HealthcareGPT

Transform your data queries with our AI-driven chatbot, seamlessly integrating Databricks' powerful analytics for instant, accurate insights.

Comment

Inspiration

The inspiration for HealthAssist stemmed from the urgent need to improve accessibility and quality of healthcare information. By harnessing the power of advanced AI, we aimed to create a tool that can provide immediate, reliable medical advice and information, reducing the burden on healthcare systems and empowering individuals with knowledge about their health.

What it does

HealthAssist is an AI-powered chatbot designed to offer instant medical and healthcare-related guidance for patients and healthcare professionals. It can answer a wide range of queries from disease symptoms and treatment options to preventative healthcare and wellness tips, making healthcare information more accessible and understandable.

How we built it

We built HealthAssist using Databricks' vector search and model serving capabilities integrated with a large language model (LLM), which is Llama3 70B. We developed a pipeline that processes user queries, retrieves relevant medical information using vector search technology, and generates coherent, context-aware responses through the LLM. The interface is powered by Gradio, allowing easy and interactive user engagement.

Challenges we ran into

We are given AWS free tier but the free space usage was quickly ran out of space no matter how small the datasets that we prepared, before we let it charge our bank account, we had to stop the model tuning section. Also our 14 day free trial is running out and we had to move everything from one accunt to the other account.

Accomplishments that we're proud of

We are particularly proud of the chatbot’s ability to understand and process complex medical questions and provide answers that are not only accurate but also easily understandable by the general public. The system's performance in terms of speed and accuracy under various tests has been a significant achievement.

What we learned

Throughout the development of HealthAssist, we deepened our understanding of natural language processing and Retrieval augmented generation applied to the medical field. We learned about the nuances of medical terminology, the importance of data privacy in healthcare applications, and the potential of AI to transform traditional industries.

What's next for HealthAssist

We plan to work on the history setup for the chatbot so it can also learn from user's previous interactions with the chatbot. We are also planning to buy more space for databricks and AWS to run so that we can have a much strong and faster chatbot model that is marketable.

Built With

Share this project:

Updates