Machine learning models for Neisseria gonorrhoeae antimicrobial susceptibility tests

Area of Science:

• Genomic Medicine
• Computational Biology
• Infectious Diseases

Background:

• Neisseria gonorrhoeae poses an urgent public health threat due to rising antibiotic resistance.
• Current treatment options are limited by widespread antimicrobial resistance.
• Rapid molecular antimicrobial susceptibility tests (ASTs) are needed to guide effective therapy.

Purpose of the Study:

• To present a framework for developing sequence-based diagnostics to predict Neisseria gonorrhoeae antibiotic susceptibility.
• To illustrate the application of machine learning models in designing genotype-based ASTs.
• To guide the selection of features for predicting antibiotic resistance phenotypes from genomic data.

Main Methods:

• Utilizing a large dataset of nearly 20,000 clinical N. gonorrhoeae isolates with genome sequence and antibiotic resistance phenotype data.
• Applying statistical and machine learning methods to identify genotype-phenotype correlations.
• Establishing assay technology, performance criteria, target population, and clinical goals for diagnostic development.

Main Results:

• A framework is presented for translating genomic data into predictions of antibiotic susceptibility.
• The approach enables the identification of genetic features predictive of antimicrobial resistance.
• The methodology is demonstrated using Neisseria gonorrhoeae as a model organism.

Conclusions:

• Machine learning applied to genomic data offers a powerful approach for developing novel ASTs.
• This framework can aid in tailoring antibiotic therapy for N. gonorrhoeae infections.
• The generalized framework can be applied to other pathogens with available genotype and resistance data.