Enhancing Liver Cirrhosis Prognosis: A Machine Learning and Explainable AI Approach

Liver cirrhosis is a final stage of many chronic liver diseases, accompanying with higher morbidity and mortality rates reduce quality of life. The aim of this study is to examine the predictive potential of predictive analytics for survival in patients diagnosed with liver cirrhosis using a large c...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2024 Third International Conference on Artificial Intelligence, Computational Electronics and Communication System (AICECS) S. 1 - 5
Hauptverfasser: Jeyabalan, Jeyalakshmi, Karthikeyan
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 12.12.2024
Schlagworte:
Decision making
Explainable AI
Genomics
Gradient Boosting Algorithms
Liver Cirrhosis
Liver diseases
Machine Learning
Machine learning algorithms
Mortality
Precision medicine
Prediction algorithms
Predictive analytics
Prognostics and health management
Survival Prediction
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Liver cirrhosis is a final stage of many chronic liver diseases, accompanying with higher morbidity and mortality rates reduce quality of life. The aim of this study is to examine the predictive potential of predictive analytics for survival in patients diagnosed with liver cirrhosis using a large clinical dataset. The employ advanced machine learning tools, like XGBoost, LightGBM and CatBoost to predict survival rates and classify patients in different prognostic subgroups. The above data preprocessing strategy and ensemble model with soft voting and weighted XGBoost work well toward the performance. It achieves of 0.4264 Logarithmic loss and an F1 score of 0.826 over 10 cross-validation splits on a Tensorflow dataset compared to the state-of-the-art genome analysis tools, These values of SHapley Additive exPlanations (SHAP) are even more important to understand the features used in model decisions. These results reflect the promise of machine learning to enable personalized medicine and clinical decision-making in hepatology by improving upon current prognostic predictions.
DOI:10.1109/AICECS63354.2024.10956250