A novel approach to antimicrobial resistance: Machine learning predictions for carbapenem-resistant Klebsiella in intensive care units

ALPARSLAN V., GÜLER Ö., İNNER A. B., DÜZGÜN A., Baykara N., KUŞ A.

International Journal of Medical Informatics, cilt.195, 2025 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 195
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1016/j.ijmedinf.2024.105751
  • Dergi Adı: International Journal of Medical Informatics
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, BIOSIS, Biotechnology Research Abstracts, CAB Abstracts, CINAHL, Compendex, EMBASE, INSPEC, MEDLINE
  • Anahtar Kelimeler: Carbapenem-Resistant Enterobacteriaceae, Intensive Care Units, Klebsiella pneumoniae, Supervised Machine Learning
  • Kocaeli Üniversitesi Adresli: Evet

Özet

This study was conducted at Kocaeli University Hospital in Turkey and aimed to predict carbapenem-resistant Klebsiella pneumoniae infection in intensive care units using the Extreme Gradient Boosting (XGBoost) algorithm, a form of artificial intelligence. This was a retrospective case-control study involving 289 patients, including 159 carbapenem-resistant and 130 carbapenem-susceptible individuals as controls. The model's predictive analysis combined a diverse range of demographic, clinical, and laboratory data, resulting in an average accuracy of 83.0%, precision of 83%, sensitivity of 88%, F1 score of 85%, and Matthews Correlation Coefficient of 0.66. Prolonged hospitalization and intensive care unit stay were significant predictors of carbapenem-resistant Klebsiella pneumoniae infection. The role of artificial intelligence role in healthcare, particularly in ICUs for managing antibiotic-resistant infections, is a major development in medicine. This study emphasizes the potential of artificial intelligence to predict antimicrobial resistance and improve clinical decisions in resource-limited settings. The study was approved by ClinicalTrials.gov (trial registration number NCT05985057 on 02.08.2023).