Inspiration
Current menstral cycle tracking apps don't go beyond tracking and we wanted to change that. The current apps tell users when their next period could be, but don’t provide meaningful information that women need to fully understand their situation. We wanted to create something that feels more personal, supportive, and actionable
What it does
Cycle Buddy reads real user inputs for an extended period of time and then uses ML to predict symtoms. It groups users into clusters with similar profiles, and gives personalized recommendations to help manage common challenges like fatigue, cramps, or mood swings.
How we built it
We built a pipeline that combines clustering, logistic regression, and natural language processing. User data is preprocessed and clustered using k-means. Logistic regression models predict next-cycle symptoms. Free-text symptom inputs are mapped to symptom categories using NLP techniques like embeddings and fuzzy matching.
Challenges we ran into
One challenge was handling messy user data, especially when users described symptoms in different ways. Another was balancing complexity with hackathon speed, we needed an ML pipeline that was powerful but also lightweight enough to integrate into Swift without long debugging sessions.
Accomplishments that we're proud of
We are proud that we built a full end-to-end system: user clustering, predictions, and recommendations all working together. We also managed to integrate our ML pipeline directly into the Swift app, creating a real interactive experience instead of just a demo.
What we learned
We learned how to combine classical ML models with NLP to make predictions more usable. We also realized that the hardest part of building a health app isn’t just accuracy it is also about presenting insights in a way that users can understand and act on.
What's next for Cycle Buddy
Next, we want to expand the dataset, add wearable integration, and make the system learn from user feedback on recommendations. Our goal is to turn Cycle Buddy into a platform that feels like a personalized health companion, not just a tracker.
Log in or sign up for Devpost to join the conversation.