[论文解读] Early Prediction of Liver Cirrhosis Up to Three Years in Advance: A Machine Learning Study Benchmarking Against the FIB-4 Score

Zhuqi Miao, Sujan Ravi+1 more|arXiv (Cornell University)|Jan 1, 2026

Artificial Intelligence in Healthcare被引用 0

一句话总结

机器学习模型使用常规电子病历数据在诊断前可预测肝硬化事件，最多提前两年（隐含最多三年），在多项指标上优于 FIB-4 和 APRI 分数。

ABSTRACT

Objective: Develop and evaluate machine learning (ML) models for predicting incident liver cirrhosis one, two, and three years prior to diagnosis using routinely collected electronic health record (EHR) data, and to benchmark their performance against the FIB-4 score. Methods: We conducted a retrospective cohort study using de-identified EHR data from a large academic health system. Patients with fatty liver disease were identified and categorized into cirrhosis and non-cirrhosis cohorts based on ICD-9/10 codes. Prediction scenarios were constructed using observation and prediction windows to emulate real-world clinical use. Demographics, diagnoses, laboratory results, vital signs, and comorbidity indices were aggregated from the observation window. XGBoost models were trained for 1-, 2-, and 3-year prediction horizons and evaluated on held-out test sets. Model performance was compared with FIB-4 using area under the receiver operating characteristic curve (AUC). Results: Final cohorts included 3,043 patients for the 1-year prediction, 1,981 for the 2-year prediction, and 1,470 for the 3-year prediction. Across all prediction windows, ML models consistently outperformed FIB-4. The XGBoost models achieved AUCs of 0.81, 0.73, and 0.69 for 1-, 2-, and 3-year predictions, respectively, compared with 0.71, 0.63, and 0.57 for FIB-4. Performance gains persisted with longer prediction horizons, indicating improved early risk discrimination. Conclusions: Machine learning models leveraging routine EHR data substantially outperform the traditional FIB-4 score for early prediction of liver cirrhosis. These models enable earlier and more accurate risk stratification and can be integrated into clinical workflows as automated decision-support tools to support proactive cirrhosis prevention and management.

研究动机与目标

• Motivate early prediction of liver cirrhosis (LC) using routinely collected EHR data.
• Develop and evaluate ML models for 1- and 2-year prediction horizons prior to LC diagnosis.
• Benchmark ML models against FIB-4 and APRI scores.
• Demonstrate improved discrimination and clinical utility for early risk stratification.

提出的方法

• Retrospective cohort study using de-identified EHR data from a large academic health system.
• Developed XGBoost models for 1- and 2-year prediction horizons.
• Applied model-specific feature selection and Bayesian hyperparameter tuning.
• Evaluated on held-out test sets using accuracy, precision, recall, F1, PR AUC, and AUC.
• Compared ML models to FIB-4 and APRI scores across metrics.

实验结果

研究问题

• RQ1 Can ML models predict incident LC up to 1 and 2 years before diagnosis using routine EHR data?
• RQ2 Do ML models outperform FIB-4 and APRI scores for early LC prediction across horizons?
• RQ3 How do performance metrics (AUC, PR AUC, etc.) differ between 1-year and 2-year prediction horizons?

主要发现

• XGBoost achieved AUC of 0.872 (1-year) and 0.839 (2-year).
• FIB-4 achieved AUC of 0.756 (1-year) and 0.723 (2-year).
• APRI achieved AUC of 0.798 (1-year) and 0.761 (2-year).
• PR AUC for XGBoost was 0.657 (1-year) and 0.562 (2-year).
• PR AUC for FIB-4 was 0.456 (1-year) and 0.373 (2-year).
• PR AUC for APRI was 0.504 (1-year) and 0.421 (2-year).