Scholarlink

This research presents an innovative machine learning framework designed for the early detection of liver disease, with the goal of enhancing diagnostic precision and improving patient care. The pre-processing phase addresses missing data by applying column mean imputation and transforms categorical variables into numerical representations for seamless analysis. Principal Component Analysis (PCA) is utilized for dimensionality reduction, ensuring the retention of critical data variance while minimizing complexity. A hybrid feature selection strategy combines the strengths of Random Forest and Particle Swarm Optimization (PSO), where PSO is used to identify the most significant features for predictive modeling. The XGBoost algorithm, recognized for its robustness and strong predictive capabilities, serves as the primary model for classification. To further enhance its performance, a grid search technique is applied for hyperparameter optimization. This comprehensive approach underscores the potential of building an accurate and efficient system for the early detection of liver disease. The proposed framework not only supports timely medical interventions but also contributes to improved clinical outcomes by streamlining the diagnostic process.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Early Detection of Liver Disease Using Machine Learning: An Integrated Approach with Feature Selection and XGBoost Optimization

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Priti V. Kale,
  • Narendra M. Kandoi,
  • Mahesh S. Shinde,
  • Chitrakant O. Banchhor,
  • Vikas Maral,
  • Dattatray G. Takale,
  • Parikshit N. Mahalle,
  • Bipin Sule

摘要

This research presents an innovative machine learning framework designed for the early detection of liver disease, with the goal of enhancing diagnostic precision and improving patient care. The pre-processing phase addresses missing data by applying column mean imputation and transforms categorical variables into numerical representations for seamless analysis. Principal Component Analysis (PCA) is utilized for dimensionality reduction, ensuring the retention of critical data variance while minimizing complexity. A hybrid feature selection strategy combines the strengths of Random Forest and Particle Swarm Optimization (PSO), where PSO is used to identify the most significant features for predictive modeling. The XGBoost algorithm, recognized for its robustness and strong predictive capabilities, serves as the primary model for classification. To further enhance its performance, a grid search technique is applied for hyperparameter optimization. This comprehensive approach underscores the potential of building an accurate and efficient system for the early detection of liver disease. The proposed framework not only supports timely medical interventions but also contributes to improved clinical outcomes by streamlining the diagnostic process.