TXT811

Using Cohere's natural language processing models, we made reaching out for medical inquiries much more accessible, on the latter of calling 811 and having to ask a busy human for advice.

Comment

Inspiration

After learning about NLP and Cohere, we were inspired to explore the capabilities it had and decided to use it for a more medical oriented field. We realized that people prefer the internet to tediously needing to call somebody and wait in long hold times so we designed an alternative to the 811 hotline. We believed that this would not only help those with speech impediments but also aid the health industry with what they want to hire their future employees for.

What it does

We designed a web application on which the user inputs how they are feeling (as a string), which is then sent onto our web server which contains the Cohere python application, from which we ask for specific data (The most probable illness thought by the NLP model and the percentage certainty it has) to be brought back to the web application as an output.

How we built it

We built this website itself using HTML, CSS and Javascript. We then imported 100 training examples regarding symptoms for the natural language processing model to learn from, which we then exported as Python code, which was then deployed as a Flask microframework upon DigitalOcean’s Cloud Service Platform so that we could connect it to our website. This sucessfully helped connect our frontend and backend.

Challenges we ran into

We ran into many challenges as we were all very inexperienced with Flask, Cohere's NLP models, professional web development and Wix (which we tried very hard to work with for the first half of the hackathon). This was because 3 of us were first and second years and half of our team hadn't been to a hackathon before. It was a very stressful 24 hours in which we worked very hard. We were also limited by Cohere's free limit of 100 training examples thus forcing our NLP model to not be as accurate as we wanted it to be.

Accomplishments that we're proud of

We're very proud of the immense progress we made after giving up upon hosting our website on Wix. Despite losing more than a third of our time, we still managed to not only create a nice web app, we succesfully used Cohere's NLP model, and most notably, we were able to connect our Frontend and Backend using a Flask microframework and a cloud based server. These were all things outside of our confortzone and provided us with many learning opportunities.

What we learned

We learned a tremendous amount during this hackathon. We became more skilled in flexbox to create a more professional website, we learned how to use flask to connect our python application data with our website domain.

What's next for TXT811

We believe that the next step is to work on our web development skills to create an even more professional website and train our NLP model to be more accurate in its diagnosis, as well as expand upon what it can diagnose so that it can reach a wider audience of patients. Although we don't believe that it can 100% aid in professional diagnosis as that would be a dangerous concept to imply, it's definetly a very efficient software to point out warning signs to push the general public to reach out before their symptoms could get worse.

Share this project:

Updates