Current students

In recent years, the growing availability of large-scale multi-modal healthcare data (clinical, genetic, imaging, …) has been paving the way for a more precisely informed health analytics and the development of tools supporting clinical decision towards more personalized approaches.
However, despite the success of single-modal methods in many relevant tasks, integrative omics is still at an early stage and lacks some essential features for clinical adoption (e.g., interpretability, possibility to work with missing data).
My research project focuses on the integration of multi-modal information for predicting cardiovascular risk. In particular, the main underlying methodological steps include the extraction of latent features from complex data and their fusion into a personalized representation of the patient’s cardiovascular profile. The modelling phase relies on the interpretability of latent factors and risk scores, to effectively drive clinical decision making and support expert knowledge in the fields of personalized medicine and early disease detection.
Among the modalities, particular attention is given to cardiac magnetic resonance imaging data and interpretability of the corresponding derived latent features.