EMBRY started with one question: can AI improve the success rate of IVF procedures? It began as a pilot to assess the quality of mature eggs, trained with embryologist Mimoza Adji-Krsteva on a diverse dataset of oocytes and embryos. It now decodes embryo development stages, covers sperm analytics, and predicts IVF outcomes.
What it does
- Follows the whole cycle. Embryologists record patient details and every step of the process; EMBRY keeps it structured and retrievable.
- Assesses quality with computer vision. Oocyte health, embryo quality, and day-specific development milestones such as Day 3 and Day 5.
- Predicts the outcome. Patient data and history are turned into an outcome probability, expressed as a percentage.
- Opens a research door. Hospitals, clinics and research centres can analyse their own data to find patterns worth investigating.
How it is built
EMBRY is built on Python and Django, designed to drop into existing hospital and clinic infrastructure. It works independently of your existing lab equipment, so it does not require a hardware refresh to be useful.
Data privacy
EMBRY stores results and figures, not personal identifiers. Each patient is tagged with an ID specific to the hospital or clinic, which keeps data private while still allowing an IVF attempt to be reassessed later.
Where it is going
Embry v2 extends the data schema across the full cycle, adds detection models for embryo, oocyte and sperm assessment, and moves clinics to fully paperless workflows.
Want to see EMBRY on your own data?