Controlled ovarian stimulation is tailored to the patient based on clinical parameters but estimating the number of retrieved metaphase II (MII) oocytes is a challenge. Here, we have developed a model that takes advantage of the patient’s genetic and clinical characteristics simultaneously for predicting the stimulation outcome. Sequence variants in reproduction-related genes identified by next-generation sequencing were matched to groups of various MII oocyte counts using ranking, correspondence analysis, and self-organizing map methods. The gradient boosting machine technique was used to train models on a clinical dataset of 8,574 or a clinical-genetic dataset of 516 ovarian stimulations. The clinical-genetic model predicted the number of MII oocytes better than that based on clinical data. Anti-Müllerian hormone level and antral follicle count were the two most important predictors while a genetic feature consisting of sequence variants in the GDF9, LHCGR, FSHB, ESR1, and ESR2 genes was the third. The combined contribution of genetic features important for the prediction was over one-third of that revealed for anti-Müllerian hormone. Predictions of our clinical-genetic model accurately matched individuals’ actual outcomes preventing over- or underestimation. The genetic data upgrades the personalized prediction of ovarian stimulation outcomes, thus improving the in vitro fertilization procedure.
Study question Is it possible to predict clinical pregnancy using Machine Learning models on day 6 after embryo transfer to an acceptable degree of certainty? Summary answer Using selected biochemical markers it is possible to build a machine learning model to predict clinical pregnancy. What is known already Human chorionic gonadotropin (β-hCG) is considered the main early predictor of pregnancy. Due to the fact that the highest percentage of spontaneous abortions occurs in the first trimester predicting clinical pregnancy outcomes at the early stages of pregnancy is particularly difficult. The level of B-hcg measured at the early stages of implantation and the dynamics of its increase can help predict clinical pregnancy and predict its further development. However, the specific cut-off points at which pregnancy can be determined vary from study to study, and a different factor is often suggested to increase the certainty of β-hCG-based predictions. Study design, size, duration 1474 single frozen embryo transfers (1205 patients between 23 and 47 years of age), January 2019 -December 2022 were used for data modelling and statistical testing. The dataset consisted of endocrinal markers:β-hCG, PRG (progesterone), E (estradiol) measured on the 6th day after ET and other practitioner-chosen data: age, PGD results, embryo quality status, RIF or RPL history, embryo maturation days as well as pregnancy outcomes. Exclusion criteria were: PRG > 50ng/ml, β-hCG >70 mIU/ml. Participants/materials, setting, methods A machine learning model for predicting clinical pregnancy was trained using the gradient boosting technique. The model used 100 decision trees with 8 leaves, maximal depth of 2 nodes and a learning rate of 0.1. The performance of the model was measured using the accuracy score metric and area under ROC (AUC). Feature importance was based on Shapley values. Additional analysis was performed using statistical testing using a two-sided Student’s t-test. Main results and the role of chance A Machine Learning naive model employed chosen variables to predict clinical pregnancy. The most important variables turned out to be the serum level of β-hCG and PRG on day 6 after embryo-transfer. The final model, limited to those variables, performs with an accuracy of around 83 % (accuracy score = 0.834), the area under ROC (AUC) was 0.927. Based on Shapley values, β-hCG followed by PRG are the most relevant for the model’s predictions. Low PRG concentrations increased the chances of clinical pregnancy. The findings were confirmed by additional statistical analysis. Observations were divided into two groups for which the cut-off point was the level of PRG of 24 ng/ml – the median for the whole group. The probability of clinical pregnancy is higher in the group where the serum PRG is below 24 ng/ml even if β-hCG levels are the same. A statistically significant difference between these groups regarding β-hCG (P value < 0.001) was noticed. In order for serum β-hCG to be used as a more accurate clinical pregnancy predictor its level should be interpreted in the context of PRG serum levels. Limitations, reasons for caution Since in the dataset, only data regarding frozen embryo transfers were included, fresh ETs need to be explored further. The model was based on endocrinal markers as measured on day six after ET hence it cannot be used to interpret those markers as measured on any other date. Wider implications of the findings In order to most accurately inform patients regarding the possible outcomes of embryo-transfers, practitioners should not limit themselves to monitoring β-hCG serum levels but also include PRG as an important prognostic of clinical pregnancy. Trial registration number Not Applicable
Although the residential property market has strong connections with various sectors, such as construction, logistics, and investment, it works through different dynamics than do other markets; thus, it can be analysed from various perspectives. Researchers and investors are mostly interested in price trends, the impact of external factors on residential property prices, and price prediction. When analysing price trends, it is beneficial to consider multidimensional data that contain attributes of residential properties, such as number of rooms, number of bathrooms, floor number, total floors, and size, as well as proximity to public transport, shops, and banks. Knowing a neighbourhood's key aspects and properties could help investors, real estate development companies, and people looking to buy or rent properties to investigate similar neighbourhoods that may have unusual price trends. In this study, the self-organizing map method was applied to residential property listings in the Trójmiasto area of Poland, where the residential market has recently been quite active. The study aims to group together neighbourhoods and subregions to find similarities between them in terms of price trends and stock. Moreover, this study presents relationships between attributes of residential properties.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.