Construction Of A Long-term Prediction Model For Connective Tissue Disease-associated Interstitial Lung Disease Based On LASSO-Cox Regression

Background and objectiveAccurate risk assessment is important for the therapeutic intervention of patients with connective tissue disease-associated interstitial lung disease(CTD-ILD).In the current research,we establish a nomogram to predict 3-and 5-year mortality based on machine learning approach and validate the ILD-gender-age-physiology(ILD-GAP)model in patients with CTD-ILD.MethodsCTD-ILD patients who were diagnosed and treated at the First Affiliated Hospital of Zhengzhou University were enrolled between February 2011 and July 2018.Cox regression with the least absolute shrinkage and selection operator(LASSO)was applied to screen out the predictors and establish a nomogram.Internal validation was performed by using bootstrap resampling.Then,the nomogram and ILD-GAP model were evaluated by decision curve analysis,the net reclassification improvement(NRI),integrated discrimination improvement(IDI),Harrell’s C index and likelihood ratio testing.ResultsA total of 675 consecutive CTD-ILD patients were enrolled in this study,during the median follow-up period of 50(quartile,38-65)months,158 patients died(mortality rate 23.4%).After feature selection,9 variables were identified:age,albumin,aspartate transaminase,rheumatoid arthritis,right atrial area,right ventricular diameter,immunosuppressive agents,the percent predicted values of diffusion capacity of lung for carbon monoxide(DLco%Pred)and honeycombing.By integrating these variables,a predictive nomogram(3ARIDH model)was generated,which provided superior mortality estimates than ILD-GAP model by the calibration plots,Harrell’s C index(0.767 and 0.652 respectively)and likelihood ratio testing.The 3ARIDH model also improved the ability of discriminating 5-year(0.136 and 0.325,IDI and NRI respectively)and 3-year(0.137 and 0.294,IDIand NRI respectively)mortality rates compared to ILD-GAP model.In addition,the 3ARIDH model was more clinically useful based on decision curve analysis.Conclusions1.Albumin,aspartate transaminase,immunosuppressive agents,honeycombing,right atrial area,right ventricular diameter,the percent predicted values of diffusion capacity of lung for carbon monoxide,rheumatoid arthritis and age correlated with mortality of CTD-ILD patients.2.The 3ARIDH model performed well in predicting 3-and 5-year mortality risk in patients with CTD-ILD than ILD-GAP model.