Applications of Artificial Intelligence in Antimicrobial Resistance Surveillance, Prediction, and Control in Low- and Middle-Income Countries: A Scoping Review

Why It Matters

AI could dramatically improve AMR detection and response in resource‑constrained settings, yet structural gaps and bias risk widening health inequities if not addressed.

Key Takeaways

• 128 studies (2013‑2026) map AI uses in LMIC AMR.
• Most models validated retrospectively; few real‑world trials.
• Infrastructure, data quality, and workforce gaps hinder deployment.
• Algorithmic bias threatens equity in rural health settings.
• Investment needed in data systems and fairness‑aware AI.

Pulse Analysis

Antimicrobial resistance remains a leading global health threat, with low‑ and middle‑income countries bearing a disproportionate burden due to fragmented surveillance networks and limited laboratory capacity. In this context, artificial intelligence and machine learning promise faster, more accurate detection of resistant pathogens and smarter stewardship of existing drugs. By leveraging pattern recognition and predictive analytics, AI can transform scattered clinical data into actionable insights, supporting clinicians and policymakers in curbing the spread of resistance before outbreaks become unmanageable.

The scoping review examined 128 peer‑reviewed and grey‑literature publications, revealing a broad but uneven landscape of AI applications. Surveillance tools accounted for 28 studies, while resistance‑prediction models appeared in 31, and rapid diagnostic or antimicrobial susceptibility testing (AST) accelerators featured in 25. Performance metrics were often impressive in controlled laboratory settings, yet the majority of work relied on retrospective datasets from urban tertiary hospitals. Critical implementation barriers emerged, including inadequate digital infrastructure, scarce high‑quality training data, limited interoperability, and a pronounced shortage of skilled personnel to maintain AI pipelines. Moreover, the review highlighted a glaring gap in prospective, real‑world validation, raising concerns about the generalizability of reported accuracies.

For stakeholders, the findings underscore that technical readiness alone will not deliver AI‑driven AMR control. Policymakers must prioritize investments in robust, interoperable health information systems, ensure equitable data collection that captures rural and community health contexts, and embed fairness‑aware model development to mitigate algorithmic bias. Collaborative implementation research, supported by international donors and local ministries, can bridge the gap between proof‑of‑concept studies and scalable solutions. By aligning AI innovation with sustainable infrastructure and governance frameworks, LMICs can harness these technologies to strengthen surveillance, guide stewardship, and ultimately reduce the global toll of antimicrobial resistance.