Inspiration
#### Everyday people suffer from new ailments without being able to be diagnosed by primary care doctors. They are then stuck with the co-pay or expensive medical bills provided with just a five minutes meeting. After all that time and money invested, one comes back with a inconclusive result. LangChain continuously is trained with new pieces of data, so symptoms that doctors may not be able to immediately diagnose can be slowly diagnosed using a vast amount of data across multiple regions of the world.
What it does
#### Individuals can add in documentation based on their symptoms, the contracts of the file are then read and the user can add-on by typing in their symptoms as well, acting as a supplement to the document content already extracted
How we built it
- PyPDF2 - python library that allows one to read pdf documents are extract content
- streamlit - python web ui interface
- langchain - large language model that allows for one obtain responses on a large dataset of questions
- openai -AI framework that focused on using chatbot responses
- os - python library that allows one to obtain information about the operating system files
- pickle - python library used to save the pdf files extract content models ## Challenges I ran into
- Adding in multiple PDF files at a time, currently one has to implement files one by one
- Keeping track of different file extensions and extract content from all of them simultaneously
## Accomplishments that we're proud of - the ability to provide individuals with different forms of medical information based on the documents provided
- the ability to create a simple but intuitive UI for individuals to navigate and obtain information correctly ## What's next for Medibot
- multiple file upload
- categorising files based on different file extensions
- setting up a connection between the LLM response and doctor office communication
- scheduling appointment based on severity of illness
- finding nearby places for treatment if the symptoms are to severe to wait for an appointment
Log in or sign up for Devpost to join the conversation.