Big Data approaches for prediction of disease treatment efficacy and related outcomes

Disease trajectories from patient derived data have shown to be highly individual and disease aetiology and treatment response in the majority of diseases are insufficiently elucidated. These individual differences in sensitivity and response to treatment are poorly understood, but increasing evidence suggests that individual factors such as different exposures in the environment, host genetics or bacteria hosted in the human gut play an important role by predisposing individuals towards treatment success or complications. Identification of biomarkers associated with short- and long-term impact of intervention and/or treatments may allow for early prediction of treatment outcomes and toxicities, which will be able to determine patient subgroups at greatest risk.
The overall objective of the PhD is to develop tools and methodologies to integrate patient derived data and apply this to improve understanding of underlying factors for disease and treatment response. This involves:
• Elucidating characterisation of biomarkers predisposing individuals response to interventions or treatment and adverse effects
• Stratification of individuals based on biomarkers by developing prediction tools to uncover features that are predictive of response to interventions and/or treatment and adverse effects
Artificial intelligence models will be developed to integrate pathways and networks-based genomics inputs with patient characteristics, disease severity, treatment and early treatment responses to predict late outcomes and perform feature selection. Machine learning will be used to predict drug response subgroups.
The PhD project will commence by applying machine learning on the 3G project. The project will later continue focusing on acute lymphoblastic leukaemia (ALL) and testicular cancer as the main projects. As the main PhD projects commence, the methods may be developed and applied to other disease areas including inflammatory bowel disease (IBD) and diabetes depending on data availability.