Utilizing a Neck Piezoelectric Sensor and Deep Learning to Predict Severe Sleep Apnea in Snorers

Monday, September 30, 2024

10:00 AM – 10:06 AM EDT

Location: MBCC, 238-239

Disclosure(s):

Li-Ang Lee, MD, PhD:

faculty for this accredited education activity has no relationship(s) with ineligible companies to disclose.

Introduction: This study develops an innovative deep learning model that employs a neck-wearable piezoelectric sensor to detect severe sleep apnea in habitual snorers, aiming to mitigate the widespread issue of sleep apnea underdiagnosis in adults.

Methods: From May 2018 to October 2020, we conducted a prospective case series at a tertiary medical center in Taiwan, enrolling 60 adult habitual snorers. Each participant underwent polysomnography while wearing a neck piezoelectric sensor, which recorded carotid artery pulsations (2–10 Hz) and snoring vibrations (70–250 Hz). The initial dataset, derived from the first 20 participants, included 1,304 snoring, 1,167 non-snoring, and 399 noise samples. We then developed a hybrid deep learning model, integrating convolutional and recurrent neural networks, to identify apneas and hypopneas. The model’s performance in predicting severe sleep apnea was subsequently tested on the remaining 40 participants using a confusion matrix.

Results: Bland–Altman plot comparisons showed a high level of agreement between the sensor data and polysomnography, with rates of 96% for heartbeats, 93% for snoring events, and 87% for sleep periods. The model achieved a 92% accuracy in differentiating snores, non-snores, and noise. Notably, the detection of apneas and hypopneas was more precise in the 2nd and 3rd sleep periods than in the initial ones, with an improvement from 85% to 90% accuracy. The optimized algorithm showed outstanding efficacy in predicting severe sleep apnea, boasting 100% sensitivity, 79% specificity, and 90% accuracy.

Conclusions: Our results suggest that the combination of a neck-wearable piezoelectric sensor and deep learning analysis offers a promising solution for screening severe sleep apnea among habitual snorers. This approach could significantly enhance diagnostic accuracy in clinical settings.

Presenting Author(s)

LL

Li-Ang Lee, MD, PhD

Professor, Departments of Otorhinolaryngology-Head and Neck Surgery, Sleep Center
Linkou Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan
Taoyuan City, Taoyuan, Taiwan (Republic of China)

Disclosure(s):
faculty for this accredited education activity has no relationship(s) with ineligible companies to disclose.

Senior Author(s)

LL

Li-Ang Lee, MD, PhD

Professor, Departments of Otorhinolaryngology-Head and Neck Surgery, Sleep Center
Linkou Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan
Taoyuan City, Taoyuan, Taiwan (Republic of China)

Disclosure(s):
faculty for this accredited education activity has no relationship(s) with ineligible companies to disclose.

Co-Author(s)

YC

Yi-Ping Chao, PhD

Associate Professor, Department of Computer Science and Information Engineering
Chang Gung University, Taoyuan, Taiwan, United States

Disclosure information not submitted.
HC

Hai-Hua Chuang, MD

Associate Professor, Departments of Family Medicine
Chang Gung Memorial Hospital, Taipei and Linkou Branches, and Chang Gung University, Taoyuan, Taiwan, Taipei, Taiwan (Republic of China)

Disclosure information not submitted.
YL

Yu-Lun Lo, MD

Associate Professor, Departments of Pulmonary and Critical Care Medicine
Linkou Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan, United States

Disclosure information not submitted.
CH

Chung-Guei Huang, PhD

Associate Professor, Department of Laboratory Medicine
Linkou Chang Gung Memorial Hospital and Chang Gung University, Taoyuan, Taiwan, United States

Disclosure information not submitted.
LS

Liang-Yu Shyu, PhD

Professor, Department of Biomedical Engineering
Chung Yuan Christian University, Taoyuan, Taiwan, United States

Disclosure information not submitted.
HL

Hsueh-Yu Li, MD

Professor
Linkou-Chang Gung Memorial Hospital
Taoyuan City, Taoyuan, Taiwan (Republic of China)

Disclosure information not submitted.