Efficacy and Accuracy of an AI-Driven Sleep Earbud for Chronic Insomnia

N/AActive Not RecruitingNCT07744581

The Chinese University of Hong Kong, Shenzhen25 enrolled

Overview

This study evaluates a new smart sleep earbud designed to help adults suffering from chronic insomnia. The device uses artificial intelligence (AI) to track a user's real-time heart rate and movement through the ear canal, automatically adjusting soothing music parameters to help the user fall asleep faster and achieve deeper sleep. Participants will spend three consecutive nights in a hospital sleep laboratory. The first night serves as a baseline screening using medical-grade sleep tracking (polysomnography) to rule out other hidden sleep conditions like sleep apnea. On the second and third nights, participants will test two different audio options in a randomized order: the AI-driven adaptive music and standard, non-adjusting music. Researchers will compare the earbud's internal sensor data against the hospital's clinical equipment to verify the earbud's tracking accuracy, and participants will complete brief touch-screen brain function tests each morning. Following the lab phase, participants will continue using the earbuds in their natural home environment for two weeks before a final check-up. The goal is to determine if personalized, AI-adjusted sound therapy can effectively treat insomnia symptoms and if a consumer ear-worn device can monitor sleep architecture as accurately as a clinical hospital system.

This randomized, double-blind, two-sequence crossover clinical trial is designed to evaluate both the therapeutic efficacy of a closed-loop acoustic neuromodulation ear-worn device and the measurement accuracy of its embedded sensors against gold-standard laboratory diagnostics. The study architecture is executed across two distinct phases: a controlled laboratory phase followed by a naturalistic home-use extension. Phase 1: Controlled Laboratory Assessment and Screening (Days 1-3) Participants undergo consecutive three-night stays within a regulated hospital sleep medicine center. Night 1 (Baseline and Diagnostic Screening): Participants are instrumented with a mobile polysomnography (PSG) system (SOMNOscreen™ plus) to capture baseline architecture across standard electrophysiological channels (EEG, EOG, EMG, ECG). This night serves to objectively screen for and exclude individuals presenting with hidden primary sleep disorders, specifically moderate-to-severe obstructive sleep apnea characterized by an Apnea-Hypopnea Index (AHI). No audio intervention is delivered. Nights 2 and 3 (Randomized Crossover Window): Eligible participants who pass the diagnostic screen are randomized via sequential opaque envelopes into one of two intervention sequences (A-B or B-A). Allocation concealment is maintained by an independent unblinded study coordinator who programs the mobile application remotely, leaving the participant and data analyst blind to the track delivery. On one night, participants receive the experimental condition (AI-driven neuromodulation utilizing the NeuroRhythm algorithm to dynamically alter acoustic masking parameters based on real-time biometric feedback). On the alternate night, participants receive the sham condition (standard, non-adaptive acoustic music). Continuous PSG tracking runs concurrently both nights to allow epoch-by-epoch matrix synchronization between the earbud's internal sensor metrics and clinical hardware. Neurocognitive and Subjective Profiling On the mornings following Nights 1, 2, and 3, participants undergo standardized tracking procedures. Automated neurocognitive performance is mapped using the Cambridge Neuropsychological Test Automated Battery (CANTAB) touchscreen system to evaluate transient shifts in sustained attention, psychomotor alertness, and working memory efficiency linked to sleep structural changes. Subjective sleep depth, freshness, and hardware comfort metrics are gathered via morning clinical diaries. Phase 2: Naturalistic Home Extension and Endpoint (Days 4-14) Upon discharge from the sleep laboratory on Day 3, participants transition into a 14-day home-use window to assess the cumulative real-world utility of the intervention. Participants utilize the wearable earbud during sleep in their home environments according to their final laboratory sequence protocol assignment. Compliance, device tolerability, and subjective rest patterns are tracked daily via electronic logs. On Day 14, participants return for a final clinical endpoint visit to complete comprehensive psychometric re-evaluations, repeat the full-length neurocognitive battery (CANTAB), turn in all hardware, and execute the data collection confirmation logs.

Interventions

AI-driven Neuromodulation Music (NeuroRhythm)DEVICE

Acoustic stimulation delivered via the ANKER soundcore Sleep A40 Pro earbuds. This experimental condition utilizes the proprietary NeuroRhythm closed-loop algorithm to dynamically adjust music parameters (including timbre, tempo, white noise, and binaural beats) in real time. The acoustic adjustments are driven by the participant's live biological feedback (heart rate variability and motion data) captured by the earbud's embedded in-ear photoplethysmography (PPG) sensor.

Standard Music (Sham Control)DEVICE

Acoustic stimulation delivered via the identical ANKER soundcore Sleep A40 Pro earbuds. This control condition plays standard, non-adaptive relaxation music pre-selected to match the participant's baseline audio preferences. The acoustic properties remain completely static throughout the night and do not adjust or respond to any real-time biometric or physiological feedback from the user.

