Abstract Study question Does the Duostim protocol, evaluated by embryologists and AI, yield similar embryo quality in both cycles, and what impact does it have on reproductive outcomes? Summary answer Duostim protocol, when assessed by both AI and embryologists, maintains consistent embryo quality, making it a promising therapeutic option in assisted reproductive technology. What is known already Duostim optimises ovarian cycles with consecutive stimulations in the follicular and luteal phases, particularly beneficial for poor prognosis and oncological patients, aiming to boost oocyte production in a short period. It’s crucial to evaluate its impact on embryo quality. AI tools like CHLOE-EQ assist embryologists by automatically generating an embryo quality score (EQ Score) from time-lapse videos and morphokinetic annotations. This study compares the embryo quality from the initial and subsequent stimulations in a Duostim cycle, examining both AI-generated EQ Scores and assessments by embryologists. Study design, size, duration In a retrospective study spanning from February to December 2022, 234 embryos resulting from Duostim protocols were analyzed. The embryos were divided into “first” (n = 112) and “second” (n = 122) stimulations. Evaluation of embryo quality utilized CHLOE-EQ scores (Fairtility Ltd), morphokinetic events (tPNa-tEB), and Direct Unequal Cleavage (DUC) automatically annotated by AI. Manual annotations by embryologists included parameters like euploidy, utilization, and fragmentation. Participants/materials, setting, methods The study included embryos from Duostim protocols. Embryo quality of each stimulation group was evaluated using EQ score and embryo development pace (tPNa-tEB), measured by 2-sample t-test. Additional analysis compared euploidy, utilization, and abnormalities (DUCs, fragmentation) between first and second stimulation and was evaluated using chi-squared tests. Main results and the role of chance Patient age averaged 40±3 years. EQ scores revealed comparable embryo quality between “first” and “second” stimulation cycles (3.2 ± 3.6 vs. 2.9 ± 3.7), with a similar distribution of low, medium, and high-quality embryos.Similarities were observed in development rates between embryos derived from the first and second stimulation; tPNA(9 ± 3 vs 9 ± 3); tPNf(22 ± 7 vs 22 ± 5); t2(27 ± 4 vs 27 ± 4); t3(37 ± 6 vs 37 ± 5); t4 (39 ± 6 vs 39 ± 6); t5(48 ± 9 vs 49 ± 8); t6(54 ± 9 vs 56 ± 10); t7(57 ± 10 vs 59 ± 10); t8(61 ± 12 vs 61 ± 12); t9(70 ± 12 vs 73 ± 12); tM(89 ± 12 vs 88 ± 14); tSB(106 ± 12 vs 107 ± 16); tB(113 ± 12 vs 116 ± 15); tEB(124 ± 11 vs 125 ± 14), p=NS. Both stimulation groups had similar utilization (39% vs. 33%), euploidy (54% vs. 45%), DUCs (19% vs. 17%), and fragmentation rates (≤25%: 95% vs. 96%; >26%: 5% vs. 4 The quantity of oocytes retrieved from the first and second stimulations (112 vs. 122) further supported these findings. Limitations, reasons for caution The sample, while substantial, may not represent diverse populations. Generalizability and potential confounding variables should be considered. To enhance applicability, assessments across diverse IVF centers with varied patient demographics are crucial. Wider implications of the findings Duostim, assessed by both AI and embryologists, emerges as a promising therapeutic option without compromising embryonic quality or euploidy. The study supports AI integration for assessing therapeutic protocols and their impact on embryonic quality, emphasizing its potential in enhancing reproductive medicine practices. Trial registration number NA
Read full abstract