This web site is no longer maintained and the content may be outdated.
Please visit www.cmpe.boun.edu.tr for up-to-date information.
 
CmpE RSS
No upcoming events...

Home / Graduate / PhD Theses Completed
 
 
 
 
  Aslı Uyar Özkaya, 2011  [download thesis]    

Thesis Title

Assessing Machine Learning Methods in IVF Process: Predictive Modeling of Implantation and Blastocyst Development


Abstract

In this thesis, we address the decision-making problems in in vitro fertilizationtreatment from the machine learning perspective aiming to increase the clinical successrates. Initially, we present a comprehensive and comparative analysis of the classifica-tion techniques in embryo-based implantation prediction. In parallel, we evaluate thepredictor effects of input features in order to eliminate the redundant variables anddecide the optimum feature subset leading to the highest prediction performance. Incontrast to the limited relevant literature, our preliminary experiments demonstratethe potential of machine learning classfiers as an automated decision support tool incritical decisions affecting the success of the treatment. Later, we focus on improvingthe classification performance either by algorithmic enhancements or by improving theinformation content of the data. First, we handle the problem of imbalanced class dis-tribution and show that decision threshold optimization and re-sampling the trainingdata produce similar results. Second, we propose a frequency based encoding techniqueto efficiently transform categorical variables into continuous numeric values. And third,in addition to the patient and embryo characteristics, we investigate the effect of in-dividual physicians as a human factor on the pregnancy outcome. Finally, we applyBayesian Networks to model the embryo growth process with the objective of blasto-cyst score prediction. We propose a novel approach to adjust the frequency estimatesfor parameter learning in conditional probability tables. The results of the experimentsshow that (i) the standard machine learning algorithms enable acceptable prediction ofimplantation and blastocyst score and ii) the prediction performance can be improvedby using the proposed techniques in this study. From the clinical perspective, our re-sults have practical implications in reducing multiple pregnancies, preventing waste ofembryos and cancelation of transfers
 
 
Boğaziçi University Department of Computer Engineering
Address: 34342 Bebek, Istanbul, TURKEY
Phone: +90 212 359 4523-24 Fax: +90 212 287 2461
general information: infocmpe.boun.edu.tr   webmaster: webmastercmpe.boun.edu.tr