skip to primary navigationskip to content

Tiago Azevedo

Machine Learning and Multi-scale Modelling of Tauopathies


Neurodegenerative diseases (NDs) can be seen as conditions which primarily affect the neurons in the human brain and include known diseases such as Parkinson's disease (PD), Alzheimer's disease (AD) and Progressive Supranuclear Palsy (PSP). Currently, no cure has been found to these maladies and treatments focus are in slowing the symptoms. All these diseases fall in a category of NDs named "tauopathy". Diseases from this category are characterised as having an abnormal aggregation of a protein called tau, a commonly known marker of neurodegeneration.


A relatively recent approach to study neurodegeneration is the study and mapping of the neural connections in the brain or, in other words, the brain connectome. So, it is clear that there are at least two scales of brain organisation: molecular and connectome scales. In order to consider these two scales of brain, the research question I want to explore during my PhD is "what model could capture the molecular interactions and macroscopic consequences of tauopathies?''. To do so, I want to have a multi-scale approach to fill the knowledge gap between the molecular and connectome scales to understand neurodegeneration disruptions.


We will be able to study what are the main disruptions that occur in the brains of patients considering not only the spatial component, but also the time variability. Also, from trained models it will be possible to make predictions: how can we predict how a certain brain will evolve with time? How integration of molecular data, such as gene expression of MAPT or spatial distribution of abnormal tau in different brain regions, correlates with all the information we have from healthy and diseased brain networks? What are the best therapeutic interventions we can think of when analysing these networks? From here, we could extend the analysis to other tauopathies.