| Resumo: | Cancer is an heterogeneous disease, with a high degree of diversity between tumours. Biomarkers, in the context of an oncological disease, allow the identification of the response from a patient to a given drug. These specific treatments have been producing results that are superior on average to broader ones. However, the relationship between a drug's response a biomarkers value is in many cases yet unknown. Some models to predict this relationship have already been built, using machine learning methods. The input are characterizations of both the drug and the tissue along with the result of the drug's use on a given tissue. The goal of this thesis is to improve on previous models and the characterization of both the drug and the tissue through the introduction of graph mining and other machine learning methods.
|
|---|