Topic: cognitive phenotypes

Modest success has been made in using resting state fMRI (rs-fMRI) to predict individualized behaviors. Here, we describe a scalable and modular system for individualized representation learning of rs-fMRI. The system includes a “deep” base with “shallow” adds-on. The “deep” base is trained through multiple stages of self-supervised learning with unlabeled rs-fMRI. After the base is trained and fixed, the “shallow” adds-on are trained with supervised learning for behavior prediction. The base includes a variational auto-encoder (VAE) for unsupervised learning of spatial representations, a Bi-d

To identify patterns of structural networks associated with cognitive phenotypes detected with robotic technology in patients with temporal lobe epilepsy (TLE). Network analysis was performed using multimodal MRI from TLE patients (n=22) and healthy subjects (n=20). Five tests with the Kinarm robot were used to evaluate five cognitive domains. Four graph theory metrics were computed to characterize structural networks. Structural and cognitive phenotypes were identified using an unsupervised learning algorithm, and the association between them evaluated. Three cognitive clusters were identifie