Smart Depression Pre-Detector

Inspiration

I personally have depression. My family thought I didn't, just "not tough enough". When I was diagonized it was too late.

What it does

Learns behavioral trends from healthy people and gives out depression when monitored individual's behaviors deviate in the direction that could be depression (but not in other cases).

How we built it

Using LSTM, Learn population trend first, fine tune on personal level, and then test the test-set data on it. Disclaimer: due to lack of real life data, and the nature of this project being a proof of concept, the data were generated using GPT with principles of Brownian motion.

Challenges we ran into

Dealing with the fact that in real life no one would be labeling new data for us. Thus LSTM, we learn trend of healthy behavior, don't try to interpret why.

Accomplishments that we're proud of

Being able to finish a minimal viable prototype in time

What we learned

a lot

What's next for Smart Depression Pre-Detector

In presentation

Built With

• python
• pytorch