A temporal adaptive dictionary-constrained LDA and Bi-calibrated dual granularity DTM framework for dynamic topic evolution analysis in academic papers.

Major depressive disorder (MDD) is a prevalent and recurring mental illness, affecting one in five adults during their lifetime and representing a leading contributor to disability worldwide. Although mood symptoms are central to the diagnosis, depressive disorder is also characterized by a wide array of cognitive symptoms, including difficulties in attention, memory, processing speed, and executive function. These cognitive deficits are not merely a secondary consequence of the depressive mood, they are a core component of the illness. Indeed, they play a key role in the psychosocial impairment experienced by people with MDD, hindering their performance at work and in education, and disrupting their social interactions. Consequently, cognitive symptoms that remain untreated are associated with a lower quality of life and an increased risk of recurrent depressive episodes. Although the precise biological origins of MDD remain complex, the impaired neuroplasticity hypothesis provides a compelling explanatory framework. Neuroplasticity ‒the neural capacity to reorganize structurally and functionally in response to stimuli‒ is fundamental for adaptation. In MDD, this process becomes maladaptive, characterized by compromised synaptic strength, dendritic atrophy, deficiencies in neurotrophic factors, glial dysfunction, and altered functional connectivity. Current first-line treatments, including antidepressants (selective serotonin reuptake inhibitors - SSRI), cognitive behavioral therapy (CBT), and electroconvulsive therapy (ECT), are believed to exert their clinical benefits by targeting and reversing these plasticity deficits in vulnerable brain regions, such as the dorsolateral prefrontal cortex (dlPFC). However, monitoring the molecular changes directly in the living human brain remains a significant challenge, underscoring the need for accessible peripheral biomarkers. MicroRNAs (miRNAs), a class of small non-coding RNAs, have emerged as powerful candidates for this purpose due to their unique ability to control gene regulatory networks. By interacting with messenger RNA (mRNA) transcripts, miRNAs orchestrate broad regulatory outcomes; a single miRNA can target hundreds of mRNA transcripts, while conversely, multiple miRNAs can converge to regulate a single mRNA transcript, acting as a true master switch for biological processes. Indeed, there is compelling evidence linking changes in the regulation of miRNAs and their mRNA targets to the onset of MDD and antidepressant efficacy. Notably, miRNAs can cross the blood-brain barrier and remain stable in the circulation, making them valuable candidates for liquid biopsies that could reflect pathophysiological states of the central nervous system (CNS). Evidence shows comparable dysregulation of miRNAs in the blood and brain tissue of patients with neuropsychiatric diseases, including MDD.