Background: Morphology-based embryo selection cannot detect aneuploidy, which is common in advanced maternal age and recurrent pregnancy loss. NICS-AI combines non-invasive chromosome screening (NICS) of cell-free DNA from spent blastocyst culture medium with AI integration of developmental day and morphology to improve embryo ranking.
Methods: This multicenter, single-blind, parallel randomized controlled trial will include 520 participants. Participants undergoing conventional IVF will be eligible if they meet either (i) female age 35-43 years or (ii) recurrent miscarriage (≥2 losses \<28 gestational weeks, including biochemical pregnancy with serum hCG \>25 IU/L). They must consent to blastocyst culture/vitrification and frozen-thawed single blastocyst transfer (SBT), and have ≥2 Day-5/Day-6, 2PN-derived blastocysts with morphology grade ≥4BC/4CB at randomization. Key exclusions include any ICSI-based fertilization or PGT-related procedures, known genetic disease meeting PGT indications, donor oocytes, untreated uterine anomalies/hydrosalpinx, or contraindications to pregnancy/ART.
Randomization/interventions: Participants will be randomized 1:1 to NICS-AI-guided selection or morphology-based selection. In the NICS-AI arm, culture-medium DNA is tested and an AI-derived composite implantation score ranks embryos; controls use morphology alone (tie-break by cryopreservation order).
Outcomes/analysis: The primary endpoint is live birth after the first SBT (delivery with ≥1 live-born infant per transfer cycle, per randomized participant). Secondary endpoints include first clinical pregnancy, early miscarriage (\<12 weeks, excluding biochemical pregnancy), ongoing pregnancy to 12 weeks, and cumulative pregnancy/live birth outcomes within 1 year (≤3 SBTs from one retrieval). Safety includes fetal malformations and neonatal outcomes through 1 year postpartum.