Sleep awake detection from leg-worn wearables using deep sensor fusion.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Sleep Medicine

Background:

• Restful sleep is crucial for overall health, but children with Attention Deficit Hyperactivity Disorder (ADHD) frequently experience sleep disturbances.
• Current sleep monitoring tools like polysomnography and wrist-worn devices have significant limitations in cost, complexity, and accuracy for this population.

Purpose of the Study:

• To develop and validate a noninvasive deep learning-based system for accurate sleep monitoring in children with ADHD.
• To assess the feasibility of using a leg-worn multimodal wearable device (RestEaze) for capturing physiological and motion data.

Main Methods:

• Utilized a leg-worn wearable device collecting photoplethysmography (PPG), motion (accelerometer, gyroscope), and temperature data from 14 children referred for ADHD evaluation.
• Developed and compared a Support Vector Machine (SVM) baseline model with two deep learning models (CNN-BiLSTM) using early and late-fusion techniques on raw multimodal inputs.
• Implemented a temporal label-smoothing method to enhance the consistency of sleep-wake state classification.

Main Results:

• The late-fusion CNN-BiLSTM model achieved a high accuracy, with an area under the ROC curve of 90.94% in five-fold cross-validation.
• The system successfully derived key sleep metrics including total sleep time, wake after sleep onset, sleep onset latency, and awakenings.
• Demonstrated the effectiveness of multimodal data fusion and deep learning for robust sleep stage classification.

Conclusions:

• Leg-based multimodal sensing combined with deep learning presents a feasible and noninvasive approach for monitoring sleep in pediatric neurodevelopmental populations.
• This technology offers a potential improvement over existing methods for diagnosing and managing sleep issues in children with ADHD.
• Further research can explore the integration of these findings into routine clinical practice for enhanced pediatric care.