Classification tree model analysis of influencing factors for hepatocyte steatosis in patients with chronic hepatitis B / 临床肝胆病杂志
Journal of Clinical Hepatology
;
(12): 476-479, 2016.
Artigo
em Chinês
| WPRIM
| ID: wpr-778567
ABSTRACT
ObjectiveTo investigate the influencing factors for hepatocyte steatosis in patients with chronic hepatitis B (CHB) and the high-risk population by classification tree model analysis, and to establish a simple method to assess the risk of hepatocyte steatosis in CHB patients. MethodsThe clinical data and pathological results of the CHB patients who underwent liver biopsy in Department of Infectious Diseases, The First People's Hospital of Shunde, from January 2006 and December 2014 were collected. The classification tree model was applied to analyze the influencing factors for hepatocyte steatosis, and index value curve, misclassification matrix, and error of estimation were applied for overall evaluation of classification results of the classification tree model. ResultsThe influencing factors for hepatocyte steatosis in CHB patients were body mass index (BMI), total cholesterol, and low-density lipoprotein, and the most important factor was BMI. This classification tree model had a sensitivity of 84.3%, a specificity of 81.5%, an accuracy of 82.9%, and an error of estimation of 0.171, suggesting that this model was well fitted. ConclusionClassification tree model analysis shows that the pathogenesis of hepatocyte steatosis in CHB patients is closely related to the influencing factors BMI, total cholesterol, and low-density lipoprotein. A simple classification method is established based on these factors to evaluate the risk of hepatocyte steatosis in CHB patients. It is necessary to conduct further clinical studies to investigate the clinical value of this method.
Texto completo:
DisponíveL
Índice:
WPRIM (Pacífico Ocidental)
Tipo de estudo:
Estudo prognóstico
/
Fatores de risco
Idioma:
Chinês
Revista:
Journal of Clinical Hepatology
Ano de publicação:
2016
Tipo de documento:
Artigo
Similares
MEDLINE
...
LILACS
LIS