Elucidating depression heterogeneity using clinical, neuroimaging and genetic data

The present thesis contributes to the understanding of depression and elucidating its heterogeneous nature. In paper I we identified five subgroups based on a data-driven clustering of depression and anxiety symptoms, which differed in terms of unique symptom profiles, which were supported by resting-state fMRI brain functional connectivity differences. We did not find any significant association between brain functional connectivity patterns and depression case-control status, nor sum scores of depression or anxiety symptoms, illustrating the heterogeneity of depression. In paper II, using a machine learning approach, we were not able to predict mental health traits relevant for depression, nor polygenic scores using several conceptualizations of brain functional connectivity in the UK Biobank. In contrast, we were able to predict age and sex with high accuracy, and robustly predict years of educational attainment and fluid intelligence. In paper III, we found no significant association with conventional categorical and dimensional measures of depression with brain components based on a multimodal fusing approach. Overall, these the thesis and these three papers shows the need for precise stratification of depression at the individual level is required to disentangle its heterogeneous nature