Abstract
Background
The global rise in antimicrobial resistance (AMR) among
Neisseria gonorrhoeae
presents a major public health threat, complicating treatment and control efforts. Traditional diagnostic methods for AMR detection are time-consuming and often limited by laboratory resources, particularly in low- and middle-income countries. The rapid evolution of machine learning (ML) models offers new opportunities for predictive diagnostics that can enhance surveillance, optimize antibiotic therapy, and reduce transmission.
Aim
This systematic review and meta-analysis aimed to evaluate the diagnostic accuracy of machine learning models in predicting antimicrobial resistance in
Neisseria gonorrhoeae
and to provide pooled estimates of sensitivity and specificity compared with conventional reference standards.
Methods
A comprehensive search of seven databases PubMed, Scopus, Web of Science, Embase, CINAHL, IEEE Xplore, and Google Scholar was conducted for studies published up to 2025. Eligible studies applied ML algorithms to genomic, phenotypic, or epidemiological datasets for predicting AMR in
N. gonorrhoeae
. Data were extracted into Microsoft Excel and analyzed using RevMan 5.4 software version 5.4.1. Quality assessment was conducted using the QUADAS-2 tool. Pooled sensitivity, specificity, and area under the SROC curve (AUC) were calculated using a random-effects bivariate model.
Results
Five eligible studies encompassing unique
Neisseria gonorrhoeae
isolates were included. The pooled sensitivity and specificity of ML models were 0.94 (95% CI: 0.92–0.96) and 0.86 (95% CI: 0.81–0.90), respectively. The SROC curve demonstrated an AUC of 0.95, indicating excellent discriminative ability. Moderate heterogeneity (
I
2
≈ 40%) was observed, largely due to variations in datasets and model architectures.
Conclusion
Machine learning models exhibit outstanding diagnostic accuracy in predicting AMR in
Neisseria gonorrhoeae
, highlighting their potential integration into surveillance and clinical decision-support systems. Broader validation and standardization of ML pipelines are essential to translate these advances into global public health practice.