prikaz prve stranice dokumenta Personalized data-driven prediction algorithms for glycemic trend
Access restricted to students and staff of home institution
doctoral thesis
Personalized data-driven prediction algorithms for glycemic trend
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. urn:nbn:hr:168:201513
Žulj, Sara
University of Zagreb
Faculty of Electrical Engineering and Computing
Department of Electronic Systems and Information Processing

Institutional repository: FER Repository

Cite this document

Žulj, S. (2022). Personalized data-driven prediction algorithms for glycemic trend (Doctoral thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:201513

Žulj, Sara. "Personalized data-driven prediction algorithms for glycemic trend." Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:201513

Žulj, Sara. "Personalized data-driven prediction algorithms for glycemic trend." Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:201513

Žulj, S. (2022). 'Personalized data-driven prediction algorithms for glycemic trend', Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 25 December 2024, https://urn.nsk.hr/urn:nbn:hr:168:201513

Žulj S. Personalized data-driven prediction algorithms for glycemic trend [Doctoral thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2022 [cited 2024 December 25] Available at: https://urn.nsk.hr/urn:nbn:hr:168:201513

S. Žulj, "Personalized data-driven prediction algorithms for glycemic trend", Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2022. Available at: https://urn.nsk.hr/urn:nbn:hr:168:201513

Metadata
TitlePersonalized data-driven prediction algorithms for glycemic trend
Title (croatian)Personalizirani podatkovno vođeni algoritmi za predviđanje glikemijskoga trenda
AuthorSara Žulj
MentorRatko Magjarević (mentor)
MentorPaulo de Carvalho https://orcid.org/0000-0002-6920-869X (komentor)
Committee memberIgor Lacković (predsjednik povjerenstva)
Committee memberAlan Jović (član povjerenstva)
Committee memberLana Ružić (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
(Department of Electronic Systems and Information Processing)
Zagreb
Defense date and country2022-06-27, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Universal decimal classification
(UDC)		004 - Computer science and technology. Computing. Data processing
Abstract
An important aspect of diabetes management is controlling the glucose level. A step towards achieving the control is monitoring and taking actions (e.g., administering insulin) to avoid dangerous levels. Predicting future values may allow the user to be alerted when an action is needed. This thesis investigates an interpretable case-based reasoning (CBR) model for predicting future glucose levels. The focus of this work lies in exploring the different hyperparameters of CBR model. The work also ev104 - 117aluates the effect of meal intake as an input variable in a data-driven way. This is in contrast to previously published research which use dedicated physiological models to include the meal. The effect of the dataset size on the prediction error is modelled. Models are often trained on very small datasets and reported error often does not reflect the large variance caused by the dataset size. Further, the personalization of the CBR model by clustering the patient data prior to building the CBR model’s case base is studied. The clustering is done using clinical, behavioral, anthropometric and demographic data. Finally, a new two-stage method for maintaining the case base of the CBR regression problem is proposed. The contributions are: • A case-based reasoning model for prediction of glucose levels, • A personalization of the model through a clustering strategy, • A method for updating and maintaining the case base.
Abstract (croatian)
Kontrola razine glukoze važan je aspekt upravljanja dijabetesom. Korak prema ostvarivanju kontrole je praćenje i poduzimanje adekvatnih radnji (npr. davanje inzulina) kako bi se izbjegle opasne razine. Predviđanje budućih vrijednosti omogućuje upozorenja korisniku kako bi na vrijeme poduzeo potrebne radnje. Ovaj rad istražuje interpretabilan model temeljen na zaključivanju na temelju slučaja (CBR) za predviđanje budućih razina glukoze. Fokus istraživanja leži u odabiru različitih hiperparametara CBR modela. Također, istražen je učinak unosa obroka kao ulazne varijable na podatkovno-vođeni način. To je u suprotnosti s prethodno objavljenim istraživanjima koja koriste fiziološke modele kako bi uključili obrok. Modeliran je učinak veličine skupa podataka na pogrešku predviđanja. Modeli često uče na vrlo malim skupovima podataka te iskazana pogreška modela često ne odražava značajnu varijancu pogreške uzrokovanu veličinom skupa podataka. Nadalje, istražena je personalizacija CBR modela grupiranjem podataka o pacijentu prije izgradnje baze slučajeva CBR modela. Grupiranje se provodi pomoću kliničkih podataka, podataka o ponašanju, antropometrijskih i demografskih podataka. Konačno, predložena je nova metoda u dvije faze za održavanje baze slučajeva CBR regresijskog problema. Ostvareni doprinosi su: • Model temeljen na zaključivanju na temelju slučaja za predviđanje razine glukoze, • Personalizacija modela kroz strategiju grupiranja, • Metoda za ažuriranje i održavanje baze slučajeva.
Keywords
Keywords (croatian)
Languageenglish
URN:NBNurn:nbn:hr:168:201513
Promotion2022
Study programmeTitle: Computer Science
Study programme type: university
Study level: postgraduate
Academic / professional title: Doktor znanosti (Doktor znanosti)
Type of resourceText
Extentiii, 130 str. : graf. prikazi ;
File originBorn digital
Access conditionsAccess restricted to students and staff of home institution
Terms of useLicence In copyright icon
Created on2022-09-05 07:42:38