Introduction:
Delphea (under Aurelia Vitals) isn’t just another period app. It’s a wearable + mobile experience designed to use biomedical data and machine learning to predict fertility windows, with the rigor needed for FDA clearance. As a founding product designer, I helped transform a rough CSV-export prototype into a scalable product with clear user flows, a design system, and approachable visuals
Estimated Reading Time: 5 minutes
Results:
74%
of test users reported Delphea would be useful to guide pregnancy planning/prevention choices
The Problem:
The Missing Link in Fertility Support.
For women trying to plan or prevent pregnancy, decisions often hinge on cycle predictions that feel unreliable. Irregular periods, missed logs, and confusing data make it difficult to know when to feel confident and when to be cautious. Existing tools often leave users overwhelmed with input requirements or under-informed with vague insights, which erodes trust.
The Solution:
Tracking temperature for accurate menstrual cycle predictions and other cycle-tracking features
Biometric wearable integration →
The app syncs with the new biometric wearable to reduce data entry
Pending approval to be the only FDA-approved wearable integrated birth control app
Machine Learning Predictions→
The app uses temperature data and other inputs to predict fertility levels in users using a machine learning model.
This method of prediction is more accurate that purely based off of user inputs due to the objective nature of the data that the wearable gathers
74%
of the 23 test users reported trust in Delphea to guide fertility choices.
Retained User Input & Logging→
In order to track other factors, such as pregnancy tests or period flow levels, the app also has functionality to input symptoms
Research:
No existing menstrual cycle tracking app combined FDA-level accuracy with automatic data collection and a friendly user experience.
User interviews (12) →
Biggest Findings:
Existing apps were either too clinical or too playful.
Daily logging was a chore
Privacy and transparency were non-negotiables.
55%
of interviewees wanted more accurate cycle predictions
Top Priorities:
Cycle Prediction & Prep
Monitor Sexual Health
Symptom Tracking
Competitor Analysis →
Flo required constant user input to make predictions. Oura leaned on biomedical tracking, but its ring-based tracking raised temperature-tracking accuracy concerns, and Natural Cycles differentiated itself with FDA clearance but still relied on user input. These insights guided us to position Aurelia as the trusted, credible, and approachable alternative.
Synthesis →
After conducting our research, we finalized on a goal for our product: to make complex, regulated predictions feel trustworthy and usable for everyday decisions about pregnancy prevention and planning.
Initial MVP Design Journey:
From messy exports to meaningful predictions.
Initial Planning and Ideation →
When I joined, the app flow was clunky: pair device → export CSV → AirDrop to computer→ lose your data because the app deleted it. Not exactly user-friendly.
So in order to create a functional app, we started by determining the necessary features, information architecture, and user flow:
Wireframing →
Given our planning of features, information architecture, and user flow, we began creating lofis to begin visualizing our product.
Lofi to High-Fi →
We tried out different ways to visualize cycles: at first, simple lists, then calendars that highlighted ovulation and period predictions, and finally graphs that revealed trends or flagged abnormalities like fever. Along the way, we added symptom logging and quick insights.
Each round of iteration pulled us further away from the janky CSV days and closer to something that felt approachable, scalable, and trustworthy. What started as a raw pipeline of numbers slowly grew into a product that could genuinely support everyday decisions about fertility.
Iteration and Initial User Testing:
Rethinking the visual identity and adapting to wearable tech/ML advancements.
Major visual overhaul →
After the initial high-fidelity prototype was built, I became the sole UX designer on the project. To validate the work, I tested the designs with users, who consistently described the app as too sterile and, therefore, unlikely to be used regularly.
This feedback pushed me to rethink the 20+ component design system and redesign over 15 screens, evolving the aesthetic into something more welcoming while maintaining trust and credibility.
Advancements in wearable technology and ML algorithms →
As the wearable’s battery life improved and we learned more about how users should wear the tracker during sleep for accurate predictions, I rethought how the device was represented in the interface. This meant removing a dedicated battery indicator from the home screen and redesigning the flow to support instantaneous data transfer.
Reflection and Next Steps:
Machine learning may power products, but design is what makes it usable, credible, and human.
What I learned→
This project reinforced that designing for AI in healthcare tracking is about trust as much as accuracy. I learned to translate messy biomedical predictions into approachable guidance. I learned that designing for new hardware is never “done,” it’s about building for the long term, as hardware constraints are discovered or solved
Next Steps →
Next, we plan to conduct user testing with updated machine learning predictions integrated into the interface, allowing us to more holistically evaluate the app’s predictive capabilities. As the wearable evolves beyond temperature tracking, we will also introduce additional health data to broaden insights. Finally, the product will undergo FDA testing as we work toward official approval.
Thanks for visiting! You can look through more of my case studies below or reach out to me here
My Other Projects:
