Short Summary/Abstract:

This project aims to externally validate supervised and unsupervised approaches for identifying treatment resistance in patients with schizophrenia spectrum disorders, utilising datasets from the PsyShareD consortium. The supervised predictive modelling will assess the performance of machine learning resistance in patients with schizophrenia spectrum disorders, utilising datasets from the PsyShareD consortium. The supervised predictive modelling will assess the performance of machine learning algorithms in predicting treatment resistance, incorporating demographic, clinical, and structural neuro imaging metrics derived from T1-weighted MRIs. The unsupervised subgroup clustering will evaluate the proportion of treatment resistance within patient subgroups identified by two distinct algorithms applied to the same datasets. Additionally, the robustness of both approaches will be examined concerning clozapine response and longitudinal treatment outcomes. Finally, critical regional metrics from the predictive and clustering models will be compared to uncover convergent insights. This project aims to highlight the clinical utility of structural MRIs and demographic data in identifying treatment resistance, facilitating more informed clinical decision-making.