Inflammatory and neuroendocrine biomarkers associated with mental health symptoms across body mass index categories: a cross-sectional machine learning analysis.

Introduction: Obesity is increasingly recognized as a systemic condition with neuroinflammatory and psychobiological implications, yet its integrated relationship with inflammatory and neuroendocrine pathways in mental health remains insufficiently characterized. We investigated the associations between obesity, circulating biomarkers, and symptoms of depression, anxiety, stress, and post-traumatic stress disorder (PTSD). Methods: In this cross-sectional study, 251 adults were stratified according to psychiatric diagnosis and body mass index (BMI). Inflammatory, neuroendocrine, and metabolic biomarkers were quantified, including cytokines, hs-CRP, cortisol, and vitamin D. Random Forest-based machine learning models were applied to identify the relative importance of biomarkers in predicting psychological symptom severity across clinical groups and BMI categories. Results: Vitamin D, IL-6, TNF-α, and hs-CRP consistently emerged as the most relevant predictors. Individuals with severe obesity but without a formal psychiatric diagnosis exhibited an inflammatory profile comparable to that observed in patients with mental disorders, suggesting the presence of a biomarker profile potentially associated with subclinical psychobiological alterations. In participants with mental disorders, the interaction between BMI and biomarkers was more complex and widespread, indicating a state of systemic inflammatory and neuroendocrine dysregulation. Discussion: These findings indicate that extremes of nutritional status were associated with a higher frequency of inflammatory and hormonal alterations related to mental health symptoms. The integration of biomarker profiling with machine learning approaches may support early identifying patterns of biomarker relevance associated with mental health symptoms.