A machine learning approach to the analysis of psychiatric neuroimaging data. (360G-Wellcome-086565_Z_08_Z)

The World Health Organization (WHO) states that one in four people experience a mental health problem during their life. The development of technologies for assisting psychiatric diagnosis and predicting treatment responses is extremely relevant to avoid inefficient treatment and to promote mental health. The purpose of this research is to develop models and tools for the application of novel machine learning techniques to the analysis of brain scan data to assist in the understanding, diagnosis and prognosis of psychiatric disorders. Key Goals: 1. Development of new machine learning techniques for the analysis of brain scans including: - Probabilistic classifiers to predict group membership (e.g. patients vs. controls and - responders vs. non-responders); - Hypothesis-driven models of brain alterations in psychiatric disorders (e.g. schizophrenia, depression); - New strategies to extract useful information from neuroimaging data (i.e. feature selection approaches). 2. Developme nt of multi-modal classification integrating data from different techniques (e.g. fMRI, MRI, EEG, genetic data). 3. Validation of the developed approaches using a variety of data sets. 4. Translation of the validated approaches into clinical practice, i.e. development of a toolbox that could be used for aid diagnosis of psychiatric disorders and other clinical/outcome goals.