Prognostic models in end stage liver disease.

Area of Science:

• Hepatology
• Medical Prognostics
• Data Science in Medicine

Background:

• Cirrhosis is a leading cause of mortality globally, presenting significant healthcare burdens.
• Accurate prognostication of end-stage liver disease (ESLD) remains difficult despite advances in understanding and management.
• Existing prognostic scores, such as Child-Turcotte Pugh (CTP) and Model for End-Stage Liver Disease (MELD), are widely used but have limitations.

Purpose of the Study:

• To review the current landscape of prognostic scoring systems for end-stage liver disease.
• To highlight the challenges in predicting disease course and long-term outcomes in ESLD.
• To explore the potential role of artificial intelligence in improving prognostication for liver disease.

Main Methods:

• Review of established prognostic scores (CTP, MELD) and their modifications.
• Discussion of limitations in current prognostic accuracy, including replicability and benefit.
• Exploration of emerging approaches, including AI for patient subgroups like acute-on-chronic liver failure.

Main Results:

• Standard scores like MELD are objective and widely used for short- to medium-term prognosis and liver transplant prioritization.
• Despite modifications, MELD and other scores face challenges in accurately predicting long-term outcomes.
• Recent scores show promise for specific patient subgroups, but broader validation is needed.

Conclusions:

• Accurate long-term prognostication in ESLD remains an unmet clinical need.
• Artificial intelligence holds significant potential to enhance predictive accuracy for disease progression and long-term outcomes.
• Future research should focus on validating AI-driven models to aid hepatologists in clinical decision-making.