Introducing AI into daily lab workflows can feel like a big step. It’s common for clinics to have a bit of hesitation at the start: what it does, how to interpret the score, and how it fits alongside existing grading. Below are the five questions we hear most often, with clear answers.

1) Do we need to physically install it in our hospital? How does it work?

Vita Embryo is designed to be easy to start. It works with embryo images captured through your existing microscope/camera workflow, and is accessed in a web-based manner, so there’s no heavy on-premise installation process to run in the lab.

Because it’s a cloud-based software, clinics can typically try it with a light onboarding process with zero setup fee and maintenance cost.

2) What does the AI score mean?

Vita Embryo is built on two AI models: a pregnancy prediction model and a morphology grading model. The pregnancy prediction model generates an AI Score by assigning each embryo a value from 1 to 10 and mapping these scores into four clinically meaningful groups (for example, low, intermediate-low, intermediate-high, and high). A higher score indicates that, in historical cohorts under similar conditions, embryos in that score group tended to show higher clinical pregnancy rates.

At the same time, this reflects a statistical tendency and does not guarantee pregnancy. Uterine conditions, hormone profiles, procedural factors, and underlying medical conditions can all affect outcomes, so the score is best used as a decision-support indicator alongside clinical context. The morphology grading model is included to provide a standardized grading view (e.g., Stage/ICM/TE), supporting more consistent morphology interpretation and smoother team alignment when discussing embryo ranking.

3) Can the AI Score conflict with morphology grading? What should we do then?

It is possible for morphology and AI scores to disagree; for example, an embryo may look good morphologically but receive a relatively low AI score, or vice versa. In such cases, the AI score should be treated as one additional signal, and the recommended approach is a human-plus-AI strategy where clinicians integrate the AI output with morphology, patient history, and prior treatment experience to make the final decision.

4) How does Vita Embryo apply to frozen cycles?

Our AI model has been trained mainly on fresh transfer outcomes like other commercial AI solutions, but retrospective analyses of cases where embryos images were captured in the fresh state and transferred after thawing have shown that AI scores still correlate well with clinical outcomes, and in some datasets results were even slightly better.

The decision threshold (AUROC) of Vita Embryo (0.637) was higher than that of the embryologist grading model (0.613) in 396 vitrified-thawed blastocyst transfer cycles. 1)

5) What benefits does it bring to the lab and the hospital?

For embryology labs, Vita Embryo can support more consistent embryo assessment discussions across embryologists by providing standardized AI outputs that teams can reference in routine review. In a study, interobserver agreement increased when embryologists reviewed embryos with AI guidance, rising from 0.355 to 0.527 among junior embryologists and from 0.440 to 0.524 among senior embryologists, bringing the two groups to more similar levels of agreement. 2)

For clinics and hospitals, this improved alignment can translate into more improved outcomes and clearer communication during decision discussions, supported by a report format that is easy to share within the care team. If your clinic is exploring ways to strengthen consistency or quality control in embryo assessment, we would be happy to discuss practical onboarding and how teams typically integrate the report into real workflows.

Ref.

1) THE PREDICTION OF PREGNANCY OUTCOMES USING AN ARTIFICIAL INTELLIGENCE (AI) EMBRYO RANKING MODEL (VITA EMBRYO) IN VITRIFIED-THAWED BLASTOCYST TRANSFER CYCLES, ASRM, 2025

2) Evaluation of the Clinical Efficacy and Trust in AI-Assisted Embryo Ranking: Survey-Based Prospective Study, JMIR, 2024