The emergence of antimicrobial resistance (AMR) poses a significant threat to global health, necessitating advanced computational approaches for early detection and prediction. This study presents a comprehensive comparative analysis of 13 machine learning algorithms across three critical tasks: antibiotic resistance prediction, multi-drug resistance (MDR) classification, and bacterial species identification. Using a large-scale dataset of 120,619 antimicrobial susceptibility testing records from 5,369 specimens collected over eight years (2016–2024), we evaluated tree-based methods, neural networks, support vector machines, and ensemble approaches. XGBoost achieved the highest performance in resistance prediction with 80.45% accuracy and 0.7925 F1-score, while Extra Trees excelled in bacterial species identification with 78.48% accuracy. Feature importance analysis revealed that antibiotic type and patient age are the most predictive factors for resistance outcomes. An ensemble method combining the top three performers achieved 80.51% accuracy, demonstrating the potential for improved prediction through model combination. These findings provide evidencebased guidance for implementing machine learning solutions in clinical antimicrobial stewardship programs and highlight the superior performance of gradient boosting methods for resistance prediction tasks.