Obstructive sleep apnea (OSA) is a common disorder in the pediatric population, primarily diagnosed through polysomnography (PSG). However, PSG can be expensive and is often limited in availability. This study aimed to develop a cost-effective diagnostic strategy by integrating a screening model with PSG.A retrospective analysis was conducted on children suspected of OSA. Screening models were initially constructed with machine learning techniques. Cost-utility analyses compared three diagnostic strategies: (1) PSG alone, (2) the screening model alone, and (3) the screening model-PSG combined, in the discovery and validation cohorts. Cost-utility was measured using the incremental net monetary benefit (INMB).690 children were included. The logistic regression model using age, tonsil scale, OSA-18 questions 1 and 2, and oxygen desaturation index 3% predicted OSA with an area under the curve of 0.91. In the cost-utility analysis, the "PSG alone" strategy, as the baseline, was the most beneficial (utility 0.9557) at Chinese Yuan (CNY) 4523.98. The "screening model alone" had 91.6% sensitivity and 59.3% specificity, offering lesser value (utility 0.9337) at CNY 6071.51 (INMB CNY -3966.43) when compared to "PSG alone". The "screening model-PSG combined" strategy increased sensitivity to 100%, specificity to 99.2%, and utility to 0.9554 at CNY 4463.36, establishing it as the most cost-effective option with an INMB of CNY 34.22. One-way sensitivity analyses and adaptation to US cost parameters confirmed the robustness of these results.Using the screening model as a triage tool for PSG enhances the cost-effectiveness of pediatric OSA management.© 2024 American Academy of Sleep Medicine.