In the time between Enigma and FOSDEM, I have been writing some musings on reverse engineering to the point I intended to spend a weekend playing with an old motherboard to have it run Coreboot. I decided to refocus a moment instead; while I knew the exercise would be pointless (among other things, because coreboot does purge obsolete motherboards fairly often), and I Was interested in it only to prove to myself I had the skills to do that, I found that there was something else I should be reverse engineering that would have actual impact: my glucometers.
If you follow my blog, I have written about diabetes, and in particular about my Abbott Freestyle Optium and the Lifescan OneTouch Verio, both of which lack a publicly available protocol definition, though manufacturers make available custom proprietary software for them.
Unsurprisingly, if you’re at least familiar with the quality level of consumer-oriented healthcare related software, the software is clunky, out of date, and barely working on modern operating systems. Which is why the simple, almost spartan, HTML reports generated by the Accu-Chek Mobile are a net improvement over using that software.
The OneTouch software in particular has not been updated in a long while, and is still not an Unicode Windows application. This would be fine, if it wasn’t that it also decided that my “sacrificial laptop” had incompatible locale settings, and forced me to spend a good half hour to try configuring it in a way that it found acceptable. It also requires a separate download for “drivers” totalling over 150MB of installers. I’ll dig into the software separately as I describe my odyssey with the Verio, but I’ll add this in: since the installation of the “drivers” is essentially a sequence of separate installs for both kernel-space drivers and userland libraries, it is not completely surprising that one of those fails — I forgot which command returned the error, but something used by .NET has removed the parameters that are being used during the install, so at least one of the meters would not work under Windows 10.
Things are even more interesting for FreeStyle Auto-Assist, the software provided by Abbott. The link goes to the Irish website (given I live in Dublin), though it might redirect you to a more local website: Abbott probably thinks there is no reason for someone living in the Republic to look at an imperialist website, so even if you click on the little flag on the top-right, it will never send you to the UK website, at least coming from an Irish connection… which mean to see the UK version I need to use TunnelBear. No worries though, because no matter whether you’re Irish or British, the moment when you try to download the software, you’re presented with a 404 Not Found page (at least as of writing, 2016-02-06) — I managed getting a copy of the software from their Australian website instead.
As an aside, I have been told about a continuous glucose meter from Abbott some time ago, which looked very nice, as the sensor seemed significantly smaller than other CGMs I’ve seen — unfortunately when I went to check on the (UK) website, its YouTube promotional and tutorial videos were region-locked away from me. Guess I won’t be moving to that meter any time soon.
I’ll be posting some more rants about the problems of reverse engineering these meters as I get results or frustration, so hang tight if you’re curious. And while I don’t usually like telling people to share my posts, I think for once it might be beneficial to spread the word that diabetes care needs better software. So if you feel to share this or any other of my posts on the subject please do so!
