We saw a rising need in today's society for quality healthcare via technology. For many people, not only in the state of North Carolina but all over the US and around the world, access to complete medical services is not feasible, and thus we decided to build another path to help them determine their next steps in healthcare. In addition, many of our close friends and family have been drastically impacted by recent events such as Hurricane Florence and the California wildfires, so this is an issue that's close to our hearts.
What it does
Medic AI has several functions. The Facebook Messenger Chatbot is intended to help users narrow down potential causes of their symptoms so they can make more informed decisions for their approach to illness or discomfort that they may be experiencing. We also implemented two interactive data visualizations that provide greater insight into the availability of drugs and pharmaceuticals in the United States as well as historical trends on mortality. This is all compiled into an accessible, user-friendly website which includes resources through FEMA that provide more information on health case studies using machine learning.
How we built it
Facebook Messenger Chatbot
We used a Node.js back-end along with a Heroku server to implement our Facebook Messenger Chatbot. The Chatbot uses IBM Watson's NLP libraries to actively process user input and direct the user toward medical information and resources.
All data visualizations were completed in Tableau using public datasets from the National Center for Biotechnology Information (NCBI) and the National Center for Health Statistics (NCHS). All gradient color schemes and data analysis is fully integrated into Tableau.
Challenges we ran into
In the process of making the chatbot, we ran into several issues with testing. Additionally, we had difficulties connecting to the Facebook server on multiple occasions, and the NGROK library proved challenging for the back-end. We also had problems with configuring and setting up Heroku because we had no prior experience with a cloud server like it.
Accomplishments that we're proud of
We're amazed that we were able to carry out our ambitious plans in our limited time frame, especially with our combined inexperience. We faced many set-backs with the chatbot, so ultimately having a working version was extremely fulfilling.
What we learned
Coming into this hackathon as a novice team, we had absolutely no experience with anything related to chatbots, and minimal experience with Tableau data visualizations. However, we managed to bridge this gap by working together to solve problems and debugging various issues as a team. Ultimately, we needed to connect all of our separate skill-sets together, and finding that common ground really allowed us to progress. We were also able to experiment with new technologies such as Node.js, Google Firebase, MongoDB, and others.
What's next for MedicAi
We hope to expand our data visualization beyond the US to offer the same kind of services to other countries. We'd also like to provide more analysis on a case-by-case basis through the chatbot to come up with more accurate healthcare suggestions.