Abstract: BACKGROUND: Unexplained infertility is associated with a higher abortion rate and lower fertilization rate, implantation rate, clinical pregnancy rate and cumulative live birth rate. It is urgent to establish a clinical prediction model related to infertility of unknown cause to solve the problems of clinical prognosis and individualized medical services, and finally achieve the purpose of increasing the cumulative live birth rate of patients with infertility of unknown cause.
OBJECTIVE: To construct and verify the prediction model of high-quality blastocyst formation in patients with unexplained infertility during in vitro fertilization.
METHODS: A total of 419 patients with unknown infertility who underwent in vitro fertilization in the Assisted Reproduction Department of Changzhou Maternal and Child Health Care Hospital from March 2017 to June 2022 were retrospectively analyzed, including 317 patients with high-quality blastocysts and 102 patients without high-quality blastocysts. A prediction model was established and used as the model group. The model group was sampled 1 000 times by the Bootstrap method as the validation group. Firstly, the univariate analysis was used to screen the influencing factors of high-quality blastocyst formation of unknown infertility, and the best matching factors were selected by the least absolute shrinkage and selection operator (LASSO) algorithm. Multiple factors were included in the progressive Logistic regression to find out the independent influencing factors and draw a column graph. Finally, the subject working curve, calibration curve, clinical decision curve and clinical impact curve were used to verify the differentiation and accuracy of the prediction model as well as the clinical application efficiency.
RESULTS AND CONCLUSION: (1) Univariate analysis of the factors influencing the formation of high-quality blastocyst of unknown infertility were age, insemination method, antimullerian hormone level, basal follicle-stimulating hormone level, basal luteinizing hormone level, human chorionic gonadotropin injection day follicle-stimulating hormone level, human chorionic gonadotropin day estradiol level, progesterone level on human chorionic gonadotropin day, the number of high-quality cleavage embryo (day 3) and the number of blastocyst formation (P < 0.05). (2) The best matching factors further screened by LASSO regression were age, insemination method, antimullerian hormone level, basal luteinizing hormone level, human chorionic gonadotropin injection day follicle-stimulating hormone level, human chorionic gonadotropin day estradiol level, the number of high-quality cleavage embryo (day 3) and the number of blastocyst formation (P < 0.05). Multifactor stepwise Logistic regression results showed that independent influencing factors on the formation of high-quality blastocysts for unexplained infertility were age, insemination method, antimullerian hormone level, the number of high-quality cleavage embryo (day 3), and the number of blastocyst formation. (3) Receiver operating characteristic curve exhibited that the area under the curve was 0.880(0.834, 0.926) in the model group and 0.889(0.859, 0.918) in the validation group. It showed that the prediction model had good differentiation. The average absolute error of the calibration curve was 0.036, indicating that the model had good accuracy. The Hosmer-Lemeshow test showed that there was no statistical difference between the prediction probability of blastocyst formation and the actual probability of blastocyst formation (P > 0.05). The clinical decision curve and clinical impact curve showed that the model group and the validation group had the maximum clinical net benefit when the threshold probability value was (0.16-0.96) and (0.08-0.93), respectively, and had better clinical application efficacy within the threshold probability range. These findings concluded that age, insemination method, antimullerian hormone, the number of high-quality cleavage embryos (day 3), and the number of blastocyst formation were independent factors influencing the formation of the fine blastocyst in patients with unexplained infertility. The clinical prediction model constructed by these factors has good clinical prediction value and clinical application efficiency and can provide a basis for clinical prognosis and intervention as well as the formulation of individual medical programs.

Key words: unexplained infertility, prediction model, high-quality blastocyst, nomogram, live birth