作者: Seyed Abolghasem Mirroshandel , Fatemeh Ghasemian , Sara Monji-Azad
DOI: 10.1016/J.CMPB.2016.09.013
关键词:
摘要: Prediction of pregnancy rate using data mining.Working on unstained sperms.Ability to work low resolution images.High accuracy and real-time processing time.Preparing a freely available dataset. Background objectiveAspiration good-quality sperm during intracytoplasmic injection (ICSI) is one the main concerns. Understanding influence individual morphology fertilization, embryo quality, probability most important subjects in male factor infertility. Embryologists need decide best for real time ICSI cycle. Our objective predict quality zygote, embryo, implantation outcome before each an cycle infertility with aim providing decision support system selection. MethodsThe information was collected from 219 patients at therapy center Alzahra hospital Rasht 2012 through 2014. The prepared dataset included 1544 injected sperms into related oocytes. In our study, transfer performed day 3. Each represented thirteen clinical features. Data preprocessing first step proposed mining algorithm. After applying more than 30 classifiers, 9 successful classifiers were selected evaluated by 10-fold cross validation technique precision, recall, F1, AUC measures. Another experiment measuring effect feature prediction process. ResultsIn zygote prediction, IBK RandomCommittee models provided 79.2% 83.8% respectively. KStar model achieved 95.9% which even better human experts. All these predictions can be done time. ConclusionsA machine learning-based would helpful selection phase improve success treatment.