Dutch Fertility Clinics Turn to AI for Embryo Selection Despite Lack of Proven Success
Amsterdam, Wednesday, 29 April 2026.
Four out of sixteen Dutch IVF laboratories now use artificial intelligence to select embryos, despite no scientific evidence that AI improves pregnancy rates. The technology photographs embryos every ten minutes and assigns scores from zero to ten, but research shows it may actually be more expensive without delivering better outcomes. With Netherlands’ IVF success rates stagnant at thirty percent for twenty years, experts warn against costly unproven technologies that could jeopardize insurance coverage for treatments.
Current Adoption and Technology Implementation
The adoption of AI-driven embryo selection represents a significant shift in Dutch reproductive medicine. As of April 2026, four of the sixteen Dutch IVF laboratories actively employ artificial intelligence for embryo selection, while an additional five clinics possess timelapse incubators that can support AI implementation [1]. The technology operates by capturing photographs of developing embryos every ten minutes throughout their development process [1]. At Dijklander Ziekenhuis in Purmerend, lab head Margarida Santos has been utilizing AI for embryo selection since approximately April 2025, working with incubators containing sixteen compartments for petri dishes [1]. The AI system assigns embryos scores ranging from zero to ten based on their viability potential [1].
The Human Cost Behind the Technology
The personal stakes driving demand for AI-assisted embryo selection become clear through patient experiences like that of 39-year-old Samantha de Heij, who endured twenty artificial attempts over eight years before achieving a successful pregnancy [1]. Her journey, which she described as requiring “blood, sweat and tears,” involved repeated hormone stimulations, punctures, embryo transfers, and numerous hospital visits [1]. De Heij expressed her desperation, stating she would have done anything for success, “Bij wijze van spreken een moord” (figuratively speaking, even murder) [1]. This emotional investment explains why patients eagerly embrace any technology promising improved outcomes, even when scientific evidence remains inconclusive.
Limited Evidence for Effectiveness Despite High Costs
Scientific research has failed to demonstrate that artificial intelligence actually improves IVF success rates. Doctoral research conducted by Dorit Kieslinger, a clinical embryologist at Amsterdam UMC, specifically examined timelapse technology combined with algorithms and found that while the technology increased costs, it did not enhance pregnancy rates [1]. This finding is particularly significant given that Netherlands’ average IVF success rate has remained stagnant at approximately thirty percent for the past twenty years, according to the Nederlandse Vereniging voor Obstetrie en Gynaecologie [1]. The AI systems represent substantial investments, with Santos comparing the cost to “a Tesla, and not the cheapest model” [1]. Despite this expense, the underlying patterns and scoring mechanisms remain opaque, as Santos acknowledges: “Welk patroon het systeem herkent en waarop de score precies is gebaseerd, blijft ondoorzichtig. Zo werkt AI nu eenmaal” (Which pattern the system recognizes and what the score is precisely based on remains unclear. That’s how AI works) [1].
Ethical Concerns and Future Implications
The implementation of unproven AI technology raises significant ethical and practical concerns among medical professionals. Kieslinger warns against the human tendency to embrace promising technologies, stating “Mensen willen te graag geloven dat het werkt” (People want too much to believe that it works) [1]. More critically, she cautions that introducing expensive, ineffective technologies could jeopardize insurance coverage for IVF treatments, referencing discussions that occurred in 2012 about limiting reimbursed treatments [1]. Stefan Buijsman, an assistant professor of Responsible AI at TU Delft, highlights additional risks, noting that AI models could systematically favor certain types of embryos or perform better for patient groups on which they were trained [1]. The technology also raises broader questions about reproductive autonomy, as ethicist Tsjalling Swierstra from Maastricht University asks: “Hoeveel controle is goed voor ons? Moeten we een designerbaby willen?” (How much control is good for us? Should we want a designer baby?) [1]. While Dutch clinics currently use AI only to support embryologists’ decision-making, international developments show more controversial applications, with some U.S. companies using AI alongside genetic testing to select embryos based on disease risk, eye color, or intelligence [1].