失眠症評估之新進展

The Assessment of Insomnia

失眠從來不只是「睡不著」這麼簡單，而是睡眠、身體、情緒、認知與行為交互失衡之後，在夜裡浮現的一種臨床綜合現象。DSM-5、ICSD-3與ICD-11逐步淡化「原發性」與「次發性」失眠的二分，改以統一的失眠障礙概念，強調在具備足夠睡眠機會與適當環境下，仍持續出現入睡困難、睡眠維持困難或過早醒來，並造成主觀困擾或日間功能受損。這樣的整合有助於臨床辨識，卻也把病程、表型、共病與生理機轉差異極大的病人，一併收攏在同一個名稱之下。因此，下一步評估不該是再多發一張問卷，而是把主觀經驗、時間脈絡、心理與行為機轉、共病風險，以及多夜的客觀訊號整合起來。近年失眠評估正由「問卷加上單夜多頻道睡眠檢查」，逐漸走向精準表型化。心理表型可依困擾程度、情緒反應、獎勵敏感性與生活史加以分層；神經生理表型則可結合客觀睡眠時間與非快速動眼期腦波頻譜，辨識出「短睡且慢波不足」「睡眠時間正常但慢波不足」，以及「宏觀睡眠結構接近正常」等亞型。同時，失眠與阻塞型睡眠呼吸中止症的高度共病（co-morbid insomnia and sleep apnea，COMISA），也提醒我們：不能只處理「睡不著」，也不能只盯著呼吸中止指數。新一代的非接觸、非穿戴技術，為我們打開了另一扇觀察窗。24 GHz雷達可在完全不接觸身體的情況下，長期量測胸壁微動與呼吸、心搏相關訊號，結合深度學習估算睡眠狀態及呼吸異常事件；我們團隊參與開發的「醫倍思」非接觸式睡眠呼吸中止症檢測系統，已於2025年底取得衛福部第2等級醫療器材許可。另外，智慧床墊等非穿戴設備則可鋪在枕頭或床墊上，擷取呼吸、心搏與體動，並進一步分析呼吸性竇性心律不整耦合（respiratory sinus arrhythmia coupling，RSA coupling），探索自律神經、慢波睡眠與情緒過度喚醒之間的關聯。本文提出一條以臨床問題為核心、由低負擔到高解析、由單次診斷到長期追蹤的分層評估路徑。這些科技此刻應被定位為篩檢、表型化與追蹤的增益工具，而不是取代臨床訪談或多項睡眠生理檢查。真正的進展，是讓失眠從一個模糊的主觀標籤，轉為一份可理解、可追蹤、可導向個別化處置的生理儀表板——讓評估不再只是比誰「更睡不著」，而是讀懂夜裡身體說出的語言。

Insomnia is never simply a matter of“being unable to sleep”; it is a clinical phenomenon that surfaces at night once sleep, the body, emotion, cognition, and behavior fall out of balance. DSM-5, ICSD-3, and ICD-11 have progressively moved away from the“primary versus secondary”dichotomy toward a unified concept of insomnia disorder, defined by persistent difficulty initiating sleep, difficulty maintaining sleep, or early-morning awakening—accompanied by subjective distress or daytime impairment—despite adequate opportunity and a suitable environment for sleep. This integration improves clinical recognition, but it also gathers under one name patients who differ greatly in course, phenotype, comorbidity, and pathophysiology. The next step in assessment, therefore, is not to administer one more questionnaire, but to integrate subjective experience, temporal context, psychological and behavioral mechanisms, comorbidity risk, and multi-night objective signals. In recent years, insomnia assessment has been shifting from“a questionnaire plus a single night of polysomnography”toward precision phenotyping. Psychological phenotypes can be stratified by the level of distress, emotional reactivity, reward sensitivity, and life history, whereas neurophysiological phenotypes can combine objective sleep duration with non–rapid-eye-movement electroencephalographic spectral power to distinguish subtypes such as short sleep with deficient slow-wave activity, normal sleep duration with deficient slow-wave activity, and near-normal sleep macrostructure. Meanwhile, the high comorbidity of insomnia and obstructive sleep apnea (co-morbid insomnia and sleep apnea, COMISA) cautions us against treating only the complaint of sleeplessness or fixating solely on the apnea–hypopnea index. A new generation of contactless, non-wearable technologies opens a further observational window: a 24-GHz radar can measure chest-wall micromotion and respiration- and heartbeat-related signals without any bodily contact and, coupled with deep learning, estimate sleep state and abnormal respiratory events; the contactless sleep-apnea detection system that our team helped develop received Class II medical-device clearance from Taiwan’s Ministry of Health and Welfare in late 2025. Non-wearable devices such as smart mattresses can likewise capture respiration, heartbeat, and body movement, and can further analyze respiratory sinus arrhythmia (RSA) coupling to probe the interplay among autonomic function, slow-wave sleep, and emotional hyperarousal. This article proposes a tiered assessment pathway that is anchored in the clinical question, that moves from low burden to high resolution, and that extends from a single diagnostic encounter to long-term follow-up. For now, these technologies should be positioned as augmentative tools for screening, phenotyping, and monitoring rather than as replacements for the clinical interview or polysomnography. Genuine progress lies in transforming insomnia from a vague subjective label into an intelligible, trackable physiological dashboard that can guide individualized care—so that assessment becomes not a contest over who“sleeps worse,”but a way to read the language the body speaks at night.