ten-minute mile pace all the way down to five-minute mile pace, increasing speeds in small increments to collect more data.  The protocol is reproduced in the table below.

Pace (min/mi)
Miles per Hour
Trial time

 I carried out this treadmill test three times, at 60%, 80%, and 100% of my effective body weight on the AlterG.  During the tests, I recorded my heart rate (as measured by a Polar chest strap / watch Heart Rate Monitor) in the middle and at the end of each three-minute segment.  

The AlterG, on the other hand, allows a 150 lb runner to run as if he weighted 100 lbs, but with all of the muscular power of a 150 lb runner.  Because of this, running at a reduced body weight should be vastly easier.  But how by how much?

The treadmill went no faster than twelve miles per hour, which was not a problem—my 5k fitness was no better than about 15:50 at the time of the tests (which is 11.8 miles per hour), so 12 mph should have elicited my maximum heart rate. 


Despite the fairly rudimentary setup, the quality of the data I collected was very good.  For all three trials, heart rate showed an almost perfectly linear relationship with running pace, which indicates that the heart rate data is reliable (seeing as that's what it's supposed to do).   The graph below shows all the data collected, alongside trend lines and crude error bars (given as two standard deviations of the variance between the 1.5 min and 3 min heart rate measurements for all segments).
The raw data from each of my three trials on the AlterG
Small non-linear regions can be observed at very low speeds for the 60% body weight condition and near "maximum" heart rate for the 100% body weight condition.  In both cases (very low and very high exercise intensity), these deviations from linear behavior are expected.  These deviations were not included when fitting the trend lines.

The only other abnormality in the data was my apparent maximum heart rate of 172 beats per minute.  I'm 26 years old, and am reasonably well-trained, so hitting a "maximum" heart rate this low was something of a surprise.  Regardless, that's all the higher my heart rate went after 2.5 minutes (didn't make the full three) at five-minute-mile pace at the end of almost 40 min of progressively faster running.  I confirmed it by manually checking my pulse for ten seconds, then multiplying by six. 

I'm also not one for using physiology to an unnecessary degree in training, so I've never actually had my maximum heart rate tested, or even used a heart rate monitor during a workout, for that matter.  When I asked a few friends better-versed in exercise physiology testing than I am, they all agreed that a max heart rate this low was very strange, and suggested I try another protocol to find out my true maximum.  Regardless, because the linear relationship between submaximal heart rate and running speed was so strong, whether or not 172 represents my true maximum heart rate or not shouldn't matter for the rest of the data analysis.

Generalizing the results: The equivalent heart rate curves

When presented as raw heart rate and running speed data, these results are only useful for one person (me).  To get more broadly-applicable results, we need to find a more general relationship between running speed, body weight, and heart rate.  Fortunately, the three trend lines allow us to do just this. 

By solving for heart rate using the 100% body weight trend line for the range of training speeds we would be interested in—5:00 mile pace, 5:20 pace, 5:40 pace, and on down—we can then plug these heart rate values back in to the 80% and 60% body weight trend lines to find the corresponding paces at those relative body weights that correspond to the same heart rate.  Then, by drawing a curve that connects each of those three points, we can interpolate for body weight percents that I didn't actually measure.  This will give us a smooth curve for every relevant pace that tells us how much faster we'd have to run at a given relative body weight to elicit the same heart rate.

This method gives only three data points for each curve, and since the relationship was clearly nonlinear, a polynomial fit was required.  It doesn't make sense to talk about goodness of fit with this kind of model, since a best-fit curve of a polynomial of degree two or higher will always be a perfect fit with only three points.  This produces some obvious shortcomings in the method—the reality of the relationship is surely not parabolic—but it's what we have to work with! Sure, I could have completed more trials (at 50, 70, and 90% body weight, for example), but I had real training to do!

The equal heart rate curves that result from this method are best depicted on a graph.  Each curve essentially answers the question, "Compared to running [speed x] on flat ground at my normal body weight, how much faster would I have to run on the AlterG to get the equivalent effort level for a given relative body weight?"
Click to enlarge
One thing that quickly becomes apparent is that the maximum speed of the treadmill limits the range of equivalent efforts that are possible to reproduce at lower body weights.  For example, it is not possible to reproduce the equivalent effort of 5:20 mile pace at any lower than about 85% body weight, since it would require treadmill speeds above twelve miles per hour. 

One way to circumvent this problem would be to increase the treadmill's incline.  I did not collect any data on running on an incline with the AlterG, but according to Jack Daniels, each percent gradient increase in incline will increase the difficulty by 12-15 seconds per mile for overground running.  Since the AlterG reduces your weight, this will probably overestimate the increase in difficulty, but it's a good place to start.  So, for example, running 5:00 pace (12 mph) at 60% body weight is equal to only 5:40 pace on flat ground.  To get this effort to be equivalent to 5:20 pace, you could increase the incline to a 2-3% gradient.  Do keep in mind that some injuries (Achilles, calf, metatarsals, possibly IT band syndrome and other knee issues) will react poorly to uphill running, so you won't always be able to use this trick.

