Inspiration
During the COVID-19 pandemic, fear of entering healthcare spaces has been rising as patients try to avoid exposure and stay at home. While staying at home has been beneficial in keeping transmission risk low, what if a patient has intense back pain and doesn’t know whether they should go in-person or seek a telemedicine appointment?
This decision to go or not go to the clinic is normally made based on a patient’s comfort level, but it should really be made healthcare providers who have the expertise to know what type of appointment patients need. But during this time of healthcare worker shortage, healthcare providers don’t have the time to call patients and guide them through these decisions. This is where our mobile app Polaris comes in.
What it does
Polaris is an AI chatbot that will recognize a patient's entered symptoms or needs, such as prescription refills, and through a series of personalized questions, correspondingly provide recommendations on whether the patient should schedule an in-person visit, book a telemedicine appointment, or go to the ER or urgent care immediately.
How we built it
To create a chatbot that would triage patients based on symptoms, we looked into platforms such as Google's DialogFlow, IBM's Watson Assistant and AWS Lex. Using Android Studio, we created a messaging UI that would integrate with our chatbot.
Challenges we ran into
While our team has some programming experience, we were all new to chatbot platforms such as Google's DialogFlow and AWS Lex and experienced a steep learning curve in how to work with them. We ran into challenges in creating the backend of the chatbot AI and integrating it with our Android Studio messaging UI. Luckily, the mentors at MedHack (special shoutout to Steven Gomberg and Dr. Vishnu Ravi) were incredibly helpful in idea formulation and guiding us towards resources!
Accomplishments that we're proud of
While we weren't able to get all the features of Polaris that we had initially planned, we are really proud that we learned how to utilize AWS Lex from the ground up and create a chatbot that could triage patient symptoms.
What we learned
We learned how crazy it is to build an app in 24 hours, gained experience with Android Studio and AWS Lex, and can now better understand chatbot functionalities. Above all, we learned the power of teamwork!
What's next for Polaris
We plan to add infrastructure to support a directory of recommended clinics and providers after a patient is successfully triaged based on their symptoms. These suggestions would be tailored to match the insurance plan and geographic proximity of the patient. How? We’ll be doing this by implementing the ability of users to make profiles, uploading their insurance details, and previous medical history, to ensure a more personalized and accurate recommendation and enable more longitudinal support. Lastly, we hope to expand our database of common symptoms and conditions, and what turns these often benign signs into dangerous red flags to ensure, again, that patients are getting only the most accurate recommendations we can offer them.
Log in or sign up for Devpost to join the conversation.