Construction of a NAFLD identification model in Xizang population and exploration of risk factors related to liver fibrosis
- 1Plateau Biomedical Big Data Mining and Bioinformatics Analysis Laboratory, Xizang Minzu University, Xianyang, 712082 China
- 2Provincial Key Laboratory of Molecular Genetic Mechanism and Intervention of Diseases related to the Tibetan Plateau, Xizang Minzu University, Xianyang, 712082 China
Abstract
Background: The global prevalence of non-alcoholic fatty liver disease (NAFLD) exceeds 32%, but current diagnostic methods based on imaging or histology are complex and costly. Most existing NAFLD models are primarily derived from low-altitude populations, while the high-altitude populations in Xizang remain relatively understudied.
Methods: Clinical data were collected through questionnaires, examinations, and laboratory tests from populations residing at various altitudes in Xizang. Patients with NAFLD and participants without hepatic steatosis were identified based on the CAP value obtained using the M probe of FibroScan and specific exclusion criteria. Differences in NAFLD prevalence across altitude, age, gender, and ethnicity were analyzed using Chi-square tests. Initially, 31 clinical indicators were screened through univariate logistic regression; variables demonstrating statistical significance (P < 0.05) were subsequently included in a multivariate logistic regression model. The model’s performance was evaluated using the area under the receiver operating characteristic (ROC) curve, visualized with a nomogram, and subgroup analyses were performed by altitude.Moreover, machine learning methods—including Generalized ar Models (GLM), Gradient Boosting Models (GBM), Extreme Gradient Boosting (XGBoost), and Random Forest (RF) identified key liver fibrosis variables, and piecewise regression analyzed biomarker risk thresholds.
Results: The prevalence of NAFLD in the Xizang exhibits significant disparities based on sex, age, and altitude. A multivariable analysis identified 15 independent predictors, with the full model demonstrating an area under the curve (AUC) of 0.898 (95% CI = 0.884–0.911), a sensitivity of 86.3%, a specificity of 77.7%, and an accuracy of 81.5%, resulting in a Youden index of 0.64. Additionally, a simplified model that included six liver function parameters, age, and body mass index (BMI) also showed strong discriminatory capability, with an AUC of 0.884 (95% CI = 0.870–0.898), a sensitivity of 84.8%, a specificity of 75.8%, and an accuracy of 79.8%, yielding a Youden index of 0.61. Among individuals aged 65 years or younger, the risk of liver fibrosis was relatively low when their permanent residence altitude ranged from 1,400 to 3,832 m. In contrast, for those older than 65 years, a lower risk of liver fibrosis was observed at permanent residence altitudes between 1,768 and 2,583 m.
Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05193-7.
Related
- Factors associated with significant fibrosis and MASH in a diabetic Caribbean population of the French West Indies
- Machine learning-based analysis of risk factors and construction of a predictive model for hyperuricemia in Chinese health examination population
- Identification of palliative care needs and prognostic factors of survival for tailoring appropriate interventions in advanced oncological, renal, and pulmonary diseases: results from a prospective observational study in an Italian population.
- Automated quantitative analysis can support the identification of progressive pulmonary fibrosis in fibrotic interstitial lung disease.
- An exploration of organisational conditions and psychosocial work factors of long-term care workers in self-managing teams
BibTeX
@article{jiang2026construction,
title = {{Construction of a NAFLD identification model in Xizang population and exploration of risk factors related to liver fibrosis}},
author = {Zhao Jiang and Kezhen Han and Manfei Jin and Zhiying Zhang and Lijun Liu and Longli Kang},
journal = {BMC Gastroenterology},
year = {2026},
month = aug,
doi = {10.1186/s12876-026-05193-7},
url = {https://doi.org/10.1186/s12876-026-05193-7},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Construction of a NAFLD identification model in Xizang population and exploration of risk factors related to liver fibrosis
AU - Zhao Jiang
AU - Kezhen Han
AU - Manfei Jin
AU - Zhiying Zhang
AU - Lijun Liu
AU - Longli Kang
PY - 2026
DA - 2026/08/07
JO - BMC Gastroenterology
DO - 10.1186/s12876-026-05193-7
UR - https://doi.org/10.1186/s12876-026-05193-7
AB - Background: The global prevalence of non-alcoholic fatty liver disease (NAFLD) exceeds 32%, but current diagnostic methods based on imaging or histology are complex and costly. Most existing NAFLD models are primarily derived from low-altitude populations, while the high-altitude populations in Xizang remain relatively understudied. Methods: Clinical data were collected through questionnaires, examinations, and laboratory tests from populations residing at various altitudes in Xizang. Patients with NAFLD and participants without hepatic steatosis were identified based on the CAP value obtained using the M probe of FibroScan and specific exclusion criteria. Differences in NAFLD prevalence across altitude, age, gender, and ethnicity were analyzed using Chi-square tests. Initially, 31 clinical indicators were screened through univariate logistic regression; variables demonstrating statistical significance (P < 0.05) were subsequently included in a multivariate logistic regression model. The model’s performance was evaluated using the area under the receiver operating characteristic (ROC) curve, visualized with a nomogram, and subgroup analyses were performed by altitude.Moreover, machine learning methods—including Generalized ar Models (GLM), Gradient Boosting Models (GBM), Extreme Gradient Boosting (XGBoost), and Random Forest (RF) identified key liver fibrosis variables, and piecewise regression analyzed biomarker risk thresholds. Results: The prevalence of NAFLD in the Xizang exhibits significant disparities based on sex, age, and altitude. A multivariable analysis identified 15 independent predictors, with the full model demonstrating an area under the curve (AUC) of 0.898 (95% CI = 0.884–0.911), a sensitivity of 86.3%, a specificity of 77.7%, and an accuracy of 81.5%, resulting in a Youden index of 0.64. Additionally, a simplified model that included six liver function parameters, age, and body mass index (BMI) also showed strong discriminatory capability, with an AUC of 0.884 (95% CI = 0.870–0.898), a sensitivity of 84.8%, a specificity of 75.8%, and an accuracy of 79.8%, yielding a Youden index of 0.61. Among individuals aged 65 years or younger, the risk of liver fibrosis was relatively low when their permanent residence altitude ranged from 1,400 to 3,832 m. In contrast, for those older than 65 years, a lower risk of liver fibrosis was observed at permanent residence altitudes between 1,768 and 2,583 m. Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05193-7.
SN - 1471-230X
ER -