Due to copyright issues, an electronic copy of the thesis must be ordered from the faculty. For the faculty to have time to process the order, the order must be received by the faculty at the latest 2 days before the public defence. Orders received later than 2 days before the defence will not be processed. After the public defence, please address any inquiries regarding the thesis to the candidate.
Trial Lecture – time and place
See Trial Lecture.
Adjudication committee
- First opponent: Research Director Leena Latonen, University of Eastern Finland,
- Second opponent: Professor Inge Jonassen, University of Bergen,
- Third member and chair of the evaluation committee: Assistant Centre Director and Group Leader, Anthony Mathelier, University of Oslo
Chair of the Defence
Professor Karl-Johan Malmberg, University of Oslo
Principal Supervisor
Professor Tero Aittokallio, University of Oslo
Summary
Cancer is a type of disease associated with various genetic changes. Omics data, such as genomics, transcriptomics, and proteomics, provide comprehensive description into the molecular mechanisms driving cancer development and progression. However, the high dimensionality of omics data presents challenges for analysis, as traditional statistical tools can struggle to handle it. Modern artificial intelligence and machine learning techniques offer a solution to this challenge due to their ability to handle high-dimensional datasets, recognize complex patterns, and improve predictive accuracy. Such multi-dimensional machine learning analysis can help identify novel biomarkers for early diagnosis, treatment response, and prognostic indicators. Therefore, it is important to explore the potential applications of machine learning to diverse omics data in the field of precision oncology, offering new insights into various cancers and providing predictive models for potential clinical use.
In this thesis, I used multi-omics data and machine learning techniques to predict clinical outcomes for three distinct cancer types. In Paper I, we developed a supervised machine learning framework using multi-omics data to investigate the genetic heterogeneity between individuals with ductal carcinoma in situ and invasive breast cancer. In Paper II, we developed a Lasso-based model to predict the treatment response of the drug umbralisib in chronic lymphocytic leukemia patients. The model was constructed using (phospho)protein profiling data and leave-one-out cross-validation. In Paper III, we investigated potential prognostic signatures for non-small cell lung cancer patients. The final biomarkers were used to construct a risk score.
As a summary, in all the three papers in this thesis, machine learning tools provided powerful and accurate modelling for multi-omics analysis. Moreover, the results revealed new biological knowledge about cancers and provided potential applications for clinical use. Additionally, given that the sample sizes are relatively small in this thesis, a discussion about machine learning analysis on small-sample data is included.
Additional information
Contact the research support staff.