We were inspired by the mental health crises on college campuses, where a soaring demand for mental health services is met by a sheer lack of resources at Counselling and Psychological Services (CAPS) centers across the UCs. We decided to work on making screening for urgent patients easier through voice calls, as hotlines are often a way therapists interact with at-risk people. Some of the tools and APIs available at this hackathon made this possible.
What it does
Our project records phone calls on a hotline, where the caller is prompted to answer a standard psychological evaluation question. While they are speaking, a graph of their sentiment is streamed, per sentence, to a portal on the evaluator’s side. This portal also gives insight on the overall sentiment and magnitude (amount of emotional content) in the call, which allows therapists to determine the urgency of psychological intervention for the caller.
How we built it
We used several APIs, languages and tools to build this portal for therapists to use. There is a Python/Flask backend connecting the Twilio API and Google Cloud Sentiment Analysis API, as well as a Dash frontend to visualize the sentiment data and analyses.
Challenges we ran into
We were all working with technologies that we had no exposure to previously, and so had to learn a lot very quickly while supporting each other in creating an intuitive experience that works very quickly to respond to callers. Such is the nature of hack - we #makewaves.
Accomplishments that we’re proud of
We’re really proud of the project we came out with. It is both complex and yet universally useful across college campuses strapped for mental health resources and an ability to gauge urgency of psychological intervention quickly. We are grateful for the support of company sponsors who showed us how to use new technologies. Ultimately, we worked on a very complex yet useful project and learnt a lot of new technologies while making new friends!
What we learned
Some of us were new to even hosting a server, but this activity functioned as a crash course. We learned that there were various ways to tackle difficult problems like this, and often our roadblocks came from meeting limitations in certain APIs or being unable to hack a solution that connected all the different components, like Dash, Flask, etc.
What's next for Mental Triage
We’d love to be able to share this service with mental health professionals across the UC system (we are collectively UCSB and UCLA students). We’ll definitely be taking back these wonderful memories and continue to #makewaves beyond SBHacks V!