Optimizing Antibiotic Stewardship through Machine Learning: Addressing the Rise of Multidrug-Resistant Infections in Clinical Settings

Abstract

Antimicrobial resistance (AMR) poses a significant global threat, with multidrug-resistant infections escalating healthcare costs and compromising patient outcomes. This study employs machine learning algorithms to enhance antibiotic stewardship, focusing on data collected from over 5,000 patient records at a major tertiary hospital. Utilizing a combination of logistic regression and random forest models, we identified key predictors of inappropriate antibiotic use, achieving an accuracy of 87% and a reduction in prescription errors by 30%. These findings illuminate critical pathways for clinicians to optimize treatment protocols in managing AMR effectively.