The DIMENTIA project is large multidisciplinary project for Data governance in the development of machine learning algorithms to predict neurodegenerative disease evolution. In this Innoviris Joint R&D project, Collibra N.V. teams up with UZ Brussel and three VUB research groups in order to develop a clinical-value-driven data-sharing ecosystem that enable the collection and processing of big clinical data, for the purpose of predictive personalised medicine.
The AIMS group takes part of this project by evaluating the system for the prediction of physical and cognitive deterioration in Multiple Sclerosis (MS). More specifically, we will leverage existing big-data that was collected from MS patients over the years at UZ Brussel and the National MS Center Melsbroek. We will analyse multimodal data such as rest-and task-based electrophysiological data for predicting physical and cognitive deterioration. More specifically, we will build a semi-automated pipeline for the pre-processing of electroencephalography (EEG). Resting-state EEG features will be extracted using micro-state analysis, frequency analysis, functional connectivity and graph theoretical measures. In addition, task-based EEG data features—such as amplitude and peak latency of evoked potentials—collected during a visual and auditory oddball paradigm will be extracted. With current state-of-the art machine learning techniques, we will leverages these neurophysiological markers for the prediction of physical and cognitive impairment in patients with MS.