Artificial intelligence in sleep diagnostics for children under 2 years of age: state of the evidence and future directions

Background Artificial intelligence (AI) has advanced rapidly in adult sleep medicine, but its role in children under 2 years old remains unclear. This scoping review maps the current evidence on AI-enabled diagnostics in neonates, infants and toddlers, identifying challenges, opportunities and research priorities. Methods A structured scoping review was undertaken up to April 2025, incorporating peer-reviewed studies, normative datasets, regulatory reports and evidence on consumer and commercial technologies. Studies focusing on children under 2 years were prioritised, with older cohorts reviewed where relevant to translation. Findings No AI tool is validated for independent diagnostic use in this age group. Research prototypes, including automated polysomnography scoring, convolutional neural networks for oximetry and multimodal wearable or radar-based systems, show feasibility but remain experimental. Commercial platforms such as SleepImage, Belun Sleep Health and WatchPAT exclude children under 2 years, while consumer monitors including Owlet, Nanit and Miku are widely used but lack clinical validation. Conclusion AI in sleep diagnostics for children under 2 years remains an emerging field with high potential but no current clinical readiness. Progress will depend on the development of infant-specific datasets, improved artefact handling, simplified monitoring strategies and human-in-the-loop workflows. Longer-term opportunities include prognostic modelling, circadian analysis and predictive closed-loop systems, but equity and regulatory clarity must guide translation to ensure safe and globally relevant impact.