Our main inspiration was that there are quite a few times when people can't access their nearby resources when they need it the most. People have to wait for a fair amount of time just to get the basic needs of mental health. And that is when our app comes into play.

What it does

Therapy ai is an online chatbot that gives people instant access to different medical needs. What sets therapy ai unique is that it can identify the user’s emotions and answers based on their emotion, it will give a helpful solution or advice to the user. For instance, if the severity of the user is critical, it will give the user access to their nearest health resource or connect them with a health expert. On the other hand, if the severity of the user is not so critical, it will give the user advice on what to do in the situation. And this really helps a user in difficult situations as they need help or advice on what to do.

How we built it

We used MERN stack (MongoDB, Express, React, Node.js) to build the web application and co:here to implement the AI chatbot feature.

Challenges we ran into

The main challenge that we ran into was trying to use co:here in our application since it was our first time using it. This was also our first time implementing a third-party API with our own custom API. Fortunately, by reading the documentation and asking questions to the co:here engineers, we were able to achieve what we wanted.

Another challenge we ran into was trying to merge the frontend and backend together.

Accomplishments that we're proud of

We’re proud that we managed to learn and implement a technology we’ve never touched before and got it to co-exist with a working application in essentially 24 hours (since we didn’t work on Friday).

What we learned

We learned a lot of new things within these 36 hours such as time management, communication, and leadership skills. And more importantly, we learned about the MERN Stack and using co:here API to make an automated chat bot.

What's next for TherapyAI

We want to improve the precision and accuracy of the response that the AI provides. In order to do this, we were considering further training the AI in two ways. Receive user feedback to check if the perceived emotion was correct and if the response was helpful Classify invalid user input (like gibberish) and handle it accordingly Use conversations from users to further train the AI

Share this project: