Inspiration
LunaTrack was inspired by my personal experience. I was diagnosed with uterine fibroids and eventually underwent a hysterectomy. During every gynecological visit, doctors asked if my menstrual bleeding was heavy — but I never knew how to answer. There was no clear standard for “normal” which I could easily use and discussing menstruation openly, even with friends or family, felt too private. When I finally learned I had been experiencing menorrhagia, I was shocked. I realized that many people share this problem without realizing it, and I wanted to create a tool to help detect it earlier — safely and privately.
What it does
LunaTrack estimates menstrual blood volume from photos of used sanitary pads. It uses a local AI model that processes everything directly on the user’s device, ensuring that no image or data ever leaves their computer. The goal is to empower users with objective, privacy-preserving information that could support early detection of heavy menstrual bleeding and improve communication with healthcare providers.
How we built it
This is my first Chrome Extension project. I built the base architecture using Claude Code for code generation and debugging support. The AI model runs locally by using Prompt API, analyzing color and absorption patterns to estimate volume. The interface was designed to make data recording and visualization simple and intuitive.
Challenges we ran into
The main challenge was calibrating the AI model. It tended to overestimate the blood volume, so prompt-tuning and testing required time and care. Ensuring local AI performance while maintaining accuracy was also technically demanding.
Accomplishments that we're proud of
I’m proud that LunaTrack can run entirely offline — protecting user privacy while providing potentially life-changing insights. It turns a deeply personal issue into actionable health awareness.
What we learned
I learned how powerful local AI can be for sensitive health applications, and how design decisions must always respect users’ privacy and dignity. I also deepened my understanding of Chrome Extension development.
What's next for LunaTrack
The next goal is to improve image detection accuracy through better prompting and/or other method and dataset refinement. I also plan to explore menstrual pattern tracking and total menstrual bleeding per cycle — always keeping privacy at the core.
Log in or sign up for Devpost to join the conversation.