The “best-looking” embryo is not always the one transferred first. At Clínica Tambre, Dr Abraham Zavala, Fertility Specialist, explains how AI can sharpen IVF decisions—from embryo ranking to donor matching—without replacing the embryologist or your doctor.
For patients trying to make sense of IVF, the biggest shift may be this: artificial intelligence is increasingly being used to help choose embryos, match donors, assess sperm, and study the uterine lining — but the final decisions still rest with people.
Fertility care is full of judgment calls. Which embryo should be transferred first when several look good? Is a donor the best physical and genetic match? Could a sperm sample that appears normal still have hidden problems? These are exactly the kinds of questions AI is being asked to help with.
With more than 45 years in assisted reproduction, Clínica Tambre has integrated AI into its workflow as a support tool while keeping medical expertise at the centre.
Still, AI in fertility is best understood as an extra layer of analysis. It can spot patterns across large amounts of data and images, but it does not know your full medical history, your diagnosis, or the tradeoffs in your treatment plan unless a clinician puts those pieces together. In practice, that means clinicians use AI as one more input alongside the rest of the medical picture.
For many patients, the most immediate question in IVF is simple: which embryo gives me the best chance? Embryo selection is where AI has attracted the most attention.
Clínica Tambre aims for a day 5 transfer whenever possible, meaning the embryo is transferred at the blastocyst stage. This is the point at which the embryo would naturally undergo implantation.
Embryo ranking takes on added importance when the plan is to transfer just one embryo. Traditional embryo grading already looks closely at how a blastocyst appears, including the quality of the inner cell mass, which later becomes the fetus and baby, and the trophoblast, the outer ring of cells that becomes the placenta. Embryologists also consider how expanded the blastocyst is and whether it reached that stage on day 5 or day 6.
Those details can lead to decisions that are less obvious than patients might expect. A day 5 blastocyst is generally considered more advanced than a day 6 blastocyst, and the best embryo for transfer may still be a slightly lower-quality day 5 embryo rather than a day 6 AA embryo. Put simply, the embryo with the best-looking grade on paper is not always the one chosen first.
AI provides one more variable in that decision. The clinic uses CHLOE, an AI system trained on a huge amount of data from many embryo culture cycles. It analyses development patterns that may not be apparent to the naked eye or even to an embryologist and gives the embryo a score that can be compared with the clinic’s own grading system. The goal is not to replace standard grading, but to add another variable when deciding which embryo to transfer next.
A major reason AI can be useful in embryo selection is that modern incubators can collect far more information than a single daily check under the microscope. With time-lapse technology, embryos remain in stable culture conditions while a series of pictures is taken frequently during development.
That allows embryologists to review development frame by frame instead of relying on isolated snapshots. They can assess division patterns, symmetry, fragmentation, and the timing of development to the blastocyst stage. When several embryos appear similarly strong, those subtle differences can help identify which one may be more appropriate for transfer.
AI works particularly well alongside time-lapse imaging. The imaging creates a detailed visual record, and AI can analyse that record for patterns linked to implantation potential. For patients, the practical takeaway is that embryo selection is becoming less dependent on a single moment of observation and more informed by the embryo’s full developmental story.
One of the easiest misunderstandings about AI in IVF is the belief that it can tell which embryo will become a healthy baby. That is not what these tools can do today.
Systems such as CHLOE are designed to estimate which embryos have the highest chance of implantation or achieving pregnancy based on development. That is useful, but it is only one part of the process. Miscarriages can happen for many reasons, including factors outside the embryo itself. Even a euploid embryo or a great-quality embryo can miscarry without the embryo being the problem.
That limitation is important for patients deciding how much confidence to place in an AI score. Right now, no tool can determine which embryo will lead to a healthy live birth. AI may help identify embryos that are more likely to attach, but it does not answer every question that matters after transfer.
Patients often wonder whether smarter image analysis could eventually reduce the need for procedures such as PGT-A, or preimplantation genetic testing. The idea is appealing: if AI can analyse many factors at once, perhaps it could someday make highly accurate predictions without invasive testing.
