The role of genetic polymorphism in predictive model of ATDH

BACKGROUND: To develop a predictive model for hepatotoxicity due to antituberculosis drugs using a machine learning approach combining general clinical features of the electronic medical record, laboratory indications and genetic features of key genes in the PXR/ALAS1/FOXO1 axis. METHODS: Using the occurrence of ATDH as the outcome variable, the data were screened for features and model construction based on general clinical features and laboratory test indications, combined with single nucleotide polymorphism characteristics of PXR, FOXO1 and ALAS1 genes, combined with Lasso regression and logistic regression to evaluate the model's goodness of fit, predictive efficacy, discrimination and consistency, and used clinical decision Curve analysis was used to assess the clinical applicability of the models. RESULTS: The best model had a discriminant efficacy C-index of 0.8164, sensitivity of 34.25%, specificity of 97.99%, positive predictive value of 78.13%, negative predictive value of 87.69%, consistency test Sp=0.896, maximum bias Emax=0.147, and mean bias Eave=0.017. In the validation set performance was close. The clinical decision curve shows the clinical applicability of the prediction model when the prediction risk threshold is between 0.1 and 0.8. CONCLUSION: The ATDH prediction model was constructed using a machine learning approach, combining general characteristics of the study population, laboratory indications and SNP features of PXR and FOXO1 genes with good fit and some predictive value, and has potential and value for clinical application.