# Serum Amino Acid and Fatty Acid Metabolites as Predictors of Sleep Disorders in Children: A Risk Prediction Model

## TL;DR

This study identifies specific amino acid and fatty acid metabolites in children's blood that can predict sleep disorders, offering a new tool for early detection and intervention.

## Contribution

A novel predictive model using serum metabolic indicators for early detection of sleep disorders in children is developed and validated.

## Key findings

- Seven amino acids were elevated in children with sleep disorders, while linoleic acid and taurine levels were reduced.
- The predictive model achieved high diagnostic accuracy with an AUC of 0.935 and good sensitivity and specificity.
- Taurine levels showed notable differences among sleep disorder subtypes, particularly in parasomnia and insomnia.

## Abstract

Objective: Adequate sleep is vital for children’s growth and well-being. This study investigates serum amino acid and fatty acid metabolic indicators in children with sleep disorders, identifies independent factors, and develops a predictive model. Methods: A total of 143 children diagnosed with sleep disorders (n = 143) were compared to 120 typically developing children (n = 120). Serum levels of 12 amino acids and 7 fatty acids were measured using liquid chromatography–tandem mass spectrometry. Differences between groups were assessed using t-tests or Mann–Whitney U tests. Independent factors were identified via multivariate logistic regression, leading to the construction of a predictive model. Its diagnostic efficacy was evaluated through receiver operating characteristic analysis, calibration curves, and decision curve analysis (DCA). Subgroup analysis of different sleep disorder subtypes was also performed to explore metabolic characteristic differences. Results: Significant differences in multiple metabolic indicators were found (p < 0.05) between these two groups. Seven amino acids were elevated, including glutamine and tryptophan, while linoleic acid and taurine levels were reduced. Analysis of four sleep disorder subtypes revealed no significant differences in most metabolic indicators among subtypes, with only taurine levels showing notable heterogeneity, the highest in parasomnia and the lowest in insomnia. Multivariate analysis revealed that arachidonic acid (OR = 0.75, 95% CI: 0.649–0.866), the ratio of cerotic acid to behenic acid (OR = 0.39, 95% CI: 0.186–0.816), aspartic acid (OR = 1.1, 95% CI: 1.040–1.164), glutamine (OR = 1.009, 95% CI: 1.004–1.014), taurine (OR = 0.985, 95% CI: 0.974–0.995), and phenylalanine (OR = 1.047, 95% CI: 1.018–1.078) were identified as independent factors for the development of sleep disorders (p < 0.05). The predictive model achieved the area under the ROC curve of 0.935 (95% CI: 0.904–0.967), with a threshold of 0.748 yielding sensitivity of 0.881 and specificity of 0.867. Ten-fold cross-validation confirmed robust generalizability (AUC: 0.927–0.916), and adjustable thresholds enabled flexible clinical application. Calibration curves and DCA demonstrated good agreement and clinical utility. Conclusions: Children with sleep disorders exhibit notable serum metabolic disturbances. The developed predictive model provides high diagnostic value and practicality for early screening and targeted interventions.