Eligibility

Sex: ALLMin age: 18 YearsMax age: 55 Years

Medical Language ↔ Plain English

Inclusion Criteria * Aged 18 to 55 years (inclusive). * Able to understand the study protocol and voluntarily sign the informed consent form. * Meets the diagnostic criteria for chronic insomnia disorder according to the International Classification of Sleep Disorders, Third Edition (ICSD-3) or DSM-5. * Duration of insomnia symptoms is greater than 3 months but less than 2 years. * Insomnia Severity Index (ISI) total score \> 8 (indicating mild or greater clinical insomnia). * Currently drug-free status: No use of any prescription or over-the-counter medications that affect sleep (including sedatives, hypnotics, antidepressants, antihistamines, or traditional Chinese medicine sleep aids) for at least 2 weeks (or more than 5 drug half-lives) prior to enrollment. * Owns a smartphone and is capable of operating mobile applications to complete electronic questionnaires. * Willing and able to tolerate wearing earplugs or earbuds during sleep. Exclusion Criteria * Comorbid primary sleep disorders: Diagnosed via screening (STOP-Bang questionnaire) or polysomnography (PSG) with moderate-to-severe obstructive sleep apnea (OSA, defined as Apnea-Hypopnea Index $\\ge$ 15), narcolepsy, periodic limb movement disorder (PLMD), or other sleep disorders that could interfere with sleep quality assessments. * Severe psychiatric or psychological conditions: A Patient Health Questionnaire (PHQ-9) score \> 10 or a Generalized Anxiety Disorder scale (GAD-7) score \> 10. * History of schizophrenia, bipolar disorder, major depressive disorder, or active suicidal ideation. * Active substance abuse or dependence: History of alcohol abuse (exceeding 14 standard drinks per week) or illicit drug use within the past 3 months. * Comorbid somatic medical conditions: Severe or unstable physical illnesses (e.g., severe heart failure, malignant tumors, chronic pain) or neurological disorders known to impair sleep monitoring accuracy. * Active ear diseases, ear canal discharge, structural abnormalities, or a history of related otologic surgeries that prevent or restrict earbud placement. * Known history of hypersensitivity or allergic reactions to silicone or plastic materials. * Disruptive lifestyle factors: Current engagement in shift work schedules or travel across more than 3 time zones within 2 weeks prior to study entry. * Inability to provide independent informed consent or successfully complete cognitive testing due to profound language or cognitive barriers.

Outcomes

Primary Outcomes

Change from Baseline in Polysomnography (PSG)-Measured Sleep Onset Latency (SOL)

The objective time, in minutes, from turning the lights off to the appearance of the first continuous epoch of sleep, as recorded by the mobile PSG system.

Time frame: Measured on Laboratory Night 1 (Day 2 morning) and Laboratory Night 2 (Day 3 morning).

Change from Baseline in Percentage of Slow Wave Sleep (N3 Stage)

The percentage of total sleep time spent in the N3 deep sleep stage (slow-wave sleep), derived from the objective PSG recordings scored according to AASM standards.

Time frame: Measured on Laboratory Night 1 (Day 2 morning) and Laboratory Night 2 (Day 3 morning).

Change from Baseline in Insomnia Severity Index (ISI) Score

The ISI is a 7-item self-report instrument assessing the nature, severity, and impact of insomnia. The total score ranges from 0 to 28, where 0-7 indicates no clinically significant insomnia and 22-28 indicates severe clinical insomnia. A reduction in score represents an improvement in insomnia severity.

Time frame: Baseline (Day 1), Post-Laboratory (Day 3), and Endpoint (Day 14).

Secondary Outcomes

Sleep Stage Classification Agreement (Cohen's Kappa)

The epoch-by-epoch classification agreement between the earbud's automated AI sleep staging algorithm and the manually scored gold-standard PSG across a 4-stage sleep model (Wake, Light, Deep, REM). Agreement is quantified using the Cohen's Kappa coefficient.

Time frame: Evaluated continuously across Laboratory Nights 1 and 2 (Days 2 and 3)

Sensor Signal Accuracy for Heart Rate Variability (HRV) - RMSSD Metric

The measurement accuracy of the earbud's photoplethysmography (PPG) sensor against the reference PSG electrocardiogram (ECG) channel. Accuracy is calculated using the intra-class correlation coefficient (ICC) of the Root Mean Square of Successive Differences (RMSSD) in milliseconds.

Time frame: Evaluated continuously across Laboratory Nights 1 and 2 (Days 2 and 3).

Change from Baseline in Psychomotor Vigilance Task (PVT) Reaction Time

Measured using the automated Cambridge Neuropsychological Test Automated Battery (CANTAB) on a touchscreen tablet. This metric captures the participant's median reaction time and lapses in milliseconds to evaluate sustained visual attention.

Time frame: Baseline (Day 1), Day 2 morning, Day 3 morning, and Follow-up Endpoint (Day 14).

Change from Baseline in Mood Symptoms (PHQ-9)

Evaluation of secondary emotional distress changes via the Patient Health Questionnaire (PHQ-9) for depression symptoms (score range 0 to 27). Higher scores indicate worse symptom severity.

Time frame: Baseline (Day 1) and Follow-up Endpoint (Day 14).

Change from Baseline in Anxiety Symptoms (GAD-7)

Evaluation of secondary emotional distress changes via Generalized Anxiety Disorder scale (GAD-7) for anxiety symptoms (score range 0 to 21). Higher scores indicate worse symptom severity.