Fig. 3From: Biomarkers of interstitial lung disease associated with primary Sjögren's syndromeLeast absolute shrinkage and selection operator (LASSO) modeling in the identification of pSS–ILD. Application of LASSO regression modeling that is based on machine learning revealed clinical risk factors and serum protein biomarkers capable of distinguishing pSS–ILD patients with moderate/advanced-ILD from those with no-/indeterminate-ILD. The corresponding ROC curve reflects performance characteristics of this model, as indicated by area under the curve (AUC). The accompanying table shows regression coefficients for clinical risk factors and specific serum proteinsBack to article page