Body weight – lean mass and fat mass

By September 13, 2011Personal health

In January 2010 when I started getting serious about my health and staying young for longer losing weight was easily the most important objective.  As you can almost see from the chart below at that point my weight was 77.8kg and 24.6% of that was fat (rather than muscle or bone).

I started recording what I ate, and counting the calories, and in a couple of steps I went down from eating 3,000-3,500 calories a day to a target of 2,000-2,100 (excluding alcohol).  The weight started dropping off pretty quickly and my body fat percentage also dropped.

I was then faced with figuring out what my target weight was, and a lot of friends were concerned that I was going too far.  I have long had this notion that 67.5-70kg (147-154lb) was a good target, but I wanted to be a bit more scientific.  The Body Mass Index or BMI was too crude for me, so I decided to define my goal as achieving a body fat percentage of 10-13%, a figure I took from Ray Kurzweil’s TRANSCENDWikipedia suggests that fit male adults should have 14-17% body fat, with male athletes at 6-13%.  Women are higher – 21-24% for fit females, and 14-20% for female athletes.

The next problem then, was to measure my body fat percentage, and that turns out to be quite a challenge, as you will see from the rest of this post.  I had the data from my Withings Scale, which is what you can see in the chart below, but I wasn’t happy that it was accurate.  The day to day fluctuations were too great to make sense and I had this nagging feeling that at around 16% all the numbers were on the high side.

image

The first thing I did was buy a handheld body fat measurement unit which uses the same electrical impedance technique as the Withings Scale, hoping that a second measurement might corroborate the first, or give a clear idea where it was wrong (e.g. if it was too high).  Unfortunately the measurements from the new unit had greater fluctuations than the Withings Scale, and the numbers were on average even higher!

My next step was to get measured in a Bopod unit in Hampstead back in June.  That showed my body fat percentage as 9.9%, which struck me as too low, and my confidence in the number wasn’t helped by the fact that the operator took an unscheduled third measurement because the first two were unusually far apart. 

However, if the 9.9% was accurate then I shouldn’t be losing any more weight, and that had an impact on my diet.

Since June I have put on 1-2kg (impossible to be more precise than that because my weight fluctuates by 1-2kg on an intra-day basis), and I was thinking the increase was due to an increase in my lean mass rather than because I had put on fat.  How wrong I was.

Yesterday I got measured in a DXA unit which uses a scanning technique, and my body fat percentage has increased to 14.3%.  The increase from 9.9% in June to 14.3% measured yesterday is explained mathematically by a 3.5kg increase in my body fat and 0.8kg decrease in my lean weight, which that 81% of the increase comes from an increase in fat.  Rather depressing.

Once again, however, I am unconvinced by the quality of the data.  Looking at myself in the mirror, I think 14.3% body fat is too high, and I have two issues with the measurements.  Firstly, I find it hard to believe that my lean mass has decreased in the three months since June.  I strongly suspect the 0.8kg decline is due to calibration or measurement errors with either the Bopod or the DXA machine.  Secondly, I weighed in at 72.5kg yesterday, which is 2.5kg up on Saturday morning, and I think that is at the top end of my day to day cycle of fluctuations.  If I had weighed 71kg, which is about average for me at the moment, and the 1.5kg difference came off the fat mass then my body fat percentage would be 12.5%, and just inside my target range.

This story illustrates just how hard it is to get accurate data on our bodies.  I have described how I have bought two different home measurement devices and gone for two scans, the total cost of which is £220 and a lot of hassle.

As I written at length in the context of startups measurement and good data are critical to good decision making.  That is also true in matters of personal health, but with the additional qualification that capturing the data must be easy and convenient if it is to be picked up by the mass market.  Whilst the Withings Scale is a step in the right direction the quality of the data coming out of it isn’t good enough once the body fat percentage gets into the 15-20% range, and we are clearly some way from the point where measuring lean mass and fat mass will go mainstream.

Enhanced by Zemanta
  • kinda too hardcore no? 

    any half-bit scientist (which includes yours truly) will tell you that your rather emotionally turbulent experiment violated the cardinal rule of research and questioning: maintain consistency in measurement methods. you measured your bodyfat at various points in time using wildly different methods each more inaccurate than the last and were surprised the results were all over the map? next time.. keep all but one variable constant: measure your weight and body fat only in the morning, at a specified time before breakfast using the same instrument each time.. Only then can you compare apples to apples. 

    edit: it doesn’t matter if it’s inaccurate, as long as *it’s inaccurate by the same amount each time*. You’re not in a competition with “averages” and whatever Wikipedia says… That is as asinine as using BMI as a bible. Rather you’re in competition with yourself, to better achieve your body’s targets as you discover them.

  • Hi Roham – I think I am only partially guilty as charged. I used the Withings Scale in the consistent manner you describe and still got wildly varying results. Apparently it gives very different results depending on levels of hydration, and it is too much effort to control those.
    The Bopod and DXA machines are supposed to be much more accurate.

    You make a good point though. Maintaining rigour in these self experiments is as important as it is tiresome. I could do better.

  • Jouko Salonen

    “too hardcore?” it may sound so, but the point is, we need easier ways to get better quality data on our body. Not only on fat percentage. Right now I’m having my morning coffee blended with 50% chaga mushroom tea. I’d like to know what is happening inside.

  • Jouko Salonen

    “too hardcore?” it may sound so, but the point is, we need easier ways to get better quality data on our body. Not only on fat percentage. Right now I’m having my morning coffee blended with 50% chaga mushroom tea. I’d like to know what is happening inside.

  • The data is slowly becoming available – the availability is exciting, but the slowness is frustrating. I guess that’s what I’m trying to say here.
    ________________________________

  • The data is slowly becoming available – the availability is exciting, but the slowness is frustrating. I guess that’s what I’m trying to say here.
    ________________________________

  • Quick question (having bought first bodyfat detecting scale and viewing it as somewhat inconsistent), are any of these tools improved by using a 1 week trailing average to calm down the weirdness?

  • Christopher McAtominey

    There is a tailoring company called Body Map Me that uses the Xbox Kinect to measure your body size for clothing. The parent company has a much more accurate laser scanner for suits but the Kinect is suitable for sizing most off the peg garments.

    It seems to me that something like the Kinect could be a good solution for measuring body fat as it can get a good (and consistent) measurement of various body parts (ie biceps, thighs etc.) and approximate from there. As @pnpegypt:disqus said: “as long as *it’s inaccurate by the same amount each time*”.

  • Yes – looking at the trendline on a graph, or going to the trouble of calculating the trailing average will smooth out differences. That worked fine for me at first, but when I got close to my targets it becomes more important to have an accurate absolute measurement, and I’m not sure the impedance technology delivers that, even if results are averaged.
    Sorry for the slow reply.

  • NicolasCacace

    If you have two devices that demonstrate a consistency in variance then I would say that they both work but need to be calibrated.