cgm-dashboard

Plotly Dash based dashboard overlaying your CGM data on your CGM percentile plot. Designed to work with nightscout data and supports mongodb and REST.

what

Who wants to check his/her CGM report? Nobody!

Still, many CGM users benefit from looking at their report, especially the Ambulatory Glucose Profile helps to quickly see at which time of the day your therapy needs attention.

cgm-dashboard was developed to make seeing your data simple and fun. Instead of limiting yourself to today's data or the AGP in your report, it combines them both. It brings your glucose data into perspective. On top of that, you can quickly take a glance on how you perform without downloading the report from your device manufacturer or opening an app.

cgm-dashboard overlays your today's cgm data with your historic percentile plot. This allows you to quickly foresee if you enter a problematic time of the day. You can think of it as using the percentile plot as a forecast, based on your history.

how to use it

The webbased dashboard can be accessed from any browser making it the perfect candidate for a left-over tablet. Just put it where you see it, anywhere in your home.

how it works

CGM Dashboard is a python based webserver that you can reach from any webbrowser. It uses plotly's dash to create a beautiful, minimalistic, java scrip based dasboard. You can run it from any machine such as your computer, or a raspbery pi as simple as calling

python app.py

And opening the dashboard from your browser

http://0.0.0.0:8080 This will work when you access the website from the same machine running the code. If you want to access it from a different machine in your network, simply replace 0.0.0.0 with the ip address of the computer running it. Also, the port 8080 might be different. Check the log when running python app.py which will output the address.

app.py

Contains all plotly dash code. It configures the graphs and connects it to the data source.

cgm.py

Provides access to the stored cgm data and performs pre-processing for visual representation.

database.py

Handles the data storage and access. Uses adapter class to request more data from remote service.

database.py

Handles the data access from remote services. For now, mongo database access and REST calls are supported.

config.ini

Here, you need to fill in your backend credentials.

installation

1. Checkout the git repository git clone git@github.com:janvv/cgm-dashboard.git

2. Downlaod the required python3 packages sudo pip install -r requirements.txt

3. Fill in your nightscout credentials ...

   [REST]
   domain = <the domain of your REST service>
   port = <the port of your REST service>

   or ...

   [MongoDB]
   user = <your mongo db user name>
   password = <your mongo db password>
   host = <your mongo db domain i.e. testuser.mlab.com>
   database = <the name of the mongo database>
   port = <the port of the mongo database>

Start upon boot

If you want the service to run in background all the time (i.e. on a raspberry pi), you can create a cronjob that starts the webserver upon boot.

First create a shell script to run the serbvice in background mode and save it as cgm-dash.sh

#Q/bin/sh
#cgm-dash.sh
cd ~/git/cgm-dashboard
python ./app.py

Then, create a cronjob to execute the shell script on boot.

1. Open cronjob file

crontab -e

2. Add following line

@reboot sh <path to shell script> <path to logfile> 2>&1

Note: 2>&1 directs error logs into the log file.
I put the tool on my raspberry pi and it looked like this. The ./venv/bin/python is because I used a virtual environment to install only the packages that I really needed.

cgm-dash.sh

#Q/bin/sh
#cgm-dash.sh
cd ~/git/cgm-dashboard
./venv/bin/python ./app.py

And the line in the cronjob file

@reboot sh /home/pi/cron/cgm-dash.sh > /home/pi/cron/cgm-dash-logs 2>&1

Trouble shooting (raspberry pi)

On raspberry pi, running the tool might result in a numpy error. Something with libatlas. For me it worked to reinstall numpy and install libatlas-base-dev again sudo apt-get install libatlas-base-dev