摘要目的:探讨Ⅰ型发作性睡病（narcolepsy type 1，NT1）与Ⅱ型发作性睡病（narcolepsy type 2，NT2）的脑电跨频率耦合（cross-frequency coupling，CFC）特征差异。方法:基于中国临床睡眠数据库（Chinese Clinical Sleep Database，CCSD）2022年10月至2023年9月的数据，纳入23例NT1患者（NT1组）和31例NT2患者（NT2组）进行分析。所有受试者均接受多导睡眠监测及多次小睡睡眠潜伏时间试验。从多导睡眠监测记录的脑电信号中提取不同睡眠期、导联组合、频带组合及耦合类型构成的CFC特征。采用弹性网络正则化方法进行特征筛选，并分析筛选出的关键CFC特征与艾普沃斯嗜睡量表（Epworth Sleepiness Scale，ESS）评分的Spearman相关性。最后，构建支持向量分类器（support vector classification，SVC）用于区分NT1与NT2，并采用留一交叉验证法评估模型的泛化能力。结果:非快速眼动睡眠1期（non rapid-eye movement sleep stage 1，N1）中，额-枕θ-α1耦合和中央区δ-α1耦合的特征系数绝对值最高，分别为1.13和1.10。在NT1组中，N1中F3-C3通道的α1-β2的虚部锁相值（imaginary part of phase locking value，iPLV）与ESS评分呈正相关（ r=0.52， P=0.012）。在机器学习分类任务中，留一交叉验证法显示SVC模型的分类准确率达85%。 结论:睡眠-觉醒交界阶段的CFC特征在区分NT1和NT2中具有重要作用，并在NT1中与日间过度思睡具有相关性。CFC有望作为发作性睡病亚型识别的潜在生物标志物，并为日间过度思睡的机制研究和临床评估提供新思路。

abstractsObjective:To investigate the differences in cross-frequency coupling (CFC) characteristics of electroencephalography (EEG) between narcolepsy type 1 (NT1) and narcolepsy type 2 (NT2).Methods:A total of 23 NT1 and 31 NT2 patients were included from the Chinese Clinical Sleep Database (CCSD) between October 2022 and September 2023. All participants underwent overnight polysomnography and a multiple sleep latency test. CFC features were extracted from EEG signals during polysomnography, encompassing various combinations of sleep stages, electrode pairs, frequency bands, and coupling types. Feature selection was performed using elastic net regularization. The Spearman correlation between key CFC features and the Epworth Sleepiness Scale (ESS) scores was analyzed. Finally, a support vector classification (SVC) model was constructed to distinguish NT1 from NT2, and leave-one-out cross-validation was used to assess the generalization performance.Results:Among all coupling features during non-rapid eye movement sleep stage 1 (N1), the fronto-occipital θ-α1 and central δ-α1 couplings showed the highest absolute coefficients, reaching 1.13 and 1.10, respectively. In the NT1 group, the α1-β2 imaginary part of phase-locking value (iPLV) of the F3-C3 pair during N1 was significantly positively correlated with ESS scores ( r=0.52, P=0.012). In the machine learning classification task, the SVC model achieved an accuracy of 85% using leave-one-out cross-validation. Conclusion:The CFC features during the sleep-wake transition stage play an important role in distinguishing NT1 from NT2 and show a significant correlation with excessive daytime sleepiness (EDS) in NT1. CFC may serve as a potential biomarker for differentiating narcolepsy subtypes and provide new insights into the mechanisms and clinical evaluation of EDS.