LabiResult

Track

Healthcare

About me

I am a Student, and for transparency, I just developed it today, February 28, 2026

Quick Docs

I have prepared a docs for understanding the problem, solution, architecture, and impact of this project. Feel free to visit: https://docs.google.com/document/d/1G5AFKSu5t4ZG5svZ45Lh_A8RWXjmwo8qXFtInT9MwA0/edit?usp=sharing

Inspiration

The inspiration for this project is the motivation to streamline the process of understanding complex medical information. I wanted to bridge the gap between the jargons for normal people to understand. However, I am not positioning this as an alternative to medical professionals' advices, but however, just a quick tool with privacy and understanding in mind.

What it does

A desktop GUI application that extracts text from lab report images using OCR (Docling), analyzes the results with a local LLM (Llama 3 via Ollama), and optionally emails a formatted summary to a recipient.

How we built it

Built with Python, GUI with CustomTkinter, Docling for OCR, and local LLM (Llama 3) pulled with Ollama

Challenges we ran into

Local processing. I do not have a high-end laptop that would allow me to run extremely large models so I had to experiment with small models. The choice to go with Llama 3 is the result of testing different small models because some of them hallucinate and set normal values as out of range.

Accomplishments that we're proud of

This is my first time fully working locally. Since medical record is very sensitive to Data Privacy issues, I really wanted it to become local-first and it works!

What we learned

I really learned a lot starting from OCR to installing local LLM with Ollama to VRAM management. I managed to make it work with my laptop!

What's next for LabiResult

The next step for LabiResult is to actually partner with clinics so that instead of sending it to some random email, users will have a list of verified email for quick summary.

Built With

• customtkinter
• docling
• llama
• python
• redmail
• tkinter