A novel model predicted liver cirrhosis constructed by ultrasound and serological in autoimmune liver hepatitis

To establish a noninvasive model based on two-dimensional shear wave elasticity (2D-SWE) technology, ultrasound feature and serological indicators to predict cirrhosis in autoimmune hepatitis (AIH) and verified. Patients with AIH confirmed by liver biopsy with liver ultrasound and serological examination were collected from January 2019 to May 2022. Patients were divided into cirrhosis and non-cirrhosis groups. Basic indexes, ultrasound indexes and serological indexes were collected. Multivariable logistic regression used for screening independent risk factors predicting cirrhosis, construct the AIH cirrhosis prediction model, named autoimmune hepatitis cirrhosis (AIHC). Determine best cutoff score according to the Youden index, verified the model's predictive efficacy. One hundred forty-six patients were collected. The following indicators were independent risk factors for predicting cirrhosis: LS (OR: 1.416, P = .015), splenomegaly (OR: 10.446, P = .006), complement C4 (OR: 0.020, P = .009). The best cutoff score was 65, with a sensitivity 88.9% and specificity 75.6%; the area under curve was 0.901, AIHC possessed a higher net reclassification index (NRI) and integrated discrimination improvement compared with other indexes, and AIHC had the best clinical decision curve. The AIHC constructed in this study has better predictive efficacy than other noninvasive indexes, and we visualized the model for easy application, which was worth further promotion in clinical practice.