I am working on some simple visualisations to help with the trend analysis for diabetes data (for my son). The website attached is automatically updated when I download data from the blood sugar reader. I intend shortly to add a trend line for carbs, exercise and insulin to help solve problem areas we are having with regulating our boys BS levels. If you have any thoughts or suggestions on the design or you know of someone doing something similar it would be great to hear about it.
with our Glucosurfer we are working at the same goal. We look for trends and offer different and in some cases unique ways to analyse diabetes data. You can drag the diagram with your mouse. Please read the description of the diagrams:
We are running our own server at a computing center. So the data is in our hands and not in the hands of google. Most importantly the service runs under full anonymity: we do not collect any data that can identify a person - not even an email address (you can see that on the page ‘Your membership’). As a result the service is offered for free and runs on donations and our private pockets if necessary (see page ‘Your donations’).
Thanks for message. I had a quick look at glucosurfer and it looks just like what I am working towards. Have you had much success in getting clinical Diabetic teams to adopt the practice of using the facility to assist tracking their groups rather than rely on phone calls and quoting BS levels from readers ?
My diabetologists have thousands of patients and most of them have compulsory health insurance. For this group the patient has to come to the doctor with his patient id card so the service can be accounted for. Right now remote assistance is not covered by these insurers (of course an accute crisis is an exception here). For patients with private insurance or self-payment the situation is different. Sometimes patients are coming from remote places and my diabetologists wrote this short introduction to get their patients interested in their services. I can not tell how successful this approach is because for us all accounts on the Glucosurfer are managed anonymously. But as you can see they are willing to adopt and I for myself can count on their help.
It would be interesting to see how the BS management was improved as a result of your work and use of the site (my statistical background kicking in). I like the visualisations you have set up especially the representation of the insulin (with the shaded overlap showing loading). I use the google apps because I am fairly limited with my own programming ability. The google motion chart comes originally from Hans Rosling’s work and if you like it it may be worth getting in touch (gapminder.org) to see if they would provide the charts (no need for google) . I still like the google time series chart a bit more as it allows you to drill down to the reference period you might be looking at to identify trend’s. Anyway time for me to get some more data on activity levels and carb intake for my boy
Felix, you may want to check out SugarStats.com, which has a lot of useful features.
Which meter are you loading the data from, and did you write any software to do that data extraction? I had started trying to track information about diabetes data formats on this wiki. If you’ve any information I can add, I’d appreciate it.
Hi Bernard. My data extraction is very crude at the moment. I download from Fionns Accu check Nano onto the PC. I then manipulate the data into the correct format using excel and syn with google docs (via automatic backup software). The visuals then run using the URL in google docs as the source data. I intend to move to smartphone blood sugar reader (as soon as available) so that I can bypass my home PC and utilise wifi to update the readings as they take place (or sms). Then the phone can go with Fionn when hes with friends or at school and the teachers won’t need to be bothered with taking down readings as they will be recorded on the website.
best regards
(general election tomorrow and getting ready to vote. I’ve been asking all candidates to commit to getting a diabetes register in Ireland as it doesn’t currently exist !)
Nice job. Another potentially useful visualization is to divide your long term data in week long chunks, and superimpose them. I’d suggest Monday through Sunday. That way, if there’s a pattern of highs on weekends but not weekdays, it will become quickly apparent.
