Beta 1


Title System Identification for Prediction of Glucose Levels in People with Type 1 Diabetes
Author Pedersen, Søren Nygaard
Hansen, Lasse Bergenholz
Supervisor Jørgensen, John Bagterp (Scientific Computing, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Finan, Daniel Aaron (Scientific Computing, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Bachelor thesis
Year 2010
Abstract In this thesis the possibility of predicting blood glucose levels in a type 1 diabetes patient, using ARX models has been explored. The Hovorka Model was used to simulate data from a type 1 diabetes patient. Using Matlab and the System Identification Toolbox, data generated by the Hovorka Model were utilized for identification and validation of single and double input ARX models. Results showed that identification data are of central importance to the qualities and versatility of the identified ARX models. As validation, the models were used to predict data from virtual scenarios and errors were evaluated through a Clarke Error Grid analysis. Single input models showed a significant lack of versatility, in many cases expressed by an alarming distribution of data points in the Clarke Error Grid, but several of the double input models gave promising results.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-B.Sc.-2010-13
Fulltext
Original PDF bac10_13_net.pdf (2.46 MB)
Admin Creation date: 2010-06-16    Update date: 2010-06-16    Source: dtu    ID: 263827    Original MXD