Building of Prognosis Model for Patients With Cirrhosis Based on Sarcopenia Assessed by Deep Learning

The goal of this observational study is to collect clinical and abdominal imaging data of patients with liver cirrhosis. The collected imaging data will be used as a model development set to develop, test, and internally validate a fully automated imaging deep learning platform for the evaluation of sarcopenia in liver cirrhosis. Subsequently, relevant data from patients with liver cirrhosis at other centers will be collected and used as an external validation dataset. The model will be externally validated by abdominal radiology experts. Furthermore, we will include sociodemographic information, clinical data, imaging data, and clinical outcomes of the aforementioned liver cirrhosis patients to predict the prognosis of these patients using the established model. This model will be used to construct a new prognostic model for liver cirrhosis incorporating imaging biomarkers such as sarcopenia, and its predictive performance will be validated.

Inclusion criteria

• Age ≥18 years
• Diagnosis of liver cirrhosis, meeting at least one of the following criteria:
• Clinical diagnosis: ICD-10-CM codes K74.100 and K74.607 from our hospital's electronic medical record system
• Liver biopsy pathology or a combination of clinical, laboratory, and imaging examinations confirming liver cirrhosis: Pathological biopsy criteria: fibrosis bridging between lobules leading to lobular structural disarray, nodular regeneration of hepatocytes, formation of pseudo-lobules
• Laboratory tests: the presence of at least 2 of the following 4 abnormal indicators suggesting liver cirrhosis:
• a) Platelet count < 100×10^9/L, with no other explainable cause;
• b) Serum albumin < 35g/L, excluding malnutrition or kidney disease as other causes;
• c) International normalized ratio (INR) > 1.3 or prolonged prothrombin time (PT) (after discontinuation of thrombolytic or anticoagulant drugs for more than 7 days);
• d)Aspartate aminotransferase to platelet ratio index (APRI) > 2.
• Availability of high-quality L3-level CT images

Exclusion criteria

• Incomplete sociodemographic, laboratory, or imaging data
• Diagnosed or highly suspected malignancy
• Severe chronic kidney disease, respiratory insufficiency, cardiovascular diseases, etc.
• Neurological diseases and muscular degenerative diseases
• Hyperthyroidism, hypothyroidism, tuberculosis, or any other diseases that may affect basal metabolism
• Diseases or conditions causing malabsorption of intestinal nutrients, such as inflammatory bowel disease or gastrointestinal surgery
• Treatment with glucocorticoids or immunosuppressants
• Pregnancy or lactation