Pediatric SleepNet: a deep learning network for reliable pediatric sleep staging across developmental stages

Manual sleep staging in pediatric populations is challenging due to developmental variability and limited scoring consistency, especially in infants and toddlers. We developed a multimodal deep learning model for pediatric sleep staging and evaluated its performance across a broad age range and diverse clinical subgroups.

We trained a U-Net-inspired encoder-decoder model (pediatric SleepNet) using 9-channel input signals: electroencephalography (EEG), electrooculography (EOG), and chin electromyography (EMG) using 35-epoch segments from clinical pediatric polysomnograms (PSGs). Models were trained separately on three age groups (<6 months, 6-12 months, >1 year) using 9150 PSGs, with 2455 PSGs reserved for validation. Evaluation was conducted on 3804 held-out test recordings. Performance was compared with U-Sleep and the Complete Artificial Intelligence Sleep Report (CAISR), and stratified analyses were performed across ages, sexes, and seven ICD-10-based disease categories. External validation was conducted on two independent datasets, CHAT and PATS.

pediatric SleepNet achieved robust performance across all age groups, with mean Cohen's Kappa increasing from 0.49 (0-6 months) to 0.72 (>12 years). It significantly outperformed U-Sleep and CAISR across early developmental stages. Three-class staging yielded mean Cohen's Kappa increasing from 0.66 (0-6 months) to 0.79 (>12 years). Sex-based differences were negligible. However, significant reductions in performance were observed in children with epilepsy, Down syndrome, hydrocephalus, and other neurodevelopmental conditions. External validation yielded Kappa values >0.69 comparable to the internal test set.

pediatric SleepNet demonstrates reliable sleep staging across pediatric development. Its robust performance across age, disease, and external datasets supports its potential for clinical and research use in pediatric sleep medicine.