Translational gastroenterology and hepatology · 2026
FibroX: a machine learning model for detecting and prognosticating advanced fibrosis in metabolic dysfunction-associated steatotic liver disease.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Njei Basile, Al-Ajlouni Yazan A, Ilagan-Ying Ysabel, Al Ta'ani Omar, Boateng Sarpong, Njei Nelvis, Kanmounye Ulrick Sidney, Lim Joseph, Banini Bubu
Аннотация
BACKGROUND: Noninvasive liver disease assessment (NILDA) tools, such as the fibrosis-4 (FIB-4) index, are widely used to identify patients with metabolic dysfunction-associated steatotic liver disease (MASLD) at risk of advanced fibrosis (≥ F3). However, their predictive accuracy for fibrosis staging and long-term outcomes remains limited. This study aimed to develop and validate FibroX, an explainable machine learning-based model for improving the detection of advanced fibrosis and the prediction of long-term clinical outcomes in adults with MASLD.
METHODS: FibroX was developed using data from adults with MASLD (N=1,487) in the National Health and Nutrition Examination Survey (NHANES) 2017-2020. Fibrosis stage was determined using a two-step approach-initial NILDA screening followed by vibration-controlled transient elastography (TE)-in accordance with 2024 American Association for the Study of Liver Diseases guidance. The model was built using extreme gradient boosting (XGBoost), optimized at a 95% specificity threshold, and internally validated using 5-fold cross-validation. External validation was performed in two independent cohorts: 337 biopsy-confirmed MASLD patients and 4,276 participants from NHANES III with up to 30 years of mortality follow-up. Model interpretability was evaluated using SHapley Additive Explanations (SHAP).
RESULTS: In NHANES 2017-2020, FibroX demonstrated superior discrimination for advanced fibrosis compared with FIB-4 [area under the receiver operating characteristic curve (AUROC) 0.97 vs. 0.62; P<0.001]. In biopsy-confirmed MASLD patients, FibroX also outperformed FIB-4 (AUROC 0.84 vs. 0.82; P<0.001). In NHANES III, FibroX predicted all-cause mortality (C-statistic 0.80) and remained independently associated with cardiovascular mortality after adjustment for established risk factors [adjusted hazard ratio (HR) 1.22; 95% confidence interval (CI): 1.01-1.47]. SHAP analysis identified platelet count, age, hemoglobin A1c, aspartate aminotransferase (AST), and estimated glomerular filtration rate (eGFR) as the most influential predictors.
CONCLUSIONS: FibroX is a transparent and accurate machine learning model that improves detection of advanced fibrosis and prediction of long-term cardiovascular mortality compared with FIB-4. These findings support the use of explainable machine learning-based noninvasive tools for risk stratification and outcome prediction in patients with MASLD.
METHODS: FibroX was developed using data from adults with MASLD (N=1,487) in the National Health and Nutrition Examination Survey (NHANES) 2017-2020. Fibrosis stage was determined using a two-step approach-initial NILDA screening followed by vibration-controlled transient elastography (TE)-in accordance with 2024 American Association for the Study of Liver Diseases guidance. The model was built using extreme gradient boosting (XGBoost), optimized at a 95% specificity threshold, and internally validated using 5-fold cross-validation. External validation was performed in two independent cohorts: 337 biopsy-confirmed MASLD patients and 4,276 participants from NHANES III with up to 30 years of mortality follow-up. Model interpretability was evaluated using SHapley Additive Explanations (SHAP).
RESULTS: In NHANES 2017-2020, FibroX demonstrated superior discrimination for advanced fibrosis compared with FIB-4 [area under the receiver operating characteristic curve (AUROC) 0.97 vs. 0.62; P<0.001]. In biopsy-confirmed MASLD patients, FibroX also outperformed FIB-4 (AUROC 0.84 vs. 0.82; P<0.001). In NHANES III, FibroX predicted all-cause mortality (C-statistic 0.80) and remained independently associated with cardiovascular mortality after adjustment for established risk factors [adjusted hazard ratio (HR) 1.22; 95% confidence interval (CI): 1.01-1.47]. SHAP analysis identified platelet count, age, hemoglobin A1c, aspartate aminotransferase (AST), and estimated glomerular filtration rate (eGFR) as the most influential predictors.
CONCLUSIONS: FibroX is a transparent and accurate machine learning model that improves detection of advanced fibrosis and prediction of long-term cardiovascular mortality compared with FIB-4. These findings support the use of explainable machine learning-based noninvasive tools for risk stratification and outcome prediction in patients with MASLD.