Smart bed based daytime behavior prediction in Children with autism spectrum disorder - A Pilot Study

Area of Science:

• Biomedical Engineering
• Neuroscience
• Pediatric Sleep Medicine

Background:

• Sleep disturbances are common in children with autism spectrum disorder (ASD), impacting their daytime behavior.
• Polysomnography (PSG), the gold standard for sleep assessment, is often unsuitable for children with ASD due to its intrusive nature.

Purpose of the Study:

• To evaluate an unobtrusive, in-home smart bed system for long-term sleep quality monitoring in children with ASD.
• To assess the utility of ballistocardiogram (BCG) signals for estimating sleep quality and predicting daytime behaviors.

Main Methods:

• Developed an inexpensive smart bed system utilizing ballistocardiogram (BCG) signals for sleep monitoring.
• Defined "restlessness" from BCG signals as a surrogate sleep quality estimator.
• Employed supervised machine learning algorithms, Support Vector Machine (SVM) and Artificial Neural Network (ANN), to predict daytime behaviors based on 1-8 nights of sleep data.

Main Results:

• Achieved over 78% accuracy in predicting behavioral issues using SVM and over 79% using ANN.
• Demonstrated the effectiveness of the "restlessness" feature in improving prediction performance.
• Validated the smart bed system's utility for in-home, long-term sleep monitoring.

Conclusions:

• The designed smart bed system offers a practical and non-intrusive method for monitoring sleep in children with ASD.
• Sleep quality, as estimated by BCG-derived restlessness, significantly correlates with and predicts daytime behaviors.
• This technology holds promise for improving interventions and support for children with ASD.