Identification and validation of a nomogram for psoriasis diagnosis: a novel biomarker combination with stable high...

Psoriasis is a complex chronic inflammatory skin disease. While typical cases are often diagnosed based on clinical features, achieving objective assessment of disease activity and precise early-stage identification remains a clinical focus. This study aims to evaluate the diagnostic value of monocytes (MONO), Orosomucoid-1 (ORM1), Orosomucoid-2 (ORM2), and Alpha-1-acid glycoprotein (AGP), and to develop a nomogram for the objective and early identification of psoriasis.

This retrospective case-control study included 140 participants, comprising 70 patients with psoriasis confirmed by the dermatology department and 70 healthy individuals who underwent routine health examinations during the same period. Demographic, clinical, and laboratory data (including MONO, ORM1, ORM2, and AGP) were collected for all participants. Potential risk factors were initially screened using univariate Logistic regression, followed by feature selection via the combination of Lasso regression and Boruta algorithm to identify features with the highest predictive value. Restricted cubic spline (RCS) plots were utilized to visually illustrate the non-linear associations between the selected variables and the risk of psoriasis onset. Finally, the selected features were incorporated into a multivariate Logistic regression model to determine the independent risk factors, and a diagnostic nomogram was constructed accordingly.

Based on the results of univariate analysis, LASSO regression and Boruta algorithm, we ultimately selected four key variables, namely MONO, ORM1, ORM2 and AGP, for the construction of the subsequent multivariate diagnostic model. The calibration curve of the model showed that the actual probability was highly consistent with the predicted probability, indicating that the model had good calibration performance. The receiver operating characteristic (ROC) curve indicated that the overall predictive ability of the model was excellent, with an area under curve (AUC) of 0.888 (95% CI: 0.835-0.941). In addition, the ROC curves of each variable analyzed separately showed that ORM2 (AUC = 0.777), ORM1 (AUC = 0.720), MONO (AUC = 0.638) and AGP (AUC = 0.673) all had different degrees of predictive ability for the risk of psoriasis.

The novel biomarker combinations with stable and high expression in the early stage have shown great potential in the research of psoriasis. These research results should be applied to clinical.