ExplainabilityAccountabilitySafetystat.ML
MAFUS: a Framework to predict mortality risk in MAFLD subjects
Domenico Lofù, Paolo Sorino, Tommaso Colafiglio, Caterina Bonfiglio, Fedelucio Narducci, Tommaso Di Noia, Eugenio Di Sciascio
January 17, 2023
arXiv.org
1 citations
Key Takeaways
-
AI can help doctors predict which patients with metabolic fatty liver disease are at highest risk of death. The MAFUS framework uses machine learning to analyze patient data like body measurements and blood tests to identify mortality risk, addressing a gap where few tools currently exist for this purpose in MAFLD patients.
-
The framework prioritizes interpretability so doctors can understand why predictions are made. The researchers performed Explainable AI (XAI) analysis to show which patient factors contribute most to mortality predictions, rather than treating the AI as a "black box"—an important safeguard for responsible medical AI deployment.
-
The required data is practical and already available in clinical settings. MAFUS uses readily accessible anthropometric and biochemical measurements, making it feasible for physicians to implement without requiring new or specialized data collection procedures.
-
Multiple algorithms were tested to identify the most reliable approach. The team trained five different machine learning algorithms and found Support Vector Machines performed best, suggesting they validated their approach rather than relying on a single untested method.
Abstract
Metabolic (dysfunction) associated fatty liver disease (MAFLD) establishes new criteria for diagnosing fatty liver disease independent of alcohol consumption and concurrent viral hepatitis infection. However, the long-term outcome of MAFLD subjects is sparse. Few articles are focused on mortality in MAFLD subjects, and none investigate how to predict a fatal outcome. In this paper, we propose an artificial intelligence-based framework named MAFUS that physicians can use for predicting mortality in MAFLD subjects. The framework uses data from various anthropometric and biochemical sources based on Machine Learning (ML) algorithms. The framework has been tested on a state-of-the-art dataset on which five ML algorithms are trained. Support Vector Machines resulted in being the best model. Furthermore, an Explainable Artificial Intelligence (XAI) analysis has been performed to understand the SVM diagnostic reasoning and the contribution of each feature to the prediction. The MAFUS framework is easy to apply, and the required parameters are readily available in the dataset.