Machine Learning-Driven Risk Prediction Models for Posthepatectomy Liver Failure: A Narrative Review
Ioannis Margaris1, Maria Papadoliopoulou2, Periklis G Foukas3
1Eugenideio Hospital, National and Kapodistrian University of Athens, 11528 Athens, Greece.
Medicina (Kaunas, Lithuania)
|February 27, 2026
Summary
Machine learning (ML) models show promise in predicting posthepatectomy liver failure (PHLF), outperforming traditional scores. Further validation is needed, but ML tools can aid early risk detection and surgical planning for liver surgery patients.
Area of Science:
- Artificial Intelligence
- Machine Learning in Medicine
- Surgical Risk Stratification
Background:
- Posthepatectomy liver failure (PHLF) is a significant cause of morbidity and mortality after major liver resections.
- Machine learning (ML) offers advanced tools for risk stratification in surgical patient populations.
Purpose of the Study:
- To systematically review and critically analyze the literature on ML-driven risk prediction models for PHLF.
- To evaluate the performance and limitations of ML models in identifying patients at risk of PHLF.
Main Methods:
- Systematic literature search of PubMed/MEDLINE, Scopus, and Web of Science databases.
- Inclusion and analysis of fifteen studies that developed and validated ML models for PHLF prediction.
Main Results:
- ML models effectively predict PHLF using perioperative clinical, laboratory, and imaging data.
- ML algorithms demonstrate high accuracy (AUC, sensitivity), often exceeding traditional risk scores.
- Limitations include small sample sizes, heterogeneity, and lack of external validation in existing studies.
Conclusions:
- ML-driven tools show potential for early and accurate PHLF risk detection.
- Integration of ML with clinical judgment can enhance personalized surgical planning and optimize outcomes.
- Further research is needed to address limitations and improve the clinical utility of ML for PHLF prediction.
