Machine-learning approaches prevent post-treatment resistance-gaining bacterial recurrences

Area of Science:

• Infectious Diseases
• Genomics
• Computational Biology
• Antimicrobial Resistance

Background:

• Recurrent infections pose a significant clinical challenge, often linked to the development of antimicrobial resistance (AMR).
• Standard susceptibility testing may not fully explain infection recurrence, highlighting a gap in understanding disease dynamics.

Purpose of the Study:

• To investigate the genomic basis of recurrent infections and identify factors contributing to antimicrobial resistance.
• To develop a predictive model for personalized recommendations to mitigate AMR at the individual patient level.

Main Methods:

• Whole-genome sequencing was employed to analyze the genetic relatedness of infecting strains in recurrent cases.
• Machine-learning algorithms were utilized to analyze genomic data and identify patterns associated with infection recurrence and resistance.

Main Results:

• Whole-genome sequencing revealed that recurrent infections are frequently caused by distinct strains, rather than solely by the evolution of the original pathogen.
• The developed machine-learning algorithm demonstrated potential in providing patient-specific recommendations to minimize antimicrobial resistance.

Conclusions:

• Understanding the strain dynamics of recurrent infections is crucial for effective treatment strategies.
• Personalized, data-driven approaches hold promise for managing antimicrobial resistance and improving patient outcomes.