Development of antibiotic regimens using graph based evolutionary algorithms

Area of Science:

• Computational biology
• Evolutionary algorithms
• Animal health
• Antimicrobial resistance

Background:

• Antibiotic use in production animals is crucial for health and growth but contributes to antimicrobial resistance.
• Developing effective antibiotic regimens that balance benefits with resistance risk is a significant challenge.
• Production animals host complex bacterial communities, including pathogens like Campylobacter spp.

Purpose of the Study:

• To develop and evaluate evolutionary algorithms for optimizing antibiotic regimens in production animals.
• To assess the impact of diversity control in graph-based evolutionary algorithms on antibiotic regimen development.
• To identify antibiotic strategies that minimize resistance while preserving animal health and productivity.

Main Methods:

• A simulation model was created integrating animal lifespan and gut bacteria dynamics.
• Graph-based evolutionary algorithms with diversity control were employed for fitness evaluation.
• Tylosin phosphate regimens were tested against Gram-positive and Gram-negative bacteria, focusing on Campylobacter spp.

Main Results:

• Optimized antibiotic regimens were identified that significantly decreased antibiotic resistance compared to conventional methods.
• The developed regimens maintained nearly equivalent weight gain and health benefits.
• Graph-based evolutionary algorithms effectively identified diverse solutions along the Pareto front for multi-objective optimization.

Conclusions:

• Evolutionary algorithms offer a powerful approach to designing safer and more effective antibiotic regimens for production animals.
• Controlling information flow in evolutionary algorithms facilitates the discovery of optimal multi-objective solutions.
• This methodology provides a framework for mitigating antibiotic resistance in livestock while ensuring animal welfare.