Machine Learning Predicts Mortality Risk in MAFLD
In a 2025 Sci Rep analysis of 4,415 NHANES participants with metabolic dysfunction-associated fatty liver disease (MAFLD), researchers trained five survival models to predict all-cause and circulatory system disease (CSD) mortality. Gradient Boosted Survival performed best for all-cause mortality and Extra Survival Trees for CSD; SHAP identified top predictors including age, gender, platelet count, HDL, smoking, BUN, and systolic blood pressure. Integrating models with SHAP enables transparent individual risk stratification.
Key Points
- 1Identified best models: Gradient Boosted Survival and Extra Survival Trees outperform others for mortality prediction.
- 2Demonstrated high predictive performance enables accurate stratification of all-cause and circulatory mortality risk.
- 3Provided SHAP-based feature importance highlighting age, BUN, SBP, platelet count, HDL, guiding clinical risk interpretation.
Scoring Rationale
Strong evidence from peer-reviewed NHANES analysis supports practical survival models, but novelty is incremental within clinical ML applications.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
