Inspiration: We don't talk about depression as much as we talk about other diseases. Depression and bad mental health have been ignored as a serious issue since ages. So, we decided to make something that can help or support people suffering from it. We believe people generally doesn't openly discuss about their mental state so a good solution will be where the user of app can get the appropriate support by staying in his/her comfort zone.
Learning: Firstly, we get to know about the seriousness of this problem by going through some of the latest related reports. We learned how patients are assessed by doctors using methods such as asking questions which we ultimately incorporated in our app. Other than this, we learned to build a project in constrained and competitive environment.
How we built: Firstly, we trained 3 supervised machine learning models and evaluated each of the 3 and saved the model which gave us the best results in as a byte-stream file. Now, we built a form that consists of 9 choice-based questions which when submitted passes the responses through the model to give the appropriate severity level. This severity level is shown on the result page and also simultaneously passed to a function to provide the corresponding recommendation.
Challenges faced: Our very first challenge was to find a relevant dataset that can be used for training the model. Once, we get the dataset, we were successfully able to train the model. For backend, we initially used Express.js but couldn't work out with it so we moved to Flask. Now to get the user responses at the frontend, we had initially decided to use typeform but we realized that we had limited functionality available as a Non-pro account user so ultimately had to discard this idea and built the form from scratch.
