use hormone-free contraception
through learned, scientifically studied NFP methods.
HTW BERLIN · CERVICAL MUCUS RESEARCH
Understanding cervical mucus.
With machine learning.
We study how machine learning can classify cervical mucus. Our work brings established cycle knowledge together with observations from our community.
A research project at Hochschule für Technik und Wirtschaft Berlin. Participants wanted.
01 / WHY IT MATTERS
A taboo.
A research gap.
Many areas of women’s health remain under-researched and underfunded. Vaginal discharge is often discussed with shame or disgust. Yet it is a normal part of the body and can offer clues about the cycle.
We want to help people understand discharge and make this knowledge more accessible. Cervical mucus is one component of discharge — and the focus of our research.
02 / OUR RESEARCH
Cycle knowledge meets
machine learning.
Observing cervical mucus has been part of natural family planning for decades. We build on this knowledge and current counselling practice: how can models help with classification, and where are their limits?
Better cycle knowledge can help people …
plan a pregnancy
by better understanding fertile phases.
make informed decisions
with a better understanding of their body.
AURA researches one part of this. The app does not currently provide a validated contraception or fertility assessment.
03 / WITH THE COMMUNITY
Without you,
there is no dataset.
Voluntary contributions from many people form an independent research dataset. It makes developing and thoroughly testing models possible. This project cannot exist without its community.
We are looking for more participants. With AURA, you can keep track of your cycle and contribute to the research.
See how to take part ↗How your contributions become research
01 / 05
Keep a record for yourself
Upload a photo in AURA and note what you observe. Your entries form your personal record. You decide whether to contribute them to the research.
Many people build one dataset
With consent, photos and observations from many participants come together in an independent research dataset. This variety is what makes the research possible.
Prepare and label the images
We check image quality, add descriptions such as the observed cervical mucus type, and prepare the images for analysis. This gives the model examples it can learn from.
Train models. Test their limits.
After training, we thoroughly test the model and examine its limitations.
A model for the community
Our goal is a model that recognizes cervical mucus across different people. We aim to make it available to the community whose contributions helped develop it. This is what we are working toward.