Developing an individual depression risk score based on traditional risk factors and routine biochemical markers

Early identification of individuals at high risk for depression is essential for effective implementation of interventions. This study utilized the UK Biobank database to construct an individual depression risk score using nomogram and explored the potential of traditional risk factors and routine biochemical markers for the auxiliary diagnosis of individual depression.

A total of 369,407 participants were included in the study and divided into training and testing sets. LASSO regression was employed to select predictive variables for depression from 16 traditional risk factors and 28 routine biochemical markers. Following variable selection, two multivariable logistic regression models were constructed. Nomograms were then generated to visualize the relationships between these variables and depression risk, and to facilitate the calculation of individual depression risk scores.

Twelve traditional risk factors and nine biochemical markers were selected for model building. Model 1, using only traditional risk factors, achieved the area under the curve (AUC) of 0.913 (95 % CI: 0.910-0.915), while Model 2, incorporating both traditional and routine biochemical markers, yielded an AUC of 0.914 (95 % CI: 0.912-0.917). Based on optimal cut-off values, Model 1 exhibited a sensitivity of 81.99 % and a specificity of 83.76 %, while Model 2 demonstrated a sensitivity of 81.54 % and a specificity of 84.31 %.

External validation is still needed to confirm the model's generalizability.

While the depression risk scoring model built using traditional risk factors effectively identifies high-risk individuals for depression and demonstrates good clinical performance, incorporating routine biochemical markers did not significantly improve the model's performance.