MedBot: a Medical Chatbot

Example conversation with MedBot

Inspiration

AI is advancing and becomes more widely used across many different industries and applications. Chatbots have gained traction amongst banking, customer service, and e-commerce. We wanted to create something similar but for medical purposes. This would help to reduce time spent commuting to the doctor's office, less money spent on treatments and care, and easier access to medical information.

What it does

Medbot is a Facebook Messenger chatbot used to answer medical built-in commands, greeting prompts, and basic prompts. The built-in commands:

/emergency <location>
/symptom <list of symptoms>
/disease <disease name>

/emergency command returns the closest hospital. /symptoms command returns possible disease/condition based on symptoms input. /disease command returns information about the user's disease input. Medbot will also basic medical prompts.

Basic prompt example

How we built it

We used Facebook's messenger API for the front end interface. The endpoint was created using Python/Flask and hosted on Heroku to receive and process Webhook notifications.

Challenges we ran into

Facebook Messenger API does not allow a working local version, we were required to host to run and test the app, so there were difficulties with testing and it was a slow-going process.
Adjusting the OpenAI davinci model’s temperature and edits to return not only relevant, but also helpful responses to the prompts it received.
Currently the different models we use all stem from OpenAI which has been having scaling and uptime issues therefore this chatbot is not optimized and may not be able to handle a high volume of queries.

Accomplishments that we're proud of

This is our first time working with the Facebook Messenger API and OpenAI's API. We're proud of creating a project using new technologies in a short time frame.

What we learned

The basics of natural language processing and how it can be applied to create a chatbot that can understand and respond to user input.
The importance of a temperature/accuracy with AI models and the challenges that come with working with sensitive medical data.
The potential applications of AI and NLP in the field of healthcare, and how these technologies can be used to improve patient outcomes and streamline healthcare processes.
The importance of working in team and effectively communicate with team members to achieve the project's goal

What's next for Medical chat bot

Work on User Interface design for a smoother experience.
Expand the functionality of the Medbot to assist users with a wider range of medical queries.
Collaborate with medical professionals and institutions to ensure the accuracy and relevance of the information provided by the Medbot.
Increase the reach and accessibility of Medbot by promoting it on various platforms and making it available in multiple languages.
Increase user engagement and satisfaction with the Medbot by regularly gathering feedback and implementing improvements.