Enhanced Preprocessing Approach Using Ensemble Machine Learning Algorithms for Detecting Liver Disease.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Computational Biology

Background:

• Global rise in liver disease incidence and mortality.
• Challenges in early liver disease detection due to limited symptoms.
• Growing effectiveness of ensemble learning over traditional algorithms.

Purpose of the Study:

• To propose a novel ensemble learning architecture for liver disease prediction.
• To enhance prediction accuracy through advanced data preprocessing techniques.
• To evaluate the performance of six ensemble algorithms on the Indian Liver Patient Dataset (ILPD).

Main Methods:

• Application of multivariate imputation for missing values.
• Utilized log1p transformation, standardization, and various scaling techniques.
• Employed univariate selection, feature importance, and correlation matrix for feature selection.
• Trained Gradient boosting, XGBoost, Bagging, Random Forest, Extra Tree, and Stacking algorithms.

Main Results:

• The proposed model demonstrated superior performance compared to existing studies.
• Extra Tree Classifier achieved the highest testing accuracy at 91.82%.
• Random Forest achieved a testing accuracy of 86.06%.

Conclusions:

• The enhanced ensemble learning model provides a robust solution for early liver disease detection.
• Advanced preprocessing significantly improves prediction accuracy.
• The developed method shows potential for real-world clinical application.