Diagnostic modalities in pediatric obstructive sleep apnea: A review article.

BACKGROUND

Pediatric obstructive sleep apnea (OSA) affects ∼3 % of children and causes adverse neurocognitive, behavioral, and cardiovascular outcomes. Polysomnography (PSG) remains the gold standard but faces accessibility constraints. This review synthesizes evidence on alternative diagnostic tools, focusing on accuracy, feasibility, and clinical applicability.

METHODS

This systematic-narrative hybrid review evaluated alternative pediatric OSA diagnostics through database searches (PubMed, Scopus, Web of Science) through July 2025. Eligible studies included children aged 0-18 years assessed with questionnaires, oximetry, portable monitors, machine learning/artificial intelligence models, smartphone video analysis, or novel modalities compared with PSG.

RESULTS

Five modality categories emerged: pulse oximetry, questionnaires, portable monitoring, machine learning/artificial intelligence, and emerging technologies. Screening tools prioritized sensitivity over specificity; diagnostic tools demonstrated the reverse. The Pediatric Sleep Questionnaire (PSQ) and smartphone video analysis (VAPS) achieved highest screening sensitivity (73-96 %), with VAPS providing objective visual documentation when parents capture representative breathing. Algorithm-enhanced oximetry offered optimal diagnostic accuracy for moderate-to-severe OSA (AUC 0.90-0.97, sensitivity/specificity >85-90 %). Portable monitoring performed well but underestimated mild disease. Machine learning models and wearables (mandibular sensors, software-generated AHI) approached PSG-level performance. Meta-analyses confirmed optimal stepwise screening combines high-sensitivity questionnaires with high-specificity oximetry.

CONCLUSION

Where PSG is unavailable, questionnaires and smartphone video analysis provide effective screening, while automated oximetry and machine learning models offer diagnostic alternatives for moderate-to-severe OSA. Tiered pathways integrating questionnaires, video analysis, and objective testing optimize detection while reducing PSG dependence.