The quantified self and accuracy

Fitness tracking and accuracy

There has been a lot of discussion recently around the usefulness of activity trackers in enabling people to be more health. Some of these are less favourable incluing https://www.independent.co.uk/news/health/nhs-apple-watch-fitbits-ai-waiting-times-gp-misdiagnosis-a8749876.html.

I thought it is time for me to share my own personal views of using activity trackers for some time. My first activity trackers were cheap things bought from Amazon which gave a sort of measurement of steps which was useful but the dashboard was not really useful for me so I upgraded to a Fitbit Charge 2. I still wear this all the time and love the data it provides me. The key thing is the data is understandable and in a form I can digest, although there are parts of it I do not record as I am not interested in them.

Overall, throughout my time using a Fitbit, I have become fitter and more aware of my own levels of health. I like the sleep tracker and although I am aware of my sleep patterns it is great to be able to see how my sleep patterns change when I am under stress or doing a lot of exercise.

What I have learned is that data is useful to support what I already know. I don't think the data from the Fitbit has enabled me to find out anything new that I did not already have a good idea about, apart from the heart rate function.

The Heart Rate Monitor (HRM)

What I like about the Fitbit Charge 2 is the ability to see my heart rate at any time at a touch of a button and a screen touch.  I am really happy I am considered to be in excellent fitness with a cardio score of over 45. I have also found that my resting heart rate is around 60BPM. Again this is very encouraging as it means I am not necessarily the decaying person I sometimes feel I am.

Being a nerd, I have regularly counted my pulse at the same time as using the Fitbit and I have always found it really accurate. Accurate at low rates, but at higher rates it is a little less reliable.

I cycle, when I can,  on a push bike.  I also belong to a local cycle group which means that every week there are people going out for group rides.  These rides started at around 20 miles and have increased for me last year to 70 miles and hopefully in June I will ride my first 100 mile sportive since becoming an adult.

The Fitbit's accuracy on rides is very patchy.  I noticed more than a year ago that when I ride to the top of a hill, which I find very hard indeed as I hate hills, my pulse skyrockets. I recall counting my pulse for a short time when we stopped once and it was going at about 170bpm but the Fitbit was showing 85bpm.  Quite a disparity, wouldn't you agree?

In order to test this out in a more scientific setting I did a small test at home on my turbo Trainer, doing hill reps. The session was only 30 mins or so but it did show a huge difference in readouts between the two devices. I should point out that I use a Wahoo Heart Rate Monitor chest strap when I cycle.

Here are the readouts:

Firstly the Fitbit readout for the morning including the ride:

As you can see my pulse went up to 90bpm according to the Fitbit during this rather strenuous session.  This is not bourne out by my own observations of my pulse which I recorded as @176 at one point during the session using my watch and fingers.

Interestingly the Wahoo HRM gave a slightly different picture and one which seems to bare my own experiences out:

The Wahoo HRM puts my pulse/heart rate at averaging 166bpm for the session, which is almost twice what it recorded my peak rate on the Fitbit.  Moreover, the Wahoo has me doing upwards of 184bmp at a few points in the session.  Which explains the reasons why when I finished I was exhausted and felt like I had undertaken a good workout.

Is accuracy important?

The big question is whether accuracy is important?

My personal response is yes and no!

No it is not totally important to be totally accurate. I know when I wake up and look at my Fitbit and it says I have walked 160 steps this is not accurate.  For me the progress I make is more important.  By this I do not mean that I have to keep beating what I did before, rather that I am acknowledging to myself and affirming through Fitbit that I have not been a couch potato all day.

To me everything is a relational scale which I use to make judgements on how well I am doing.  in 2016 I had a year where I was really unwell for much of the time and if I had had this data then I might have gained some solace that I was improving, at times.  Even today when I have bouts where my condition resurfaces, I can use the data from the Fitbit to affirm I am still alive and continuing to move.

There is an issue though; How can a Fitbit be so wildly inaccurate during an exercise session.  I recall looking at the Fitbit on many occasions during the Turbo session and each time it recorded my pulse as below 90 where as the display from the Wahoo had me at at least 170 at the times I was looking. This could be a problem if I was not able to think for my self and relied totally on the output from the Fitbit.

If for example I had a heart condition, which hopefully I do not, then it is really important to keep you heart rate within clear parameters. As it is, I try to do this in any case which is why I purchased the Wahoo for cycling as when you are out and about, it is more difficult to tap a watch from the position of riding a bike.

Should data be God?

Poor heading, I know, but I could not think of anything better to explain the question. Data, via the quantified self, in which we are constantly measuring ourselves to determine who health we are, is the new fad.  It saves time at the doctors by allowing us as people to take control over our own health... or so the great plan in the sky suggests. But is is really possible to be accurate with health data. We constantly use short hand phrases which have no scientific backing as messages from on high. for example, 10,000 Steps,  five a day, or my BMI is too high. Although these well known quantifications are used readily and without doubt have been useful to allow people to change their behaviours, there is little scientific substance to either.

Using fitness tracking devices does mean that we are allowing ourselves to accept poor data. But this is not a problem unless the data is being used by people who do not understand it is poor or take the data as some form of Gospel truth.

Any data is flawed. The bigger the dataset the more accurate the results, does not always ring true.  I realise I am diverging into a mathematical discourse but what I hope any reader can take from this is that instead of letting the machine determine your health, you can determine it and use the machine to support you in your quest.

Personally I don't care whether I walked 3000 steps of 10000 steps or  two million steps, I am only interested in the fact that I am active every day as much as I can be.  Technology can support this but it does not control it.  I do as I hope you control your own.

Finally, No I have no intentions of ditching my Fitbit just yet and have not found any more accurate wrist health monitors around.  I am sticking with Fitbit and Wahoo.