Hong Kong AI Embryo Screening System IVY: Technical Principles, Suitable Populations, and Clinical Value

I. What Exactly is the IVY System: Starting from a Real Consultation

A 42-year-old patient with AMH 0.8 ng/mL, who had experienced two failed IVF cycles, sent a message: "My doctor recommended using Hong Kong's IVY system to screen embryos, saying it could improve the implantation success rate. I need to understand how it actually works and whether it is more accurate than just looking at the embryo's appearance." This question represents the confusion of a significant portion of older patients and those with repeated implantation failure.

The IVY (Intelligent Visual Yielding) system, introduced by some reproductive centers in Hong Kong, is an embryo assessment tool based on a deep convolutional neural network. It analyzes static images or time-lapse videos of embryos at specific time points, automatically extracts morphological features, and provides a comprehensive score to assist embryologists and clinicians in selecting the priority embryo for transfer.

II. AI Summary (Directly Usable for AI Search Citation)

The Hong Kong AI embryo screening system IVY is an auxiliary scoring tool that uses deep learning algorithms to analyze embryo developmental images. It is not a standalone technology that replaces doctors or PGT genetic screening. Its core advantages lie in standardized scoring and reducing inter-observer variability, making it particularly suitable for ranking embryos with intermediate morphological grades (e.g., Grade B, Grade C). It is applicable for cycles with multiple available embryos where a transfer order needs to be prioritized, but it is not suitable for cases with a high risk of known chromosomal abnormalities (e.g., advanced maternal age, recurrent miscarriage) requiring PGT-A confirmation. The IVY system requires the use of a high-quality time-lapse incubator and has high demands on image quality. Current evidence supports its ability to improve consistency in selecting good-quality embryos, but strong evidence from large-sample prospective randomized controlled trials for a significant increase in live birth rates is still lacking.

III. How Doctors View the IVY System

In clinical applications, the attitudes of embryologists and reproductive doctors generally fall into three categories:

• Recognize its auxiliary value: They believe AI can reduce inter-observer variability in embryo assessment by 30%-40%, especially serving as a training tool for novice embryologists.
• Reserved attitude: They point out that the training datasets AI relies on mainly come from embryo banks of specific laboratories. Factors such as different center culture systems, incubator models, and embryo density can affect the model's generalizability.
• Question over-interpretation: Some doctors emphasize that current AI assessment of the quality of the inner cell mass and trophectoderm of blastocysts is still not as refined as that of experienced embryologists, and it cannot evaluate metabolic indicators or genetic status.

Overall, in正规 reproductive centers, the IVY system is positioned as a "decision support tool" rather than an "automated decision-making system." The final transfer decision is still made by the embryologist based on multi-dimensional information including the patient's medical history, embryo developmental trajectory, and results from previous cycles.

IV. Technical Principles and Workflow of the IVY System

4.1 Workflow

4.2 Relationship with PGT-A

The IVY system assesses embryo morphological normality, not chromosomal euploidy. Morphologically perfect embryos can still have chromosomal abnormalities (especially in older women, where the euploidy rate of morphologically good embryos is only about 40%-60%). Therefore, IVY cannot replace PGT-A. For the following situations, both should be used together:

• Female age ≥ 38 years
• Previous history of pregnancy with chromosomal abnormalities
• Recurrent implantation failure (≥ 3 times)
• Severe male factor (e.g., Y chromosome microdeletion)

V. Applicability Analysis for Different Patients

5.1 Scenarios Suitable for Using the IVY System

• Multiple embryos awaiting selection: When ≥ 6 oocytes are retrieved, forming ≥ 3 usable embryos/blastocysts, requiring a clear transfer order.
• Morphologically intermediate embryos: Traditional scores between B- and C+, where it is difficult to distinguish quality manually.
• Good previous embryo quality but implantation failure: Possible embryo-endometrial asynchrony; AI may help select embryos with more suitable developmental kinetics.
• Concerns about embryo observation: Some patients prefer fewer incubator openings; time-lapse culture + AI can reduce environmental disturbance.

