AI embryo selection tools analyze time-lapse images and developmental data to score embryos, helping embryologists make more consistent and objective assessments. Some observational studies show improved implantation rates, but the largest randomized trial to date found AI did not outperform experienced embryologists in clinical pregnancy rates. AI is a promising support tool — not a silver bullet. It enhances consistency in the lab but doesn’t replace human expertise, and it’s not yet proven to improve live birth rates in rigorous trials.
📝 Key Takeaways
- AI scores embryos, it doesn’t pick them. These tools provide a numerical score based on developmental patterns. Your embryologist still makes the final recommendation.
- The biggest trial was humbling. A 2024 Nature Medicine randomized trial across 14 clinics found iDAScore achieved 46.5% clinical pregnancy vs. 48.2% for standard embryologist assessment. AI didn’t win.
- Consistency is the real value. AI reduces variability between embryologists and clinics. The same embryo gets the same score every time, regardless of who’s on shift.
- It’s usually bundled with time-lapse. AI scoring is part of time-lapse incubation systems many clinics already use. It may add $500–$1,500 to your cycle cost, or be included in standard pricing.
- Don’t choose a clinic solely for AI. Lab quality, physician experience, and success rates matter far more than any single technology.
How AI Embryo Selection Actually Works
Traditional embryo selection relies on an embryologist looking at your embryos under a microscope (or reviewing time-lapse images) and grading them based on how they look — cell number, symmetry, fragmentation, and blastocyst quality. It’s skilled work, but it’s inherently subjective. Two embryologists can grade the same embryo differently, and the same embryologist might grade it differently on Monday versus Friday.
AI embryo selection tools work differently. They analyze thousands of images from time-lapse cameras that photograph your embryos every few minutes as they develop over five to six days. The AI has been trained on datasets of tens of thousands (or even hundreds of thousands) of embryos with known outcomes — did they implant? Did they lead to a pregnancy? Using deep learning, the AI identifies patterns in development that correlate with success and assigns each embryo a numerical score.
The key thing to understand: the AI score is an additional data point for your embryologist, not a replacement for their judgment. Your embryologist considers the AI score alongside their own assessment, your medical history, your age, and other clinical factors when recommending which embryo to transfer.
The Major AI Platforms in Clinics Today
iDAScore
Developed by Vitrolife. Analyzes full time-lapse image sequences across embryo development. Trained on over 115,000 embryos from 18 IVF centers. The subject of the largest randomized controlled trial to date (Nature Medicine, 2024). Score reflects probability of successful implantation.
ERICA
Embryo Ranking Intelligent Classification Algorithm. Predicts both implantation potential and chromosomal status (ploidy). Achieved a positive predictive value of 0.79 for euploidy. Developed at Weill Cornell Medicine. Aims to non-invasively predict what PGT-A testing would reveal.
STORK / STORK-A
Developed at Weill Cornell Medicine using Google’s Inception deep learning model. Trained on ~50,000 images. STORK predicts blastocyst quality with >0.98 AUC. STORK-A extends this to predict ploidy status with ~69% accuracy — promising but not yet clinically reliable for standalone use.
AIVF / EmbryoScore
Commercial platform used in clinics globally. Reports 65% implantation rate for high-scoring embryos at partner clinics. Claims to improve frozen transfer pregnancy rates by up to 30%. Named to CB Insights’ 50 Most Promising Digital Health Companies.
What the Research Actually Shows
This is where it gets nuanced — and where honest reporting matters.
The Good News
AI tools consistently demonstrate high accuracy at grading embryo quality when compared to expert embryologists. They reduce subjectivity and inter-observer variability, meaning your embryo gets a more consistent assessment regardless of which embryologist is working that day. Some observational studies and clinic-reported data show improved implantation rates and clinical pregnancy rates when AI scoring is used alongside standard assessment.
The Humbling News
The most rigorous test of AI embryo selection to date was a multicenter randomized controlled trial published in Nature Medicine in 2024. Over 1,000 patients across 14 clinics in Australia and Europe were randomized to either standard embryologist selection or iDAScore AI selection. The result: AI achieved a 46.5% clinical pregnancy rate compared to 48.2% for trained embryologists. AI did not demonstrate noninferiority — meaning the study could not confirm that AI was even as good as, let alone better than, experienced humans.
This doesn’t mean AI is useless. It means the technology isn’t yet proven to improve outcomes in the gold-standard type of study that matters most. The tools may still add value in specific scenarios — like clinics with less experienced embryologists, high-volume labs where consistency matters, or as a non-invasive supplement to genetic testing.
The Emerging Frontier
Perhaps the most exciting application is non-invasive ploidy prediction. If AI could reliably predict whether an embryo is chromosomally normal from images alone — without the need for a biopsy and PGT-A testing — it would save patients thousands of dollars and avoid the small risk associated with biopsying embryos. STORK-A and ERICA both show promise here, but accuracy is currently around 69–79%, which isn’t high enough for standalone clinical use. This is an area to watch closely.
