ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Analyzing Alzheimer's patients' data with machine learning methods

Igor Murgić (2017) Analyzing Alzheimer's patients' data with machine learning methods. EngD thesis.

Download (2995Kb)


    The aim of the diploma thesis is to analyze the data concerning patients with Alzheimer’s disease and to use the predictive models constructed through machine learning methods. The collected data was analyzed and the laws between attributes were defined. The data attributes were presented in the form of an undirected graph. The most relevant attributes were determined using the constructed models, the attributes that caused overfitting were eliminated. The models thus obtained were tested through cross-validation and the accuracy of each model was calculated. The constructed models and the comparisons between them showed that certain attributes were more distinctive than others. These attributes would enable us to simplify and expedite the establishment of the diagnosis of the disease, conducting fewer tests. Doctors deem the elimination of certain tests unreasonable, though, since a lot of information on the patient’s condition can be deduced from them. We could, however, modify the sequence of the tests, which would lead to more rapid establishment of the diagnosis.

    Item Type: Thesis (EngD thesis)
    Keywords: machine learning, Alzheimer's disease, data analysis, clustering, classfication, cross-validation, decision trees, undirected graphs
    Number of Pages: 60
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    izr. prof. dr. Matjaž Kukar267Mentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1537618627)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 4003
    Date Deposited: 19 Oct 2017 12:31
    Last Modified: 25 Oct 2017 12:27
    URI: http://eprints.fri.uni-lj.si/id/eprint/4003

    Actions (login required)

    View Item