HealthPort AI

Inspiration

Some of our teammates have family members who regularly get blood tests and other health related tests done. There are a lot of values and information on the reports which may take time to learn about.

What it does

Scans an input pdf and gives an analysis generated by Gemini AI

How we built it

We built it using React.js to code the website, Vite framework to develop the webpage, FastAPI for file uploads, Tesseract OCR for processing the lab documents uploaded to text, and Gemini API to provide analysis of the pdf report. In addition, we used MongoDB as our database for storing the user and lab report information. We deployed the frontend of our webpage using Vercel and the backend through Render. We also used Cloudflare for future scalability.

Challenges we ran into

Throughout this project, we ran into multiple issues mainly due to the steep learning curve as our team members had limited experience with MongoDB. Apart from that, we ran into issues while hosting the webpage on our domain, but we were able to resolve these issues with the help of online resources and LLMs such as Gemini and ChatGPT. We also ran into a lot of merge conflicts which required us to learn more about resolving conflicts and merging branches to deploy our code.

Accomplishments that we're proud of

This project was very fun to develop. We learned a lot and also accomplished a lot. Our main accomplishment is being able to get accurate analysis of example blood reports. In addition, we were able to host our website on our domain after long hours of effort.

What we learned

In the process of developing the project, we learned a lot. We learned more about web development using React, we learned database management using MongoDB. In addition, we learned how to use Tesseract OCR for file processing. All in all, this hackathon was a great learning experience.

What's next for HealthPort AI

In the future, we would love to develop the backend better. We would also love to include different kinds of scans and report analyses on our website for more accurate and detailed analysis.