The Quest for Automated Pediatric Sleep Scoring: Are We There Yet?
Automated sleep scoring algorithms have transformed adult sleep laboratories over the past decade, but a new article in the journal Sleep asks whether the same technology is ready for pediatric patients. The answer, according to pediatric sleep specialist Alex Gileles-Hillel and biomedical engineer Joachim A Behar, is more complicated than a simple yes or no.
Published online June 27 by the Sleep Research Society, the article brings together clinical and technical perspectives from Hadassah Medical Center, the Hebrew University of Jerusalem, and the Technion-Israel Institute of Technology. Gileles-Hillel leads the Pediatric Pulmonology and Sleep Unit at Hadassah, while Behar brings deep expertise in machine learning applied to physiological signals.
Why pediatric sleep scoring is different
Children are not small adults when it comes to sleep. Their sleep architecture differs substantially from adults: more slow-wave sleep, different EEG morphology, and age-dependent normative values that shift rapidly during development. An algorithm trained on adult polysomnography data can systematically misclassify pediatric sleep stages, leading to inaccurate clinical assessments.
Compounding this, pediatric sleep laboratories often have limited access to the large, well-annotated datasets needed to train robust machine learning models. Adult sleep scoring benefits from decades of accumulated data and established commercial systems; the pediatric pipeline is far less mature.
What remains to be solved
The article highlights several unresolved challenges. First, data scarcity: most pediatric sleep datasets are small, institution-specific, and lack standardized labeling protocols. Second, validation standards: no consensus exists on what constitutes sufficient clinical validation for an automated pediatric scoring system before deployment. Third, developmental variability: algorithms must account for the rapid changes in sleep EEG across infancy, childhood, and adolescence, which few current models handle.
The authors’ multidisciplinary lens reflects a growing recognition that the path forward requires collaboration between sleep clinicians who understand pediatric physiology and data scientists who can build algorithms robust enough for clinical use.
Why it matters
Accurate sleep scoring is the foundation of pediatric sleep medicine. Misclassification can delay diagnosis of sleep-disordered breathing, parasomnias, and other conditions that affect development, behavior, and quality of life. If automated tools can be validated for children, they could expand access to objective sleep assessment in settings where manual scoring by trained technicians is unavailable or prohibitively expensive.
For now, the question posed in the title remains open. But by framing the gaps clearly, Gileles-Hillel and Behar provide a roadmap for the work still needed.
Source
Gileles-Hillel A, Behar JA. The Quest for Automated Pediatric Sleep Scoring: Are We There Yet? Sleep. 2026 Jun 27:zsag174. doi: 10.1093/sleep/zsag174. PMID: 42364168.