Research Area: Data Mining
The research area of the thesis is the application of data mining in healthcare and medicine. When applying data mining in medicine, additional problems such as varied information representation formats, semantic interoperability and patient privacy have to be resolved. The object of the dissertation research is the process and methods of data mining in medicine. The following topics are directly associated with this subject: medical data preprocessing methods, medical images processing, and multi-relational data mining.
The key goal of the thesis is to develop and explore methodology for the application of data mining methods in medicine and healthcare, which would increase the efficiency of data analysis. Achieving this goal, the following tasks have been completed: analysis of the existing methodologies and process models, creation and trial of the data mining application methodology for the medical domain, developing the supporting diagnosing models and medical data processing methods.
In this thesis, a new application methodology CRISP-MED-DM was developed, which is based on the industry standard Cross-Industry Standard Process for the Data Mining. The CRISP-MED-DM was successfully applied for predictive modeling in cardiology and oncology domains.
In addition, a new blood flow electrocardiography image processing technique was developed, which enables semi-automation of aortic valve stenosis degree diagnostics. In addition, a new similarity measure for multi-relation clustering was proposed. The research results of the work revealed new opportunities in the application of data mining methods in the medical domain.
Name of the Researcher: Olegas Niaksu
Name of the Supervisor(s): Olga Kurasova
Year of Completion: 2015
University: Vilnius University
Thesis Link: Home Page Url