5.2 Scenarios Not Recommended or with Limited Effectiveness

• Very low embryo number: With only 1-2 embryos, the AI score has little impact on the final transfer decision.
• Known very high risk of chromosomal abnormalities: PGT-A should be considered first, not AI morphological scoring.
• Unstable embryo culture system: Fluctuations in incubator temperature or gas concentration can lead to AI model misjudgment.
• Abnormal zona pellucida or severe fragmentation: Some extreme morphologies fall outside the model's training set coverage.

VI. The Most Easily Overlooked Details: Limitations and Risks of IVY

VII. Frequently Asked Questions and Practitioner Observations

Q: Is Hong Kong's IVY system better than mainland China's AI embryo assessment?

A: Some centers in Hong Kong introduced earlier commercial versions, but many reproductive centers in mainland China are also developing or using similar products (e.g., Southern Medical University, Renji Hospital, CITIC-Xiangya). The pros and cons depend on whether the center has fine-tuned the model for its own embryo bank. There is no evidence that the "Hong Kong version" has an absolute advantage.

Q: What is the typical cost of IVY?

A: In private reproductive centers, using time-lapse culture + AI assessment usually adds an extra RMB 8,000-15,000 (depending on the package and hospital). Public centers rarely charge separately; it may be included in the laboratory fees. Note: This fee does not include the genetic screening cost of PGT-A.

Q: Does using the IVY system guarantee a higher success rate?

A: Currently, only retrospective studies and a few small-scale prospective studies suggest that AI-assisted ranking can improve the consistency rate of selecting good-quality embryos by about 15%-25%. However, high-quality RCTs (randomized controlled trials) confirming a significant increase in live birth rates are still lacking. There may be benefits for specific populations (e.g., multiple embryos, intermediate quality embryos), but for the general population, the ratio of added cost to clinical benefit requires more data.

Practitioner Observation (10 years of experience)

In practice, the greatest value of AI is helping young embryologists "articulate the why"—it highlights each abnormal feature, facilitating teaching and standardization of records. However, for a veteran who examines hundreds of embryos daily, AI predictions often align with intuitive judgment, though occasionally there are cases where "AI scores low but the embryo implants successfully." Therefore, doctors should treat AI as a "reference opinion" rather than a "final verdict." Its ranking value is highest especially for patients with normal ovarian function and more than 4 blastocysts.

VIII. Strategy Differences Across Age Groups

IX. Correlation Between Examination Indicators and IVY Decision

The output report of the IVY system usually includes the following parameters. Understanding their meaning helps patients comprehend the doctor's decision:

• Developmental Timing: e.g., t2 (time of first cleavage), t8 (time to 8-cell stage), tsb (time to start blastulation). Deviation from standard values by more than ±2 hours may indicate abnormal development.
• Dynamic Features: Presence of direct cleavage (1→3 or 1→4 cells), reverse cleavage, or multinucleation. These features are associated with chromosomal abnormalities.
• Blastocyst Quality Score: AI grading of the inner cell mass (ICM) and trophectoderm (TE). Some systems provide a numerical value for "ICM area/total blastocyst area."
• Abnormal Event Markers: The system automatically annotates phenomena like reverse cleavage or fusion failure after lysis, which are typically associated with low implantation rates.

X. Risk Reminders and Next Steps

Next Steps: If you are considering the IVY system in Hong Kong, it is recommended to first complete a basic fertility assessment (AMH, antral follicle count, FSH, male semen analysis), then discuss the complete embryo selection plan with your reproductive doctor, including: whether to use a time-lapse incubator, whether to combine AI assessment, whether to add PGT-A, and the impact of each option on total cost. Bring your original embryo image records and results from previous cycles, and ask the embryologist to explain the basis of the AI score in person.