First Ever RCT of its Kind in the US. 7 IVF centers. 444 patients. Non-Inferiority, AI in IVF.
Skepticism toward artificial intelligence in embryo selection has largely stemmed from a lack of prospective, controlled evidence. That gap is beginning to narrow as a randomized controlled trial across seven U.S. IVF centers demonstrates that AI-supported selection can match outcomes achieved through traditional morphological grading. The study, which included an AI tool developed by Alife Health, introduces a level of clinical validation that has been largely absent from prior evaluations.
The trial enrolled 444 patients between June 2022 and October 2024, with 442 randomized and 359 completing embryo transfer in the intention-to-treat population. Participants, ages 21–43, were assigned to embryo selection based either on conventional morphology or an AI-generated score derived from day 5–7 blastocyst images. Clinical pregnancy, defined as fetal cardiac activity at 6–8 weeks, served as the primary endpoint.
See the RCT That Put AI to the Test
Most tools claim they’ve been tested. Alife Health did it.
In the first prospective randomized controlled trial of its kind, AI-supported embryo selection was shown to be non-inferior to traditional morphological grading—without risking worse outcomes.
What the data shows:
First RCT meeting endpoint for AI-supported embryo selection
Supports safe, real-world clinical integration
Demonstrates the potential for AI to standardize selection without sacrificing performance
👉 See the evidence, review the outcomes, and decide if AI belongs in your lab.
72.9% Clinical Pregnancy Rate Meets Non-Inferiority Threshold
In the modified intention-to-treat analysis, clinical pregnancy occurred in 68.0% of patients in the morphology arm compared with 72.9% in the AI-supported arm . The study met its pre-specified 10% non-inferiority margin across all analysis groups, with no safety concerns reported . While not powered to demonstrate superiority, the AI arm consistently exceeded 70% clinical pregnancy rates across subgroup analyses.
The design of the study reflects a higher evidentiary standard than most prior work in this space. Randomized controlled trials remain the benchmark for establishing cause-and-effect relationships in clinical practice, particularly in IVF where multiple variables can influence outcomes. Baseline characteristics, including age, BMI, hormonal profiles, and laboratory variables, were comparable between groups, supporting the integrity of the comparison .
Embryologists maintained oversight in the AI arm, with the ability to override the algorithm’s recommendation when clinically justified . In the randomized adherence to protocol cohort, this occurred in 18% of cases, indicating that clinical judgment remained an active component of decision-making alongside algorithmic input.
Randomized Data Reframes Adoption Decisions in IVF Labs
The introduction of randomized evidence alters how AI tools are evaluated within IVF laboratories. Retrospective analyses and model validation studies have offered performance signals, but have not resolved concerns around real-world clinical impact. By demonstrating non-inferiority in a controlled, multicenter setting, the trial provides data that directly looks to address whether AI-assisted selection is safe and changes outcomes in practice.
This shift carries implications for how clinics approach standardization and variability in embryo selection. Morphological grading has historically relied on individual embryologist assessment, introducing potential variability across operators and centers. The study suggests that algorithm-supported selection can produce consistent outcomes without introducing measurable risk, even when clinicians retain the ability to intervene.
For clinic directors and laboratory leaders managing increasing cycle volumes and staffing constraints, the presence of randomized data introduces a different level of consideration. Rather than evaluating AI tools solely on predictive accuracy, ALife moves toward how these systems perform within clinical workflows under controlled conditions.
AD HEADLINE HERE
AD SUB-HEADLINE HERE
Lorem Ipsum is simply dummy text of the printing and typesetting industry.
Lorem Ipsum has been the industry's standard dummy text
Lorem Ipsum
Lorem Ipsum
Lorem Ipsum
Lorem Ipsum is simply dummy text of the printing and typesetting industry
<call to action> - short explanation of what will happen when they click the button