PhD Thesis Final Defense to be held on 12 June 2018 at 14:15 (Multimedia amphitheater, Central Library of NTUA)
Thesis Title: Bioinformatics, Computational Systems Biology and Machine Learning Methods applied on In Silico Oncology
In the context of the present thesis, the scientific disciplines of Bioinformatics, Computational Systems Biology and Machine Learning have been studied, aiming at exploiting them in the framework of the constantly growing and evolving field of In Silico Oncology. This field has as its central objective the development of computational models capable to simulate the growth of cancerous tumors as well as their response to therapy. The In Silico Oncology and In Silico Medicine group of the Institute of Communication and Computer Systems of the National Technical University of Athens (ICCS-NTUA) has nodal involvement in this through its efforts on developing a family of models, the Oncosimulators. These specific Oncosimulators, are focusing mainly on the simulation of phenomena of the cellular and higher levels. In the present thesis, the aspects in which methods of the fields of Bioinformatics, Computational Systems Biology and Machine Learning may contribute in the extension of the Oncosimulators to the molecular space as well as in the exploitation of them for predictive procedures regarding the personalized response of tumors to therapy. In order for these to be presented, specific applications has been selected focusing on the case of Acute Lymphoblastic Leukemia through the usage and the study of the extendibility of the Leukemia Oncosimulator. Specifically, a Systems Biology oriented model for the simulation of the biochemical regulation of the cell cycle in Acute Lymphoblastic Leukemia as well as a Pharmacokinetic model for the Drug Prednisone, capable to provide input to the Oncosimulator, have been developed and trained through computational optimization methods. Moreover, through the usage of Bioinformatics and Machine Learning methods, a Hybrid Leukemia Oncosimulator System has been developed, which consists of procedures for automated adaptation of the Oncosimulator on patients data as well as the prediction of the response of patients to therapy through training machine learning models. More specifically, the prediction of the response to Prednisone of Acute Lymphoblastic Leukemia pediatric patients, which has a central role in the patient stratification, has been attempted. The original effort and the adequate success in the prediction by the Hybrid Oncosimulator consist of a crucial step for the further development of the field of In Silico Oncology.
Supervisor: Stafylopatis Andreas-Georgios, Professor
Co-Supervisor: Stamatakos Georgios
PhD student: Ouzounoglou Eleftherios
The examination is open to anyone who wishes to attend.