⚠️ The Patient Trust Gap
A 2025 study surveying 200 IVF patients found that while 93% were familiar with AI and 55% supported its use in medicine, only 46% trusted AI-informed reproductive care. If your clinic uses AI and you have concerns, you have every right to ask how the AI score will factor into embryo selection decisions — and whether you can opt out.
Questions to Ask Your Clinic
If your clinic uses AI embryo selection (or you’re considering one that does), here are the questions that will help you understand what you’re getting.
💬 Questions Worth Asking
- “Which AI system do you use, and how is it integrated into your embryo selection process?”
- “Does the AI score override the embryologist’s assessment, or is it one factor among many?”
- “Has your clinic seen improved outcomes since adopting AI scoring? Can you share data?”
- “Is AI scoring included in the standard IVF cost, or is it an additional charge?”
- “Do you use time-lapse incubation? Is AI scoring built into that system?”
- “Would you recommend PGT-A genetic testing alongside AI scoring, or can one substitute for the other?”
- “Can I see how my embryos were scored and understand what the numbers mean?”
The Bottom Line for Patients
AI embryo selection is a legitimate and growing part of modern IVF, but it’s important to keep perspective. The technology is best understood as a tool that improves consistency and provides an additional data point — not a guarantee of better outcomes. The things that matter most for your IVF success remain the same: the quality of your medical team, the lab environment, your individual biology, and a well-designed treatment protocol.
If your clinic offers AI scoring, it’s a positive signal that they’re invested in technology and evidence-based practice. But don’t choose a clinic solely because of AI. And don’t let anyone tell you AI guarantees success — no technology can make that promise.
What AI can do is give your embryologist a more objective, data-driven perspective on your embryos. And in a field where a single embryo decision can mean the difference between a pregnancy and another cycle, every additional piece of good information helps.
📚 Prepare for Your Best Cycle
While technology in the lab evolves, the basics of egg quality still matter. These evidence-based resources help you prepare.
It Starts with the Egg
The definitive guide to improving egg quality through supplements and lifestyle changes. Updated with current research on CoQ10, antioxidants, and environmental factors.
View on Amazon →CoQ10 (Ubiquinol, 200mg)
Research supports CoQ10 for mitochondrial function and egg quality. The ubiquinol form is better absorbed. Most REs recommend 400–600mg daily for 2–3 months before a cycle.
View on Amazon →Prenatal with Methylfolate
Start a quality prenatal at least one month before IVF. Look for methylfolate over folic acid for better absorption, especially if you have MTHFR variants.
View on Amazon →Related Guides on ConceiveGuide
Technology evolves. Your questions matter.
Understanding what your clinic offers — and what it costs — is the first step to an informed treatment journey.
Understand IVF Costs →Frequently Asked Questions
The evidence is mixed. Some studies show improved implantation rates, but the largest randomized trial (Nature Medicine, 2024) found AI did not outperform trained embryologists in clinical pregnancy rates (46.5% vs. 48.2%). AI adds consistency and objectivity, but it hasn’t been proven to improve live birth rates in rigorous trials.
AI scoring is typically bundled with time-lapse incubation, adding roughly $500–$1,500 to a cycle. Some clinics include it in standard pricing. It’s rarely billed as a separate line item. Ask your clinic specifically whether their time-lapse system includes AI scoring and how it affects your total cost.
No. AI provides a score; the embryologist makes the decision. Your embryologist reviews AI scores alongside their own expertise, your medical history, and clinical context. Think of AI as a second opinion from a very consistent, data-driven colleague — not a replacement for the human expert in the room.
Not yet. Some AI tools (STORK-A, ERICA) show promise in predicting chromosomal status from images, but current accuracy (~69–79%) isn’t high enough to replace PGT-A, which remains the gold standard for embryo chromosomal screening. AI may eventually reduce the need for biopsies, but we’re not there yet.
AI is one positive signal among many. Lab quality, physician experience, success rates, patient support, and communication matter more than any single technology. A well-run clinic without AI will likely deliver better outcomes than a mediocre clinic with it. That said, AI adoption often reflects a clinic’s commitment to innovation and evidence-based practice.
Sources & References
- Berntsen J et al. — Deep learning versus manual morphology-based embryo selection in IVF: a randomized, double-blind noninferiority trial. Nature Medicine, August 2024.
- Bormann CL et al. — Deep learning enables robust assessment and selection of human blastocysts (STORK). npj Digital Medicine, 2019.
- Bormann CL et al. — STORK-A: Embryo Ranking Intelligent Classification Algorithm (ERICA). Reproductive BioMedicine Online, 2020.
- AIVF — aivf.co
- MIT Technology Review — 10 Breakthrough Technologies 2026: Embryo Scoring (May 2026)
- Salih M et al. — Artificial Intelligence in Reproductive Medicine: Transforming ART. Journal of IVF-Worldwide, October 2025.
- Berntsen J et al. — Robust and generalizable embryo selection based on AI and time-lapse image sequences. PLoS ONE, 2022.
- Cromack et al. — Patient Trust in AI-Informed Reproductive Care (2025 survey, n=200).