Construction of a diagnostic model for rotator cuff injury based on serum inflammatory markers and routine biochemical parameters

Abstract Objective Rotator cuff injury is a common cause of shoulder pain and its diagnosis relies on MRI, which is costly. We aimed to develop a diagnostic prediction model using routine serum markers and clinical features. Methods This retrospective case-control study includes 150 patients with MRI-confirmed rotator cuff injury and 150 healthy controls. Clinical data and routine serological markers are collected. Participants are randomly divided into training (n = 212) and validation (n = 88) sets. Variable selection is performed using univariate logistic regression, LASSO regression with 10-fold cross-validation and the Boruta algorithm. Independent predictors identified by multivariable logistic regression are used to construct a nomogram. Model performance is evaluated by ROC curves, calibration plots, Hosmer-Lemeshow test and decision curve analysis, with internal bootstrap validation. Results Six variables—gender, age, total cholesterol, globulin, gamma-glutamyl transferase and lymphocyte-to-monocyte ratio—are independent predictors. The nomogram achieves an AUC of 0.880 (bootstrap-corrected 0.879) in the training cohort and 0.921 in the validation cohort. Calibration is satisfactory (Hosmer-Lemeshow P = 0.941 and 0.147, respectively). Decision curve analysis confirms clinical utility. Conclusion The nomogram shows robust discrimination, calibration, and clinical net benefit, providing a promising non-invasive tool for early screening of rotator cuff injury.