A responsible AI framework for infection surveillance in low-resource settings: ethics, opportunities and threats for LMICs (EOT-LMICs)

Comprehensive Summary

This study, presented by Alzyood, examines the role of artificial intelligence (AI) in infection surveillance within low- and middle-income countries (LMICs), proposing a new framework called EOT-LMICs (Ethics, Opportunities, and Threats). The framework was developed through synthesis of existing literature on AI adoption, digital health implementation, and infection surveillance, alongside expert discussions to ensure relevance to real-world LMIC contexts. The authors argue that current frameworks lack LMIC-specific guidance and fail to integrate ethics, implementation, and surveillance into a unified model. The EOT-LMICs framework introduces three core components: (1) ethical foundations (equity, data governance, contextual validity), (2) opportunities (e.g., workforce support, standardized surveillance, earlier outbreak detection), and (3) threats (e.g., misclassification, infrastructure limitations, model drift, and regulatory gaps). The article emphasizes that ethical preconditions must be satisfied before implementation, and risks must be mitigated alongside recognizing benefits. It also integrates the “4Ps” model (precision, partnership, practice, people) as a pathway for implementation, stressing that AI must be context-specific, human-centered, and aligned with local healthcare systems.

Outcomes and Implications

This framework is important because it provides a structured, ethics-first approach to integrating AI into infection surveillance in resource-limited healthcare systems, where improper implementation could worsen inequities or compromise patient care. Clinically, AI has the potential to improve early detection of outbreaks, enhance surveillance accuracy, and reduce the burden on limited healthcare workforces—particularly in settings lacking specialized infection control personnel. However, the article highlights that without proper safeguards, AI could lead to harmful misclassification, missed diagnoses, or overreliance on flawed systems. In practice, this framework could guide policymakers and healthcare institutions in deciding whether and how to adopt AI tools, ensuring that systems are validated with local data and integrated into existing workflows. In the near term, its application may influence pilot programs and policy development in LMICs; longer term, it could shape global standards for ethical AI in public health. Ultimately, the framework reinforces that AI should augment—not replace—clinical judgment, preserving human oversight while improving infection surveillance outcomes.