The Team

COVID Chat was created by 4 first time hackathon-goers, Sruthika Baviriseaty, Eileen Duong, Andrei Mitrofan, and Shubha Vasisht for PennApps 2020. This project was worked on remotely from Tampa, Florida; Los Angeles, California; Houston, Texas; and Princeton, New Jersey over the course of three days.


The concept behind the hack was to increase awareness of general information and resources for the COVID-19 virus that has significantly impacted everyone’s lives. In the midst of current events, there are several sources of misinformation that lead to fear-mongering, distrust, and apathy. We wanted to develop a resource that could direct users to credible sources and sites that could improve understanding and if necessary, prompt action. Given the current circumstances, we wanted to create a chatbot that helps connect people with and without symptoms of COVID-19 with personalized chat and resources. We were inspired by Google Cloud's interactive and personalized space and wanted to create something that is useful for people that would like a quick check for COVID-19 symptoms.


The chatbox application is initiated when a user opens the chatbot window. The chatbot greets the user and prompts them for input. This starts a conversation between the user and chatbot, with each user input, prompting follow up questions from the bot. Using both linear and non-linear dialogue, the chatbot iterates through a series of Intents and actions, that allows for a natural, uninterrupted conversation. The inputs from the conversation are then compiled to assess COVID-19 related needs and direct users to resources that best fit their conditions.

The COVID chat was created primarily with the Google Dialogflow interface and integrated with to enhance usability and functionality. The chatbot was then embedded into a React App, which was then published through the Heroku platform and connects to Google Cloud .

Features and Take-Aways

A key part of our project was the user interaction. Throughout the coding process, we had to be consider how a user can approach a prompt. We learned through trial and error that there are several different ways to answer a simple question, varying with phrasing, syntax, and slang. Using aspects of Machine Learning and Artificial Intelligence from Dialogflow, the chatbot learns to respond and react to the different ways a user may answer a question and allows the program to proceed and execute the tasks accordingly.

Future Applications

The concept of a chatbot can be applied beyond just this pandemic. While the scope of our project focuses on COVID-19, the idea behind it can be used in healthcare, by health care professionals, to efficiently diagnose or evaluate patients virtually. The concept can be optimized for televisits as a medium for pre-diagnosis and can be extended to store patients’ data. This is especially beneficial for increasing healthcare access as it eliminates the need for mobility, allowing more people to receive care.

Built With

Share this project: