A Year of Diabetes Data

From Personal Science Wiki
Jump to navigation Jump to search
Project Infobox Question-icon.png
Self researcher(s) Doug Kanter
Related tools Google sheet, blood glucose monitor, Photos
Related topics Metabolism, Food tracking, Blood glucose tracking, Insulin intake, Heart rate, Activity tracking

Builds on project(s)
Has inspired Projects (0)
Show and Tell Talk Infobox
Featured image A-year-of-diabetes-data.jpg
Date 2013/10/10
Event name 2013 QS Global Conference
Slides A-year-of-diabetes-data.pdf
UI icon information.png This content was automatically imported. See here how to improve it if any information is missing or out outdated.

A Year of Diabetes Data is a Show & Tell talk by Doug Kanter that has been imported from the Quantified Self Show & Tell library.The talk was given on 2013/10/10 and is about Metabolism, Food tracking, Blood glucose tracking, Insulin intake, Heart rate, and Activity tracking.

Description[edit | edit source]

A description of this project as introduced by Quantified Self follows:

Doug Kanter is a Type-1 diabetic who’s been thinking about how self-tracking influences his diabetes control for a few years. While in graduate school at the Interactive Telecommunications Program (ITP) at NYU he started experimenting with visualizations that helped him understand his blood sugar and insulin dosing. In 2012 he began adding more data to his exploration in order to better understand how diet played a role in his diabetes self-management. Watch this great talk to learn more about Doug’s journey and his ongoing diabetes project.

Video and transcript[edit | edit source]

A transcript of this talk is below:

Doug Kanter

A Year of Diabetes Data

Hi everybody my name is Doug Kanter. My project is called Databetes. I’m very interested in finding new ways to use data to help diabetes patients to self-manage better. I do this in large part because I’m a diabetic myself; I’ve been living with Type 1 diabetes for 27 years. These are three of the devices I use every day to help m manage my condition. So I have two needles in my stomach at all times. One of them is hooked up to an insulin pump and then one of them is a senor hooked up to a continuous glucose monitor, and the third device is a standard blood glucose monitor. So I’m generating tons of medical data every day. I have about 300 data points a day just of this stuff, but as great as this is I think you really need to have the lifestyle data as well to help contextualize all these readings and help you learn from it. And so I always thought that if I can get all these readings together I can live healthier, and so throughout 2012 I decided to test that. For the entire calendar year I aggregated every single data point that influenced my diabetes control. I used all these devices and all these services as well. I had a Google doc spreadsheet of everything I ate and I had about 3500 entries, a couple of thousand pictures of my food, trained and ran a marathon, and as a result 2012 was the healthiest year of my life. These are my A1C blood test results and so the lower the better. 0.7, that red line across the top is the recommended level and I dropped almost a full point, and so I’ve never had reading as good in my life. As far as the lifestyle data goes, I think the food and exercise data is the most interesting. I’ve spoken previously about exercise data at the QS in New York, so this time I wanted to speak a little bit about the things I learned from my food data. And the first thing I wanted to look at was a low carb diet. My endocrinologist has never really pushed me to have a low carb diet, and I’ve never been drawn to it myself but I wanted to see if it worked. And so I took 100 days’ worth of my data and I created a data visualization called Insulin Onboard. So it has my insulin dosages along the bottom and has blood sugar readings across the top. So I wanted to show when the insulin is actually kicking in. it has a little bit of a latency, so this is like a small insulin dose over here on the left and then a large insulin dose on the right. and the way that I did the visualization is that the blood sugar you can’t see them if they’re in range, but they get darker and bigger the worse that they are. And so 100day worth of data had about 25,000 blood sugar readings and being able to print it out I did a 10 foot print out and was sort of able to step back and look at it. And what I noticed pretty quickly was that the low carb diet definitely worked for me. And days that I was doing that, here’s about a week of doing low carb it’s noticeably low blood sugar readings and then along the right is several meals worth. So that was pretty easy to draw the correlations there. And then this is another week and it’s not like my blood sugar was perfect, but if they did go a little bit out of range they quickly came back down. One of my worst patches was on Super Bowl Sunday, this was like on a Friday, Saturday, Sunday and this is me on nachos and beer and chicken wings and you can see there for a couple of days it wasn’t very good readings. Surprisingly of the 100 days I took from January 1 to April 9 my worst day was actually March 23, and it was a low carb day but it was also a day that I had a big presentation that stressed me out. And my blood sugar went up and I couldn’t get it to come down until after the presentation was over, so that was a sort of an interesting point for me. and it’s definitely one of the main takeaways is that a new future system would definitely have to factor in things like stress, but a low carb diet was definitely helpful. And I always thought glycemic index was something really important to look at, but with speaking to other patients I also learned about TAG, Total Available Glucose which is another way to look at it which was helpful. And the second way I looked at the data was this idea of a meal memory system. So for me it’s always frustrating to carbs count meals that I had eaten before, but I can’t remember of how much insulin I took and what happened with it. So what I did was I went back into the Google doc spreadsheet and looked at all the meals from restaurants. I had about 3,500 entries and I had 273 different restaurants that I ate out of over the year. And so I applied them across the year and saw I was that I was pretty evenly distributed. This is all my lunches and dinners; they’re colored black if I ate out that day. And I decided first to see whether this was even worth it. Was I eating out enough that it was worth the trouble of self-tracking all these things, and although almost my breakfasts were at home my lunch and dinner was considerable and so the total lunch and dinner was over 50%. So I decided that it was definitely worth it. Enough of my meals were happening when I was eating out. And so then I started to grapple those meals, so of all those different restaurants that I went to, and I noticed that there was about six of them that I went to to over a dozen times during the year and then it was a sort of classic power curve. The deli down the street I went to over 30 times. But there was a lot of places that I went to more than once but you know less than a dozen times and I really wanted to focus on those restaurants. And I saw that they took up about half of my meals that I went out. A quarter of them I ate there once and a quarter of them I ate their all the time. On example was this restaurant that served Cambodian sandwiches in New York called Num Pang, and I go there. Notice that I went there 10 times throughout the year pretty evenly distributed throughout the year. and being a creature of habit I tend to eat the same sandwich every time that I go there, which in this case is a brisket sandwich. And for some reason I always remember this a a restaurant and it’s a meal that always tends to throw me off for some reason. Even though I’m carb counting the rolls it’s still I always have bad results. So I wanted to create a system that I could get better at it. And so the first thing I did was just to take pictures of all of my meals and then I created a system where I could see the blood sugar across the top of my photos across the bottom. And I could quickly find the meal that I wanted to look at and also see which meals were spiking. And throughout the year I was able to create a system that by the end of the year I was dosing much better for a lot of these restaurants. And I also adapted from a web base and working at a mobile app which is much easier to get all of the location data. So it’s just taking a picture of the food, putting a blood sugar reading in, right when I eat it and two hours later when I get a text message to record the post meal, and so this system has helped me even more. And so I definitely think that this meal memory system there’s a lot of value there, and it doesn’t just have to apply for restaurants, I want to apply it in the future to recipes, and I’m also curious about whether the data that I’m generating can be shared among the whole patient community. Not just the carb data itself but also the process of doing it could be helpful.

I’m doing an office hour at 3 O’Clock in case anyone wants to talk more about this, but that’s my project thank you.

About the presenter[edit | edit source]

Doug Kanter gave this talk.