Inspiration
Suicide is the third leading cause of death of young people between the ages of 15 and 24. Around 5,000 young people commit suicide in the U.S. each year. With these statistics we feel it is a very important issue to target and attempt to improve and reduce self harm attempts and overall deaths.
What it does
What our app does is that it takes in a text and is able to detect if the person is at risk for suicide. This will allow for parents or guardians to become aware of their child's issues and get them the necessary help.
How we built it
We used a logistic regression model with count vectorization and a tf-idf transformer that learned from a data set we found to train our model in order to be able to read and predict if a texter has potential for self harm.
Challenges we ran into
While running the logistic regression we would have some errors with our weights which would heavily impact or precision of our results thus skewing our accuracy.
Accomplishments that we're proud of
One accomplishment we have is that when testing our app/code we had a 93.89% prediction accuracy when testing it. Another accomplishment is that our app dynamically updates whenever new text is inputted.
What we learned
We learned how to train a logistic regression model with text instead of numerical values. In addition to this, we also learned how to build apps with React and used websockets to communicate with the server and the application.
What's next for Teen self help
This can be further integrated with more time to be able to run throughout a person's cell phone in order to track any messages or texts that could be related to self harm and will send a notification to their parents or legal guardian thus bringing awareness to them and providing them with the help/care they need.
Log in or sign up for Devpost to join the conversation.