mentalist

Inspiration

We were inspired by the belief in technology that it would assist us a lot in mental health. We believe that by using state-of-the-art NLP machine learning would profit our progression in exploring potential mental issues.

What it does

We built a Mental Health classification app via user's input text to recognize if that person has the tendency of 1 in 5 common mental issues "Bipolar, ADHD, Anxiety, Depression, or PTSD", using DistilBERT and Logistic Regression

How we built it

The built is specified in the README file on our github repo

Challenges we ran into

The complexity of the NLP model and the new web app framework we were trying to utilize (DJango and React)

Accomplishments that we're proud of

We finished our baseline NLP model and our most basic feature of our web app

What we learned

A lot of things around designing a model, the project architecture, and the new exciting techonology

What's next for mentalist

We wanted to add more features in our web app as well as increase our model precision even further

Built With

Submitted to

StormHacks 2022

Created by

I read some of the state of the art natural language processing papers and came up with the idea of using DistilBERT and Logistic Regression for our Mental Health Classification task. I implemented the model with the cross validation accuracy of > 85% and highly efficient execution. I helped with integrating the project together between the UI and the model.

Viet Anh Duong
I worked on the web application and tried to integrated AssemblyAI to use its speech-to-text feature to record the user's voice as input to the NLP model. The feature is still buggy at the end and doesn't work properly, but I felt eager to learn the API, it was an interesting API to me. I think it's a nice and great experience for me overall.

Long
Most of my time this weekend was spent on setting up and building a web application to communicate and deliver NLP Classification results in order to help people who feel unwell mentally identify possible mental diseases in early stages, and may potentially help psychiatrists speed up clinical diagnostic procedures. The web application is a REST API built with Django REST Framework and Python for the backend server, while connecting to the frontend app built using React and TypeScript.

Ha Thu Nguyen

Updates

Viet Anh Duong started this project — Feb 20, 2022 11:55 AM EST

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.