Perspective: Limiting Antimicrobial Resistance with Artificial Intelligence/Machine Learning

Area of Science:

• Clinical Microbiology
• Artificial Intelligence
• Machine Learning

Background:

• The increasing global threat of antimicrobial resistance necessitates advanced strategies for effective antibiotic use.
• Early expert systems like Mycin (1976) showed promise but faced limitations in clinical integration.
• Machine learning (ML) offers a solution to overcome expert system constraints by incorporating complex variables.

Purpose of the Study:

• To review the evolution of computer applications in clinical microbiology over 60 years.
• To highlight the potential of AI and ML in guiding antibiotic selection for bacterial infections.
• To discuss the role of AI in discovering novel antimicrobial agents.

Main Methods:

• Historical review of computer applications in clinical microbiology.
• Discussion of AI/ML program design incorporating diverse clinical and epidemiological variables.
• Exploration of AI's role in antimicrobial drug discovery.

Main Results:

• Early AI systems like Mycin were limited by rule-based constraints.
• Comprehensive AI/ML programs can integrate multiple factors influencing bacteria/drug interactions.
• AI holds promise for identifying new antimicrobial therapies.

Conclusions:

• AI and ML are crucial for optimizing antibiotic selection and combating antimicrobial resistance.
• Future AI/ML applications in clinical microbiology can enhance treatment outcomes and drug discovery.