Keyword: Artificial Intelligence
1 result found.
Review Article
International Journal of Evidence-Based Medicine, 1(3), 2026, jebm015, https://doi.org/10.63946/jebm/19463
ABSTRACT:
Background: Antimicrobial resistance (AMR) surveillance is essential for estimating resistance burden and guiding public-health responses, yet coverage remains uneven across geographic regions, populations, healthcare settings, and One Health domains. Low observed resistance may therefore reflect limited detection and reporting rather than genuinely low resistance.
Objectives: To map the nature and determinants of AMR under-surveillance, examine computational approaches for identifying surveillance blind spots, and develop an evidence-informed framework for AI-enabled detection of hidden AMR surveillance gaps.
Methods: PubMed and Dimensions were systematically searched for literature published from 2015–2026. Eligible studies addressed AMR surveillance coverage, capacity, representativeness, completeness, or computational approaches relevant to surveillance gaps. Study selection and synthesis followed JBI scoping-review methodology and PRISMA-ScR guidance.
Results: Twenty-two studies were included. Under-surveillance encompassed geographic and population underrepresentation, diagnostic and laboratory limitations, fragmented information systems, delayed reporting, weak genomic coverage, and incomplete One Health integration. Computational approaches included spatial and statistical modelling, machine learning, geographic information systems, genomic analytics, digital laboratory systems, and environmental surveillance. These approaches could identify or partially address surveillance incompleteness but remained constrained by data quality and representativeness. The synthesis informed a framework for detecting discordance between expected AMR risk and observed surveillance intensity.
Conclusions: AMR under-surveillance is a systems-level problem that can distort the apparent distribution of resistance. AI should complement microbiological surveillance by identifying uncertainty and prioritizing settings requiring strengthened surveillance.
Objectives: To map the nature and determinants of AMR under-surveillance, examine computational approaches for identifying surveillance blind spots, and develop an evidence-informed framework for AI-enabled detection of hidden AMR surveillance gaps.
Methods: PubMed and Dimensions were systematically searched for literature published from 2015–2026. Eligible studies addressed AMR surveillance coverage, capacity, representativeness, completeness, or computational approaches relevant to surveillance gaps. Study selection and synthesis followed JBI scoping-review methodology and PRISMA-ScR guidance.
Results: Twenty-two studies were included. Under-surveillance encompassed geographic and population underrepresentation, diagnostic and laboratory limitations, fragmented information systems, delayed reporting, weak genomic coverage, and incomplete One Health integration. Computational approaches included spatial and statistical modelling, machine learning, geographic information systems, genomic analytics, digital laboratory systems, and environmental surveillance. These approaches could identify or partially address surveillance incompleteness but remained constrained by data quality and representativeness. The synthesis informed a framework for detecting discordance between expected AMR risk and observed surveillance intensity.
Conclusions: AMR under-surveillance is a systems-level problem that can distort the apparent distribution of resistance. AI should complement microbiological surveillance by identifying uncertainty and prioritizing settings requiring strengthened surveillance.
