Category Archives: geeky

The Doctor wanted more data? Well, here it is!

The challenge

Back in… what? 2006,  I went to the doctor for a routine checkup, he was concerned that my blood pressure might look a tad bit high. He continued to explain sometimes the stress of going to the doctor’s office might artificially increase your blood pressure, and I should go to a pharmacy where you can get your blood pressure measured for free.

The experiment

Well, I decided why not one-up that idea and buy a cheap personal blood pressure/heart rate monitor? I recorded my heart rate and blood pressure almost every day for about a year, sometimes more than once a day. I know, they say you should make these measurements all at the same time, usually just after waking up, but I… got lazy ok? As a side note, I got very little exercise over this period, I’m curious how these plots will look now, since I exercise pretty much every day.

Heart rate over 1 year of measurements (2006), not taken at the same time every day.

The time series of my heart rate (above) seems noisy. Let’s have a look at the histogram of that:

Histogram of heart rate for 2006 (180 measurements). Looks like a Ricean distribution with mean~=64 bpm

Mean of 64 bpm? That is definitely not cool. I’ll have to start taking some data for 2010 and hope that exercising 5+ times a week brought that average down!

Let’s have a look at blood pressure. Below is a time series of both systolic and diastolic blood pressure:

Time series for SYStolic and DIAstolic blood pressure for 2006

We can look at the distribution for both these measurements as well:

Systolic blood pressure distribution appears quite gaussian with a mean of 133

Diastolic blood pressure distribution for 2006. Mean is 79

So over a 1 year average, my blood pressure was 133/79 and average heart rate 64 bpm. It’s not horrible, but something like 120/80 with a heart rate of, say, 50, would be way cooler. Perhaps that is the goal for 2010?

Instrument Calibration

Getting back to the original story line, I showed these graphs to the doctor. Needless to say, he was not impressed. I insisted that this data needs to be traceable back to a ‘national standard’, so to speak. So the doctor used his standard-issue blood pressure device, while 10 minutes later I used my electronic gadget to make (hopefully) the same measurement. It turns out the Doctor’s device was always about 10 points below mine. He decided to use this one measurement to refute all my data: obviously my device is wrong.

After he made his one measurement, he declared that my blood pressure is fine and there’s nothing to worry about. Bah.

Clock drift of 2 wall clocks in my office

Over the last six months I have very carefully synchronized two wall clocks together in my office. I then proceeded to make observations every couple of days and record how many seconds these clocks have drifted from UTC(NIST). Since these measurements are made with my eyeball, they are accurate to about 1 second RMS. As a result of this inaccuracy, I can’t tell you the wall clock stability over short periods of time, but this 1s RMS error becomes insignificant when computing the clock RATES over long periods of time.

 Over 6 months, they seems to have drifted 45 and 30 seconds away from truth. With only this one point, it is possible to establish the clock rates to be about 2e-6 seconds/second and 3e-6 seconds/second. That basically means both clocks gain 2 and 3 microseconds for every second. Another way to put it, is that the clocks have a frequency error of about 2 or 3 parts per million over many days – which is about what we expect for simple Quartz oscillators. Another important factor that I can’t measure with my 1s RMS error is the stability of the clocks. Sure, there is what appears to be a linear rate offset of the two clocks from UTC(NIST), but what is the jitter/stability of this 2 and 3ppm measurements? This jitter would give us a clue of the alan deviation of the clock, but alas, I can’t measure this easily with my eyeball.


A casual bike through streets of Pasadena, California, USA

I decided to take my shiny new toy (Canon SD940 digital camera) for a spin on my road bike. I taped the thing to my handle bars and took 20 minutes of 1280×760 video. It filled up about 4GB of disk space at 30fps. It’s truly beautiful watching the uncompressed video – unfortunately something horrid happened when I go from 4GB to 50MB with MPEG4 compression, I guess the laws of physics are very unforgiving.


Bike with camera 'installed'

Here is the movie (below) (and a link as well), sorry for the poor quality. I already burnt a large part of my saturday trying to get all the software to play with each other, time to go do other fun things – so long, suckers!

Pasadena: Rose Bowl and Neighbouring Areas (on bike) from Stephan Esterhuizen on Vimeo.