That possibility is still ahead, not here. The current view at the clinic is that AI may eventually play one of the roles in PGT-A assessment and could hopefully lead to very precise predictions in a noninvasive manner. Noninvasive PGT-A is seen as a step in the right direction, but it is not yet at the point where it can replace existing approaches.
AI is expanding what can be inferred from embryo development, but it has not eliminated the need for established testing strategies when those are clinically indicated.
AI is not only being used after eggs are retrieved and embryos are created. It is also being used earlier in the process, especially in donor programs.
Under Spanish regulation, donor matching takes into account weight, height, hair colour, hair texture, ethnicity, skin type, eye colour, and blood type. Fenomatch goes further by comparing the donor’s picture with the patient’s picture to improve physical matching.
Technology is also part of genetic matching, especially carrier screening and the detection of shared mutations between donor and patient. For patients using donor eggs or donor sperm, that means matching is not just about appearance. It also includes efforts to reduce genetic risk where shared mutations might otherwise go unnoticed.
It is one of the clearest practical uses of AI in fertility care, tied to a concern many patients bring into donor treatment. Patients often want a donor match that feels medically responsible and personally reassuring. Better matching tools cannot remove every uncertainty, but they can make the process more precise.
Embryo quality is only one part of whether the treatment works. The sperm matters. The uterine lining matters. The broader medical context matters. That is why AI is being folded into other parts of fertility care as well.
On the male side, the clinic uses methods to better select sperm samples, particularly in cases where sperm quality appears normal but fragmentation analysis suggests additional issues. Techniques mentioned include the ZymoT chip (Fertile Chip) and the sperm slow media technique. These are part of a broader effort to avoid relying on a standard semen result alone when the clinical picture suggests more may be going on.
Through the MATRIS project, images of prepared uterine linings for embryo transfer are being collected so AI can help determine which linings are best suited for implantation. This work is still in an initial phase, but it reflects a larger shift in fertility medicine: success is increasingly being approached as a whole-system problem, not just an embryo problem.
That broader approach matters especially for patients with recurrent implantation failure, recurrent miscarriages, diminished ovarian reserve, endometriosis, adenomyosis, infertility of unknown origin, or women over the age of 38. In these cases, choosing the embryo is important, but it is rarely the only issue worth investigating.
This may be the most reassuring point for patients uneasy about machines entering fertility care: the embryologist makes the final call.
AI can generate a score or highlight a pattern, but it does not carry responsibility for the outcome. Regulation is one reason. If something goes wrong, there must be a responsible person, and that cannot be the machine. The larger reason is clinical judgment. Fertility decisions are made in context, and context is exactly where human expertise still matters most.
The same logic extends to patients who come in with information gathered online or through chatbots. Those tools may help explain fertility topics in a simpler or more detailed way, and many patients already use them before a consultation. But fertility advice only becomes useful when it is filtered through a professional who understands the patient’s profile, problem, and goal. An answer that fits one person may be completely wrong for another.
In that sense, AI inside the clinic and AI outside the clinic pose the same challenge. Both can be informative. Neither should be treated as a substitute for individualised care.
Not immediately, and that may disappoint patients who are already stretched by treatment costs.
Over time, more automated processes may reduce some costs, and there is clear hope that AI will improve affordability because access to fertility treatment is limited worldwide by cost. But fertility care still involves human cells, medical procedures, and specialist oversight. Those realities do not disappear just because software becomes better.
For now, the case for AI is less about making IVF inexpensive than about making decisions more efficient and more informed, which could help avoid wasted time and repeated disappointment.
AI is changing IVF in the places where uncertainty has always been hardest: choosing between similar embryos, refining donor matches, spotting hidden sperm issues, and learning more about the uterine lining before transfer. At its best, it offers a more detailed view of the biology without pretending to control every outcome.
That balance is worth keeping in mind if you are considering treatment. AI may improve how fertility clinics sort information and prioritise options, but it does not replace medical judgment, and it does not erase the unpredictability of pregnancy. The hopeful part is that these tools are being used to make care more precise and, whenever possible, to make the road to parenthood shorter.
