Artificial Intelligence in Helicobacter Pylori Infection: Diagnostic Applications and Emerging Treatment-Related Predictive Uses—A Systematic Review

Background: Artificial intelligence (AI) has shown growing potential in the diagnosis of H. pylori infection, particularly through automated analysis of endoscopic images. Emerging studies have also explored treatment-related predictive applications, although this evidence remains limited. The aim of this systematic review was to synthesize current evidence on the use of AI in H. pylori infection, with the primary emphasis on diagnosis and secondary consideration of predictive therapeutic applications. Methods: A systematic review was conducted in accordance with PRISMA 2020 guidelines through searches in PubMed, ScienceDirect, EBSCO, and the Cochrane Library, including articles published between 2020 and 2025. Six studies that employed deep learning or machine learning models, primarily convolutional neural networks and predictive classifiers, were selected. Results: Artificial intelligence models showed consistent diagnostic performance, with accuracies ranging from 79.2% to 94%, sensitivities from 62.5% to 96%, and specificities from 79.4% to 93.4%. Convolutional neural network-based systems generally demonstrated diagnostic performance comparable to or better than that of human endoscopists, particularly among less experienced operators. Limited evidence suggests a possible role for artificial intelligence in predicting treatment failure; however, this finding is based on a single included study. Conclusions: Artificial intelligence appears to be a promising complementary tool for the diagnosis of H. pylori infection, particularly in endoscopic imaging. However, evidence regarding treatment-related and resistance-related applications remains limited and indirect, and these potential uses should therefore be considered preliminary.