For practical use, a table is more informative.  The color-coded table below presents the same data, with speeds in miles per hour rounded to the nearest tenth.  This is the same precision offered by the treadmill inside the AlterG.

Click to enlarge

There are a number of important points to consider before using these results when designing an AlterG training protocol for an injured athlete. 

First, since these data were collected on just one person (me), there's bound to be some margin of error.  Even my own heart rate measurements varied by about ± 3 bpm at constant speeds.  My guess is that, even following these curves, heart rate could vary by around 5%.  You'll be interpolating a bit anyways if the pace you want is not a multiple of twenty seconds per mile.  More accurate data would require a larger number of subjects and a greater range of samples along the relative body weight spectrum.

Second, heart rates will only be equivalent across the curve for "submaximal" speeds—basically, 5k pace or slower.  The data can probably still be used for maximal/supermaximal speeds since the relationship probably still holds, but any heart rate data you are using won't be indicative of effort anymore, since heart rate is maxed out already.  

There is also a phenomenon called "cardiovascular drift," where heart rate drifts higher over the course of a long continuous run (30-60min), even if it's at the same speed.  The magnitude of this increase in heart rate can be quite large (up to 15%) and is exacerbated by exercising in hot conditions.  The AlterG I was using was in a small room without a fan, so cardiovascular drift was surely a factor.  Of course, any time you'd run on this particular AlterG, you'd be running under the same conditions, so at least for my own purposes, this should not be too much of an issue.

Additionally, these curves only consider the physiological demands of running at a given speed and body weight combination.  There is a compelling argument to be made that, even though your heart rate is lower when running 5:00 mile pace at 70% of your body weight, you are still getting some neurological benefits from your legs moving at that speed (though I would point out that absorbing and producing the relevant impact and active forces, respectively, is a significant part of the specificity of training). 

Finally, these curves do not represent equal levels of impact or stress on the body.  I'm not sure how impact and stress scale on the AlterG.  It's easy to see why running 7.5 miles per hour at 80% of body weight imparts less stress on your body than 7.5 miles per hour at 100% of body weight.  But how does the impact of 8.1 miles per hour at 80% of body weight relate to the impact of 7.5 mph at 100% body weight (an equivalent effort in terms of heart rate)? I'm not sure.  Intuition tells me that it should still be lower, but not as low as the impact forces at 7.5 mph at 80%.  I would need more advanced equipment (a force plate or tibial accelerometer) to answer this.

Ultimately, you will have to use pain, or lack thereof, as the primary guide for what speeds and body weight percentages are acceptable for an injured athlete when running on the AlterG. 


Heart rate monitoring can be used to create a mathematical model which describes equivalent metabolic efforts for running at different combinations of speed and relative body weight on an AlterG anti-gravity treadmill.  This model can be useful when trying to determine appropriate training paces while running at reduced body weight, but it does not describe the variance in impact and active forces along the continuum of speed and relative body weight.  Future research using data from force plates or tibial accelerometry will be necessary to describe that relationship.  More precise data on the metabolic cost of running could also be collected using respiratory exchange ratios instead of heart rate.


  1. Fascinating article and a great piece of work.

    One question and I apologise as it's minor in the context of the article. You refer to your max heart rate and write "am reasonably well-trained, so hitting a "maximum" heart rate this low was something of a surprise." Based on the various rough formulas for estimating max hr that are around, this would appear to be so but is max hr actually trainable? I'm under the impression that max hr is a fixed variable albeit one that declines with age. Training does not increase it, what training does is increase stroke volume so the heart pumps more blood per beat. Any thoughts? Once again, apologies for digressing, overall article is, as I said, fascinating.

  2. Yes, that's correct - to my knowledge, maximum heart rate is not really trainable. My thinking was more along the lines of "maybe if I was untrained/out of shape, I would get muscularly fatigued from the volume of fast running before hitting my heart rate max and getting fatigued from oxygen debt." Kinda like how if you haven't run in a while and you go out and try to do a ten-mile run, you'll be exhausted by the end, but probably not at your HR max.

  3. Hi John, talking about HR max, do you know any study mentioning the decrease of our HR Max according we will getting older ? For example, when I was 23, my HR max used to be 210 and now with 35 and training in the same intensity and volume my HR max is 200. What's your opinion about this ? Tks

  4. Running is a tradeoff between potential energy (COM bouncing up and down with each stride) and kinetic energy (swinging the legs back and forth). Lower cadence results in more potential energy and less kinetic, higher cadence vice-versa. The former is proportional to gravity, the latter is not, so I'd expect, given the potential energy cost of low cadence is reduced, the optimal cadence with reduced "weight" would be lower, i.e. more of a bounding gait. Note this does not reduce mass, just weight: the downward force but not the momentum associated with a leg swinging forward (or backward; it's a treadmill). Did your cadence change?

  5. I am agree with your post. I hope you are always shared me this type of post. Thanks a lot. fitness bicicletas estáticas

  6. The possible objects would help students to bring around all those possible prospects which must have been understood by the individual to regard about.