National Technical University of Athens (NTUA)
Ευφυή συστήματα υποστήριξης εξατομικευμένων ιατρικών αποφάσεων για τη διαχείριση του σακχαρώδους διαβήτη
Abstract
dc:descriptionThe scope of the present thesis is the design, development and evaluation of intelligent medical decision support systems, aiming at optimizing the treatment of patients with Diabetes Mellitus (DM). Specifically, within the framework of the present thesis, several methods have been developed for the analysis and processing of data related to medical electronic health records, laboratory measurements and continuous glucose and insulin records, towards the design and the development of: i) an intelligent Insulin Infusion Advisory System (IIAS), able to provide real time estimations of the appropriate insulin infusion rates for type 1 DM patients using continuous glucose monitors and insulin pumps (Artificial Pancreas), in order to maintain glucose levels within the physiological range, and ii) models for the risk assessment of long-term complications of Type I and Type II DM, focusing on diabetic retinopathy. In the first part of this study, a simulation model of the glucose - insulin metabolism system of Type I DM has been developed, based on the combined use of compartmental models (CMs) and Neural Networks (NNs). The model is incorporated into a Model Predictive Controller (MPC) in order for the latter to estimate insulin infusion rates. In order to evaluate the performance of the IIAS, several in silico trials have been performed. Moreover, a clinical trial has been conducted under controlled conditions, the results of which provided valuable input for SSEI improvement. In particular, a tuning algorithm based on fuzzy logic has been developed, for the on line adaptation of the NMPC parameters. The enhanced SSEI has been tested against its ability to handle meal disturbances, fasting conditions, delays, noise to the glucose measurements, interpatient variability, and erroneous meal size estimations. The second part of the present thesis, concerns the development of models able to estimate the risk of people with Type I or Type II DM to develop over time, long-term diabetes complications, with particular focus to diabetic retinopathy. Towards this direction, advanced classification techniques are applied based on the wavelet neural networks. For the development and the evaluation of the aforementioned medical decision support systems, data from Type I and Type II DM patients, have been used, acquired from the Department of Pediatrics, Diabetes Center, P&A Kyriakoy Children’s Hospital, Athens and the Athens Hippokration Hospital.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2011
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Zarkogianni, Konstantia
- Ζαρκογιάννη, Κωνσταντία
Subjects
dc:subject × 24- Σακχαρώδης διαβήτης
- Γλυκόζη
- Ινσουλίνη
- Αμφιβληστροειδοπάθεια
- Τεχνητό πάγκρεας
- Υποδόρια οδός
- Ελεγκτής βασισμένος σε μοντέλο πρόβλεψης
- Αυτοσυντονιζόμενος ελεγκτής
- Διαμερισματικά μοντέλα
- Νευρωνικά δίκτυα με ανάδραση
- Diabetes mellitus
- Glucose
- Insulin
- Retinopathy
- Artificial pancreas
- Subcutaneous route
- Model predictive control
- Autotuning controller
- Compartmental models
- Recurrent neural networks
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/24790