A psychometric examination of the career construction model of adaptation in working adults and students.

Mood disorders, including major depressive disorder (MDD) and bipolar disorder (BD), represent a significant and growing global health burden. MDD carries a lifetime prevalence of 5 to 17 percent, with an average of around 12 percent (), while the aggregate lifetime prevalence of BD is estimated at 2.4 percent (). Despite their prevalence, objective and reliable tools for distinguishing between these conditions and from healthy controls remain limited in clinical practice. Diagnosis continues to rely predominantly on clinical interview and symptom-based criteria, which are susceptible to subjectivity and overlap between conditions. This diagnostic ambiguity contributes to delays in treatment, misclassification, and suboptimal patient outcomes. Accurate differentiation between MDD and BD is of substantial clinical importance, as the two conditions require markedly different pharmacological management; misclassification, particularly the under-recognition of bipolar depression as unipolar MDD, can lead to inappropriate antidepressant monotherapy and is associated with poorer long-term outcomes, including increased risk of mood destabilisation (, ). The integration of diverse medical data modalities into machine learning (ML) predictive models has emerged as a promising avenue for improving diagnostic precision in psychiatry, as well as understanding wider differences between MDD, BD, and healthy individuals (–). In the past decade, digitisation of health data has grown substantially across healthcare sectors, encompassing electronic health records (EHR), medical imaging, multi-omics data, and environmental variables (, ). Integrating these data sources has become increasingly important for enhancing prediction, diagnosis, and treatment planning, as each modality captures aspects of a patient’s health not fully represented by any single source. In Alzheimer’s disease diagnosis, combining imaging data with cognitive assessments and demographic variables has consistently outperformed single-source models (). Similar advantages have been reported in diabetic retinopathy prediction () and acute respiratory failure detection from chest X-rays and EHR data (), indicating the broader applicability of multimodal fusion across medical imaging tasks. Structural brain MRI provides objective measures of brain morphology that differ across psychiatric conditions, while polygenic risk scores (PRS) derived from genome-wide association studies (GWAS) capture inherited genetic liability for specific disorders (). These two modalities offer complementary information: neuroimaging reflects structural differences between MDD, BD, and healthy control groups, including some which may reflect consequences of illness, while genetic risk scores reflect predisposition independent of disease stage or treatment history. The combination of these modalities holds particular promise for mood disorder classification, yet their systematic integration remains underexplored in the psychiatric ML literature.