Abstract: Objective To develop and validate prognostic models for tuberculosis patients who had the fixed dose compound preparation with drug-induced liver injury.Methods The clinical data of 208 patients with drug-induced liver injury were collected to develop prognostic models．The patients were randomly divided into the modeling sampling group (n=156)and the validation sampling group (n=56). Modeling sampling were used for univariate and Logistic regression analysis to identify factors related with drug-induced liver injury. According to univariate and Logistic regression analysis results, a prognostic model was established. The validation sampling was used to compute the accuracy, sensitivity and specificity of the prognostic models.Result Logistic regression analysis revealed that living environment,working environment and the diagnose result were independent risk factors for drug-induced liver injury．The prognostic model based on the risk factors of admission had favorable performance (P>0.05 for Partial chi square test,0.884 for AUC). The accuracy of the prognostic model was 84.6%, sensitivity was 55.56% and specificity was 93.02%.Conclusion The established model,which is convenient to manage,can timely and accurately predict the prognosis of tuberculosis patients with drug- induced liver injury,and can help in decision-making in clinics.

Key words: Tuberculosis, Drug-induced liver injury, Fixed-dose combination, Prognostic models, Validation