Multi-Agent Project

Healthcare Assist

Inspiration

Under/Over investment in healthcare is a major challenge today across the world. Whilst accessibility to healthcare is debated more, the fundamental issue of investing into core healthcare remains the root of the problem.

What it does

Allow government agencies to decide on where to invest based on open data and proprietary data available in data.gov and other relevant domains

How we built it

We first looked at mimilabs data for enrollments across state and cities from a speciality disease angle. We then ask AI agent to categorize this into defined 4 categories - Cardio, Renal, Mental, Diabetic. We then provide this data to another agent which then looks at the population data from data.gov and a dummy data we created for city and state investments for 2024. The AI agent then recommends investment spread across cities assuming the budget is same as 2024.

Challenges we ran into

Understanding the mimilabs data. We used mimilabs bot for same. We then had to use Databricks to prepare the data in a standard format to feed to agent. We had to scan relevant information from data.gov for population data. We couldnt find the investment data into healthcare so had to create a dummy data for same

Accomplishments that we're proud of

Building a working prototype using Databricks AI capabilities.

What we learned

In AI world - Getting the right data, asking the right prompts is the key.

What's next for Multi-Agent Project

We intend to now create multi-agent architecture which scans more key open datasets to allow government agents to chat with data and write automatic policy recommendations which are data driven.