Scholarlink

Antimicrobial resistance (AMR) weakens the body’s defense against infections, posing severe risks to patients undergoing medical treatment. The growing prevalence of resistant pathogens such as Staphylococcus aureus, Enterococcus spp., Klebsiella pneumoniae, and Pseudomonas aeruginosa has raised major clinical concerns. Leveraging rich medical data and predictive modeling now enables the development of evidence-based frameworks for AMR management. This study presents a comprehensive AMR prediction pipeline integrating feature selection techniques, classifier variants, and class balancing. Empirical results show that Random Forest and Gradient Boosting outperform other models, while SHAP analysis identifies tet(D), tRNA, and Contigs as key determinants of resistance offering interpretable insights into patient-level AMR risk.

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

Feature Selection and SHAP-Based Interpretability in ML Models for AMR Prediction in Klebsiella Pneumoniae

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Lov Kumar,
  • Vikram Singh,
  • Nikita

摘要

Antimicrobial resistance (AMR) weakens the body’s defense against infections, posing severe risks to patients undergoing medical treatment. The growing prevalence of resistant pathogens such as Staphylococcus aureus, Enterococcus spp., Klebsiella pneumoniae, and Pseudomonas aeruginosa has raised major clinical concerns. Leveraging rich medical data and predictive modeling now enables the development of evidence-based frameworks for AMR management. This study presents a comprehensive AMR prediction pipeline integrating feature selection techniques, classifier variants, and class balancing. Empirical results show that Random Forest and Gradient Boosting outperform other models, while SHAP analysis identifies tet(D), tRNA, and Contigs as key determinants of resistance offering interpretable insights into patient-level AMR risk.