ClearCare

Inspiration

As we delved into the complexities of healthcare plans, we quickly realized just how overwhelming and confusing they can be. This inspired us to create a solution that simplifies the process—ensuring that everyone, regardless of income, or health status can confidently navigate, estimate, and select the best coverage for their needs. By harnessing data-driven insights and AI-powered guidance, we aimed to transform healthcare decision-making, making it more transparent, equitable, and personalized for all.

What it does

Our system efficiently matches users to the most suitable healthcare plans through a tailored questionnaire that assesses both health and financial factors. User responses are processed by a RAG-backed generative AI model, leveraging the U.S. Healthcare dataset from the Healthcare track—specifically the Star Rating of Healthcare Plans with HOS-CAHPS measures—to rank plans within the context of each user's unique background. We then provide the top three recommended healthcare plans, complete with estimated yearly costs and a breakdown of pros and cons for each option, ensuring informed and confident decision-making.

How we built it

Our solution is powered by a modern full-stack architecture, combining a responsive frontend, scalable backend, and AI-driven intelligence to deliver personalized healthcare plan recommendations.

Frontend: We built the frontend with React and styled it using TailwindCSS, ensuring a clean, user-friendly interface with seamless navigation.
Backend & Database: We implemented the backend using FastAPI for high-performance API handling, while Firebase managed authentication and data storage.
AI & Retrieval: We leveraged OpenAI’s embeddings and Pinecone hosted on AWS cloud services for fast, efficient vector search, enabling RAG-backed generative AI to process user inputs and retrieve relevant healthcare plan insights.
Structured Data Handling: We used Pydantic objects to ensure structured, validated output, making responses more reliable and interpretable.

This tech stack allows us to match users to the best healthcare plans with speed, accuracy, and transparency, bridging the gap between complex healthcare data and user-friendly decision-making.

Challenges we ran into

With four team members working simultaneously, ensuring alignment on a centralized design structure was a significant challenge. We needed to seamlessly integrate our model’s output with the frontend, maintaining consistency through well-defined data structures and API endpoints.

Another major challenge was the model itself. Our initial approach involved clustering users based on dataset information to classify them into groups for healthcare plan recommendations. However, upon realizing that patient-specific data was unavailable, we had to quickly pivot. This led us to explore alternative strategies, ultimately driving us to adopt a RAG-based model, which allowed us to leverage existing healthcare data more effectively for personalized recommendations.

Accomplishments that we're proud of

Our greatest strength was how we combined our domain-specific expertise to build a seamless, well-integrated solution. Throughout the weekend, each of us faced and overcame individual challenges, contributing to both our personal growth and the success of our project.

What we learned

Our team split responsibilities between frontend and backend development, allowing us to specialize and grow in our respective areas. The frontend team explored and implemented new UI/UX design strategies, ensuring a user-friendly experience, while the backend team focused on writing clean, structured code to maximize reliability and efficiency. By combining our efforts, we successfully built an application that seamlessly integrates both components, reinforcing the importance of collaboration and technical synergy.

What's next for ClearCare

While we accomplished a lot, there are several features we envisioned but didn’t have time to implement. Moving forward, we aim to enhance medical cost budgeting tools to help users plan their expenses more effectively. Additionally, we want to develop a more comprehensive and accessible summary of healthcare plan options, allowing users can easily compare and understand their choices.