Prediction model for developing neuropsychiatric systemic lupus erythematosus in lupus patients.

OBJECTIVE: This study aimed to construct a predictive model for assessing the risk of development of neuropsychiatric systemic lupus erythematosus (NPSLE) among patients with SLE based on clinical, laboratory, and meteorological data.

METHODS: A total of 2232 SLE patients were included and were randomly assigned into training and validation sets. Variables such as clinical and laboratory data and local meteorological data were screened by univariate and least absolute shrinkage and selection operator (LASSO) logistic regression modelling. After 10-fold cross-validation, the predictive model was built by multivariate logistic regression, and a nomogram was constructed to visualize the risk of NPSLE. The efficacy and accuracy of the model were assessed by receiver operating characteristic (ROC) curve and calibration curve analysis. Net clinical benefit was assessed by decision curve analysis.

RESULTS: Variables that were included in the predictive model were anti-dsDNA, anti-SSA, lymphocyte count, hematocrit, erythrocyte sedimentation rate, pre-albumin, retinol binding protein, creatine kinase isoenzyme MB, Nterminal brain natriuretic peptide precursor, creatinine, indirect bilirubin, fibrinogen, hypersensitive C-reactive protein, CO, and mild contamination. The nomogram showed a broad prediction spectrum; the area under the curve (AUC) was 0.895 (0.858-0.931) for the training set and 0.849 (0.783-0.916) for the validation set.

CONCLUSION: The model exhibits good predictive performance and will confer clinical benefit in NPSLE risk calculation. Key Points • Clinical, laboratory, and meteorological data were incorporated into a predictive model for neuropsychiatric systemic lupus erythematosus (NPSLE) in SLE patients. • Anti-dsDNA, anti-SSA, LYM, HCT, ESR, hsCRP, IBIL, PA, RBP, CO, Fib, NT-proBNP, Crea, CO, and mild contamination are predictors of the development of NPSLE and may have potential for research. • The nomogram has good predictive performance and clinical value and can be used to guide clinical diagnosis and treatment.