Inspiration
I am diabetic so when I was looking for my data just to see it and try and better my health it was unable to be found because I was on a linux machine. I thought about it and learned all the data is stored in easily accessible CSVs that as long as you have an account set up are downloadable on your phone. I decided to try and parse through the information to see if I could get everything I need out while also loading the information into graphs and showing how much your data has changed from 2 different data sets.
What it does
It allows the user to put in a CSV in dexcoms format that then gets parsed and averaged over the time period which allows for easy and concise viewing of the data that doesn't require a doctors office or even a full blown app. It is not a replacement for doctors checkups but it does help patients who are trying to better them selves see what has changed in a recent time period so they can see if they are on the right track
How we built it
I started by looking at what data structure dexcom has publicly available to it's users and getting the exact information out of that and then I went through that data and averaged all the different points at every time so the graph wouldn't look too crazy and be easy to read
Challenges we ran into
The data sets were very large being gathered every 5 minutes 24 hours a day the 14 day version had over 4000 data points that had to be parsed through quickly and dealt with. Another issue is that since this isn't start at X and continue on a constant upwards with it going up and down throughout the day and after meals I had to make the decision to use averages throughout each minute boiling down each time of the day to a single point in a 10 minute interval because of how many points were on the graph and if I went by minutes every time the dexcom transmitter is changed the clock resets so I had some data points where it was 3 or 4 points at a certain time of day because of how the clock worked causing for some wonky outliers.
Accomplishments that we're proud of
Learned how to use pandas better to organize the data set's easier along with some basic touching into Streamlit and Android-Studio to see how the implementation would work on those devices.
What we learned
How to parse through big data using DataFrames and get useful data out of big data sets
What's next for Blood Glucose Monitoring
The next step would be setting up a software that could learn the dataset and try to predict the information and give recommendations to the user based on recent history because of the bodies constant changes things like insulin to carb ratio needs to be constantly tweaked depending on a lot of things like stress, sickness, exercise amount, and just time. This could help people who are type 1 diabetics get a set up and be better equipped to managing their diabetes without as much risk or the need for constant doctors appointments.
Log in or sign up for Devpost to join the conversation.