Depression is a heterogeneous mental illness with substantial personal and societal burdens, yet its diagnosis still relies heavily on subjective assessments. Recent advances in blood-based metabolomics have opened new avenues for identifying objective biomarkers associated with depressive symptoms. This review highlights key findings from multicenter clinical and translational research that demonstrate reproducible associations between specific plasma metabolites—such as 3-hydroxybutyrate, betaine, citrate, creatinine, and γ-aminobutyric acid (GABA)—and the severity of depressive states. Several metabolites also appear to be linked to distinct symptom domains, including suicidal ideation (SI), a critical risk factor for self-harm. Notably, combinations of citrate and kynurenine have shown potential for SI severity estimation through machine learning models, suggesting a basis for minimally invasive risk stratification. In parallel, rodent models of stress-induced depression reveal consistent alterations in tryptophan and alanine metabolism, providing insight into possible causal mechanisms involving neurotransmitter biosynthesis and intestinal absorption under stress. Personality-based biotyping and artificial intelligence further refine the stratification of depressive phenotypes, offering prospects for more personalized diagnostics. Although methodological standardization and broader validation remain necessary, accumulating evidence supports the clinical utility of blood metabolomics as a complementary tool for early detection, subtype classification, and suicide risk assessment in depression.

A research team presents XFruitSeg, a deep learning model designed to accurately segment three-dimensional computed tomography (CT) images of plant fruits.