Public defence in Computer Science, M.Sc.(Tech) Heli Julkunen

Title of the thesis: Machine Learning for Precision Medicine

Doctoral Student: Heli Julkunen
Opponent: Professor Maik Pietzner, Precision Healthcare University Research Institute, Queen Mary University of London & Berlin Institute of Health (BIH), Germany
Custos: Professor Juho Rousu, Aalto University School of Science, Department of Computer Science

Developing and applying machine learning frameworks to predict drug combination treatments and disease risks

Precision medicine is an emerging approach to healthcare that tailors prevention and treatment strategies by taking individual patient differences into account. Its implementation is becoming more feasible due to advances in the scalability and cost-effectiveness of various molecular profiling technologies, such as genomics, transcriptomics, proteomics, and metabolomics. These advances have not only expanded the amount of molecular data measurable from individuals but also increased the availability of large datasets for research. This has created opportunities to discover more effective treatments, identify disease biomarkers, and develop models for predicting disease risks. However, transforming this data into practical insights for precision medicine requires advanced computational methods.

This thesis develops and applies computational methods to address various aspects of precision medicine. First, this thesis presents a machine learning framework to predict the effects of drug combinations across varying doses, providing an improvement over existing methods by enabling precise dose-specific predictions. This method demonstrated high predictive accuracy and identified novel drug combination synergies, which were subsequently experimentally validated. This method thus provides an efficient tool for systematic pre-screening of drug combinations, particularly to support the development of combination therapies for cancer treatment.

Next, the thesis expands the current understanding of blood biomarkers in disease risk prediction by analyzing population-scale metabolomic data. These studies identified novel metabolomic biomarkers and highlighted their potential in predicting the risks of various diseases, including diseases where metabolomics had not previously been studied at scale. Finally, the thesis presents a machine learning method aimed at improving disease risk prediction by incorporating comprehensive interaction effects among risk factors. This method demonstrated improved accuracy in risk prediction compared to standard methods across multiple diseases and different data sources, thereby supporting the development of more accurate tools for risk assessment.

In summary, the computational methods and insights presented in this dissertation advance the translation of molecular data into prevention and treatment strategies in precision medicine.

Keywords: precision medicine, machine learning, predictive modelling, survival analysis, risk prediction, metabolomics, drug combinations

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/ 

Contact information: 

Doctoral theses of the School of Science: https://aaltodoc.aalto.fi/handle/123456789/52