Applications of Machine Learning and Normative Modeling to Mental Disorders

Mental disorder prediction using MRI biomarkers and advanced machine learning techniques holds great promise. However, despite advances, the predictions at the level of individuals remain inadequate. The project, "Applications of Machine Learning and Normative Modeling to Mental Disorders," aims to improve such predictions by utilizing two state-of art methods: normative modeling and machine learning using large datasets. Specifically, the project seeks to (1) develop and implement machine learning pipelines that can accurately classify diagnostic categories and identify neuroimaging-based biomarkers from MRI data; (2) utilize deep learning techniques to augment task-based fMRI data, which will be used in normative modeling analyses; and (3) perform normative modeling to map between brain readouts and demographic and clinical scores to investigate heterogeneity in neuroimaging-based biomarkers related to mental disorders. 

Members: Emin Serin
Head/Supervisor: Prof. Henrik Walter, Prof. Kerstin Ritter, Prof. Andre Marquand
contact: Emin Serin