Construction and verification of a predictive model for depression risk of patients with somatization symptoms

Patients with somatization symptoms are at elevated risk of depression, yet underdiagnosis persists due to cultural tendencies (e.g., in China) to express psychological distress via physical complaints. Existing predictive models lack integration of sociocultural and physiological factors, particularly in non-Western populations.

To develop a culturally tailored risk-prediction model for depression in patients with somatization symptoms, emphasizing early identification and personalized intervention.

A prospective cohort study included 200 somatization patients (SSS≥38, PHQ-2<3) from a Chinese hospital (May 2020-August 2022). LASSO regression identified predictors from 18 variables, followed by multivariate logistic regression to construct a nomogram. Model performance was assessed via ROC-AUC, calibration curves, Hosmer-Lemeshow test, and decision curve analysis (DCA). Internal validation used 200 bootstrap resamples.

Five independent predictors were identified: advanced age (OR=1.11, 95% CI: 1.02-1.20), poor self-rated health (OR=2.07, 95% CI: 1.04-4.30), lack of co-residence with children (OR=1.63, 95% CI: 1.10-2.42), low income (OR=1.45, 95% CI: 1.05-2.01), and self-medication (OR=1.32, 95% CI: 1.01-1.73). The nomogram demonstrated strong discrimination (AUC=0.810, 95% CI: 0.728-0.893) and calibration (Hosmer-Lemeshow p=0.32). DCA confirmed clinical utility: at threshold probabilities >5%, the model provided higher net benefit than "treat-all" or "treat-none" strategies.

This model integrates sociocultural (e.g., family structure) and behavioral factors to predict depression risk in somatizing patients, particularly in East Asian contexts. It offers a practical tool for clinicians to prioritize high-risk individuals, reducing diagnostic delays and healthcare burdens. Future multicenter studies should validate its generalizability and incorporate biomarkers (e.g., inflammatory markers) to enhance mechanistic insights.