Exploring the Use of Artificial Intelligence-Based Innovations in Water, Sanitation and Hygiene Sector under Humanitarian Response in Fragile Settings: A Scoping Review

Abstract

Access to safe water, adequate sanitation, and appropriate hygiene services remains among the most critical and chronically unmet needs in humanitarian crises globally. Fragile settings, characterised by the intersection of armed conflict, institutional weakness, population displacement, and environmental stress, present the most severe WASH challenges while simultaneously being the contexts most resistant to conventional humanitarian response modalities. Artificial intelligence technologies, encompassing machine learning, deep learning, predictive analytics, Internet of Things-integrated sensor networks, natural language processing, and remote sensing, are undergoing rapid global diffusion into WASH management. However, the extent, character, and quality of their application in humanitarian fragile settings specifically remains poorly mapped in the academic and grey literature. This scoping review systematically characterises the current innovation landscape at the intersection of artificial intelligence and WASH under humanitarian response, examining what AI-based innovations have been deployed or piloted, across which WASH domains, by which actor types, with what documented outcomes, and against what structural and operational barriers. A systematic search across six databases and grey literature sources yielded 21 studies and reports meeting inclusion criteria after the screening of 312 initial records. Findings reveal that AI-WASH innovation in humanitarian fragile settings is nascent, geographically uneven, and concentrated in water quality monitoring and epidemiological disease prediction, with applications in sanitation management, hygiene behaviour change, governance support, and supply chain logistics remaining substantially underdeveloped. Critical barriers, including data scarcity in fragile settings, digital infrastructure deficits, power supply constraints, community trust challenges, and the absence of ethical frameworks governing AI deployment among vulnerable populations, systematically impede the translation of globally available AI-WASH tools into humanitarian operational practice. The review proposes the AI-WASH Humanitarian Integration Framework and evidence-based recommendations for actors across the humanitarian, research, and technology sectors.

Keywords