Objective To investigate the risk factors of dermatomyositis (DM） complicated with interstitial lung disease (ILD）, and to establish and validate a prediction model.

Methods One hundred patients with DM treated from August 2022 to May 2024 were retrospectively analyzed, and all the data of the patients were collected. LASSO regression was used to screen the risk factors of DM complicated with ILD. The adjustment parameter λ in LASSO regression was verified by 10-fold cross-validation method. Variables with λ not equal to 0 were selected and included into multivariate logistic regression model to establish the prediction model for the influencing factors of DM complicated with ILD. According to the results of the prediction model, the receiver operating characteristic (ROC） curve was drawn, and the area under the curve (AUC） was calculated. Bootstrap repeated sampling (1 000 times） was used to verify the model internally, and the consistency index (C-index） was obtained.

Results A total of 100 patients with DM were included, including 33 patients with ILD and 67 patients without ILD. Univariate analysis identified 8 indicators with significant differences. LASSO regression was used to reduce dimensionality, and 5 optimal modeling indicators were selected. Subsequent logistic regression analysis confirmed 4 independent risk factors: Gottron papule, positive anti-Jo-1 antibody, positive anti-MDA5 antibody, and elevated level of serum ferritin (P ＜ 0.05 to P ＜ 0.01）. The risk prediction model was constructed as follows: Logit(P) = 2.312 + 1.235 * Gottron papule + 0.331 * anti-JO-1 antibody + 0.514 * anti-MDA5 antibody + 0.219 * serum ferritin. Model evaluation demonstrated a C-index of 0.916 and an AUC of 0.903 (95%CI: 0.837–0.942）, indicating excellent predictive performance.

Conclusions Based on the LASSO-logistic analysis, the established prediction model for risk factors of DM complicated with ILD demonstrates good predictive performance and had good clinical application value.