Πανεπιστήμιο Στερεάς Ελλάδος
Intelligent mining of biomedical data for the creation of integrated physiological models
Abstract
dc:descriptionIn the context of contributing to the creation of Integrated Physiological Models, three novel approaches are proposed and developed, targeting at different levels of biomedical modelling. At the lower molecular level, the integration of metabolic pathways in a way capable of deriving descriptive and predictive dynamic models resulted in the creation of the KEGGconverter tool. KEGGconverter is capable of producing integrated analogues of metabolic pathways appropriate for simulation tasks, by inputting only KGML files and having an SBML model as output. The automated introduction of four case-specific default kinetic mechanisms in the models provides for the addition of the layer of kinetic equations, which is directed by a rule-based algorithm that was implemented for the specific task. At a higher level, the novel GOrevenge algorithm exploits the Gene Ontology to map genes to specific cellular pathways and vice versa in order to further infer the putative functional role of specific genes, through an associative context, starting from the results of various statistical enrichment analyses. Although this implementation utilizes the GO annotations, the algorithm can be generically extended to accommodate any biological ontology or controlled vocabulary definition. This is due to the exploitation of the encapsulated underlying topological network information and the interplay among the annotations and the annotated subjects. Lastly, a novel methodology on multi-modal data fusion regarding separate datasets has been developed utilizing a dataset of features from skin lesion images and a dataset regarding microarray data. Both datasets were the output from studies on cutaneous melanoma, but involved different patients. The multivariate analysis applied to the unified datasets that were produced by this method indicated the better discrimination performance achieved in predicting the class of healthy/disease samples. This could lead not only to the creation of better analytical models of the specific disease, but also in dealing with modelling other complex diseases having multi-modal datasets as well.
Degree
thesis:*- Grantor dc:publisher
- Πανεπιστήμιο Στερεάς Ελλάδος
- Year dc:date
- 2013
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Moutselos, Konstantinos
- Μούτσελος, Κωνσταντίνος
Subjects
dc:subject × 28- Εικονική ανθρώπινη φυσιολογία
- Μεταβολικά μονοπάτια
- Κινητικές εξισώσεις
- Συστημική βιολογία
- Οντολογία γονιδίων
- Συγκερασμός δεδομένων
- Μοντέλα
- Τυχαία δάση
- Δερματικό μελάνωμα
- VPH
- KEGG
- SBML
- GO
- Virtual physiological human
- Metabolic pathways
- Kinetics
- Data fusion
- Models
- Random forests
- Cutaneous melanoma
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Επιστήμες Μηχανικού και Τεχνολογία
- Βιοϊατρική Μηχανική
- Natural Sciences
- Computer and Information Sciences
- Engineering and Technology
- Medical Engineering
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/28595