med-bot

• 

  med-bot response example-1

• 

  med-bot response example-2

Inspiration

Main inspiration for this that we can't when we have the medical emergency. Medical Emergency can't come with hint. so we have to prepare when ever there is the emergency come.

What it does

I used the azureOpenAi api to make the RAG model that is well trained on the large medical data. which gives the accurate results to questions.

How we built it

used the on device embedding model to embed the data to embedding values these values are stored in the pine cone data store which helps in doing the similarity search and based on the results of similarity search the AzureOpenAi make current sentence that matches with the input.

Challenges we ran into

1) storing the large data into pinecone vector store. 2) building the similarity search 3) building the model

Accomplishments that we're proud of

Achieved best in class model that performs only on the medical data and questions that not related to the medical field will give "Not related to me" answer.

What we learned

My first project in generative Ai field , successfully completed the project by learning new thing like pinecone vector store, AzureOpenAi integration, similarity search,

What's next for med-bot

I think it will be upgraded with some real image examples and practical remedies for the problems that needs med-bot

Built With

• flask
• generative-ai
• python
• similarity-search
• vector-stores