Development and Validation of the Nurse Unit Manager Competency Scale (NUM-CS): A Multicenter Study.

Major depressive disorder (MDD) is a highly prevalent psychiatric condition that affects more than 332 million individuals worldwide in 2021 and imposes a substantial burden on both patients and society [, ]. The first-line treatment for MDD is typically antidepressant medication, particularly selective serotonin reuptake inhibitors (SSRIs) [, ]. However, the efficacy of pharmacotherapy for MDD varies between individuals [], with 40–50% of patients not responding to treatment in clinical practice []. The study demonstrated that early symptomatic improvement during antidepressant treatment was predictive of later clinical remission. Conversely, MDD patients who showed no response after 2 weeks of treatment achieved a final remission rate of only 4% when the treatment regimen remained unchanged []. Retrospective analyses of multiple clinical trials indicated that a lack of response after 2 weeks of antidepressant treatment reliably predicted unfavorable outcomes at 6–8 weeks []. A study demonstrated that lack of response to SSRI therapy at 2 weeks is a robust predictor of poor clinical treatment outcome at the end of the standard treatment course []. Therefore, elucidating the neuropathological mechanisms underlying MDD and exploring potential neuroimaging correlates of early treatment response could provide valuable insights for optimizing individualized therapeutic strategies in clinical practice. Several resting-state functional MRI (rs-fMRI) studies have reported functional differences between patients with MDD and healthy controls. MDD patients have lower node strength in the right hippocampal CA3/4 subregion and a reduced clustering coefficient in the right dentate gyrus, while the clustering coefficient in the right amygdala’s central nucleus correlates positively with depression severity []. Compared to healthy controls, MDD also show reduced functional connectivity between the left amygdala and both the ventromedial prefrontal cortex (vmPFC) and the right anterior insula. In contrast, the right amygdala has stronger connectivity with the bilateral hippocampus and left middle temporal gyrus []. Individual differences in functional indices are often linked to treatment outcomes. High neuroticism is linked to increased connectivity within the salience, executive control, and somatomotor networks, which predicts poorer treatment outcomes. After 8 weeks of treatment, this connectivity was reduced []. Longitudinal studies on first-episode adolescent MDD patients have shown that the amygdala’s connectivity with the superior temporal gyrus and cingulate gyrus increases before treatment. After eight weeks of SSRI treatment, these connections strengthen further, alongside symptom improvement []. The functional connectivity between the hippocampus and dorsolateral prefrontal cortex (DLPFC) is negatively correlated with cognitive symptoms, such as inattention and memory issues, but improves with treatment []. Several studies suggest that neuroimaging markers, such as cortical thickness in the anterior cingulate cortex (ACC) or early changes in functional connectivity in the subgenual ACC, may predict relapse or remission []. For instance, in an escitalopram-treated cohort study, higher subgenual anterior cingulate cortex (sgACC)–frontoparietal connectivity in baseline distinguished remitters from non-remitters with 72.64% accuracy []. However, the results across studies are still inconsistent, highlighting the need for further work to confirm which brain activity patterns predict antidepressant treatment outcomes. Recently, machine learning has shown significant potential to assist in predicting individualized treatment responses in MDD by identifying latent multivariate patterns in neuroimaging data. Rather than relying on traditional univariate comparisons, researchers are increasingly applying advanced algorithms (Gaussian Naive Bayes, AdaBoost, Support Vector Machines, Random Forest, ensemble learning, and deep learning) to high-dimensional structural and resting-state functional MRI metrics [, ]. For example, some pioneering studies have utilized regional volumetric features or intrinsic functional connectivity within specific large-scale brain networks to construct predictive classifiers for symptom improvement and clinical remission [, ]. Furthermore, recent multi-center neuroimaging consortiums and systematic reviews have further confirmed that integrating multi-metric fMRI data with machine learning frameworks can capture subtle neurobiological heterogeneity, thereby achieving preliminary stratification of antidepressant efficacy prior to treatment initiation []. Although these studies robustly demonstrate that neuroimaging features contain valuable predictive signals, translating these models into clinical practice still faces major methodological challenges.