SentiMate is a suite of tools that have been developed to fight depression. We have developed a machine learning model, that can detect depression using textual data, for example tweets. We have also built a micro-blogging web-app, StuClan, that provides a platform for students to talk about specific topics like mental health, education and much more. With these platforms especially dedicated to students, SentiMate aims to bring an end to depression through technology.
Sentiment Analysis is popular among natural language processing (NLP) researchers,
but depression detection using NLP is not as common. Furthermore, there exists no
depression detection system, that can identify depressed users on social media
using just text. SentiMate does that; it offers an alternative to traditional surveys. This web interface is based on multiple machine learning and NLP
models, all developed by us, to make a system that can detect depression with an
accuracy over 96%. This makes the model not only unique, but also accurate for our purpose.
Apart from great accuracy, the model performs well on real world data as well. Below are some sample tweets that "capture what depression feels like" and the model does an excellent job in correctly identifying them.
-Do you ever have those days where no matter how much you force yourself to do things you still feel awful -every day i begin from feeling worthless and proceed from there -i'm sad -dude, life is depressing -even after all this i’m still afraid of sadness -she died as she lived…wondering whether she was tired or dying
Nevertheless, this tool is not a substitute to consulting professional help. The objective of this tool is provide quick and accurate identification of depression for free, based on which one may consult therapists and doctors.