SUMMARY ANSWER: Age, ethnicity, obesity (BMI ≥ 30 kg/m2), and polycystic ovarian syndrome (PCOS) significantly impacted serum AMH levels, with the rate of decrease accelerating as age increased; a concentration of 4.0 ng/ml was the optimal cut-off for diagnosis of PCOS.
WHAT IS KNOWN ALREADY: There are significant differences in ovarian reserve among women from different races and ethnicities, and Asian women often have poorer reproductive outcomes during assisted reproductive treatment cycles.
STUDY DESIGN, SIZE, DURATION: A population-based multi-nation, multi-centre, multi-ethnicity prospective cohort study of 4613 women was conducted from January 2020 to May 2021. Infertile women of 20-43 years of age were enrolled. The exclusion criteria included: age <20 or >43, non-Asian ethnicity, and missing critical data.
PARTICIPANTS/MATERIALS, SETTING, METHODS: Participants were Asian women of Chinese, Japanese, Korean, Thai, Vietnamese, Malay, Indian, and Indonesian ethnicities from 12 IVF centres across Asia. These women were all naïve to ovarian stimulation cycles and attended IVF centres for fertility assessment. The AMH measurement was performed using an AMH automated assay on a clinically validated platform.
MAIN RESULTS AND THE ROLE OF CHANCE: A total of 4556 infertile Asian women were included in the final analyses. The mean ± SD for serum AMH concentrations (ng/ml) across specific age groups were: overall, 3.44 ± 2.93; age <30, 4.58 ± 3.16; 30-31, 4.23 ± 3.23; 32-33, 3.90 ± 3.06; 34-35, 3.21 ± 2.65; 36-37, 2.74 ± 2.44; 38-39, 2.30 ± 1.91; 40 and above, 1.67 ± 2.00. The rate of AMH decrease was ∼0.13 ng/ml/year in patients aged 25-33 and 0.31 ng/ml/year in women aged 33-43. The highest rates of PCOS were found in Indians (18.6%), Malays (18.9%), and Vietnamese (17.7%). Age (P
RESULTS: The oocyte AI model demonstrated area under the curve (AUC) up to 0.65 on two blind test datasets. High sensitivity for predicting competent oocytes (83-88%) was offset by lower specificity (26-36%). Exclusion of confounding biological variables (male factor infertility and maternal age ≥35 years) improved AUC up to 14%, primarily due to increased specificity. AI score correlated with size of the zona pellucida and perivitelline space, and ooplasm appearance. AI score also correlated with blastocyst expansion grade and morphological quality. The sum of AI scores from oocytes in group culture images predicted the formation of two or more usable blastocysts (AUC 0.77).
CONCLUSION: An AI model to evaluate oocyte competence was developed using federated learning, representing an essential step in protecting patient data. The AI model was significantly predictive of oocyte competence, as defined by usable blastocyst formation, which is a critical factor for IVF success. Potential clinical utility ranges from selective oocyte fertilization to guiding treatment decisions regarding additional rounds of oocyte retrieval.
DESIGN: In total, 10,677 oocyte images with associated metadata were collected prospectively by eight IVF clinics across six countries. AI training used federated learning, where data were retained on regional servers to comply with data privacy laws. The final AI model required a single image as input to evaluate oocyte competence, which was defined by the formation of a usable blastocyst (≥expansion grade 3 by day 5 or 6 post ICSI).