Modelling MEG-based functional connectivity and EMG activity to predict inter-individual differences in motor tasks

In the last years, several neuroscience studies focused on individual characterization based on functional connectome. Indeed, profound inter-individual differences can be detected across different individuals and they act as fingerprint to identify each subject among others. The main objective is investigating how these differences are reflected in individual behaviour. This thesis work is linked with ERC HANDmade project (SH4, ERC- 2017-STG) directed by professor Viviana Betti, which investigates how natural hand usage shapes behaviour as well as intrinsic and task-evoked brain activity. In particular, this thesis work aims at modelling MEG-based functional connectivity and EMG activity to predict inter-individual differences in motor tasks. Two different multiple linear regression models will be employed to find the relationship between the brain connectome at rest and EMG features extracted by a set of elementary movements, as finger tapping and toe squeezing. A regression model is based on a LOO prediction technique which does not exploit previous knowledge on EMG feature typical values, the second one is a standard Multiple Linear regression which exploits a previous knowledge on EMG data. Results indicate that standard multiple linear regression model outperforms the LOO prediction model, especially considering the whole connectome, the visual peripherical network and motor network.