Bipolar is a mental health condition that affects %1 of the population in Canada which amount to 3.7 million people. One in eight adults will show symptoms of a mood disorder. My father was diagnosed with bipolar disorder and refused medication on the basis that he didn't feel himself. A standard medication for bipolar is lithium which is known to have this effect. Because this is a problem that affects a large number of Canadians, we wanted to develop a non-medicated solution. Additionally, Engineering is known for having a high rate of depression and we wanted to give people a tool to assist in finding activities that can improve their mental health.
What it does
Our application is a one time set up device that uses AI to collect a rolling average of your mood everyday to create a mood chart. A mood chart is a tool used now for helping people with bipolar determine when they are in a manic or depressive swing. It's can also be used to identify triggers and predict potential swings. By prompting people to log about what happened to them during an extreme mood, we can better identify triggers for mania and depression, develop methods against them, and predict upcoming swings. Our application includes a function to log activities for a day which can be later examined by a user, psychiatrist, or AI. We wanted to include a notification system that could ask you how things were going (like a good friend would) when you swung out of your neutral state to do this.
How we built it
We built a device using the DragonBoard 401c and developed a node application (to run on the dragonboard linux or a laptop windows machine) that can take pictures of a user and determine their emotional state using the Azure Face API. We collect data periodically over the month (a few shots a day) to get a better representation of your mood for that day without the hassle of having to log it. If you are very depressed or manic, it can be hard to maintain a routine so it was essential that this device not require consistent attention from the user. The mood chart is managed by a node server running on Huroku. We then have an andriod app that allows you to access the chart and see your moods and logs on a calendar.
Challenges we ran into
It was our first time setting up a dragonboard device and we initially didn't have a way to interface with it due to a mistake by the MLH. We lost an hour or two running around to find a keyboard, mouse, and camera. This was the largest UI project Gary Bowen has ever been in charge of and we ran into some issues with formatting the data we were transmitting.
Accomplishments that we're proud of
June Ha is proud of setting up a public server for the first time. Gary is proud of the implementing a smiley faced dropbox and all his debugging work. I'm proud of setting up the capture software on the Dragonboard and my laptop. And I'm proud that we managed to capture data with a cheap webcam on a dragonboard, use the mircosoft api to get the emotion, send that data back to our server for processing, and delivering all that in a handy mood chart on your phone.
What we learned
I learned a bit about working my way around a linux machine and the Dragonboard 401c. Gary learned a lot of good debugging tools for android development, and June learned he should not be awake for 24 hours.
What's next for palpable
We'd like to properly implement the notification function on our app. Develop a more cost effective wireless camera device for processing emotion. We'd also love to implement additional AI for smart suggestions and to possibly diagnose people that have a mood disorder and may not know it.
https://www.canada.ca/en/public-health/services/chronic-diseases/mental-illness/what-should-know-about-bipolar-disorder-manic-depression.html https://en.wikipedia.org/wiki/Population_of_Canada https://www.canada.ca/en/public-health/services/chronic-diseases/mental-illness/what-depression.html