Transition and dynamic reconfiguration in late-life depression based on hidden Markov model.

Late-life depression is characterized by persistent emotional distress and cognitive dysfunction, yet understanding the specific brain dynamics and molecular mechanisms involved remains limited. Here, we employed a hidden Markov model to analyze resting-state functional magnetic resonance imaging data from 154 patients with late-life depression and 147 healthy controls. This analysis revealed 12 recurring brain states with distinct spatiotemporal patterns and identified atypical dynamic features across several networks. Notably, patients exhibited significantly higher transition probabilities for entering, exiting, and maintaining in the positive activation state of the default mode network, with genes linked to this state mainly enriched in regulation of neuronal synaptic plasticity and cognitive processes. Hierarchical clustering further found a critical entry and exit point between two high-level meta-states with opposing activation patterns, highlighting large-scale network dysfunction and potential molecular mechanisms associated with late-life depression through the decoding of brain states.