Annotating sleep states in children from wrist-worn accelerometer data using Machine Learning

TL;DR

This paper explores machine learning techniques to automate and improve the accuracy of sleep state annotation in children using wrist-worn accelerometer data, aiming for scalable and precise sleep analysis.

Contribution

It introduces multiple ML models, including deep learning approaches, for sleep event detection and evaluates their performance with a novel EDAP score metric.

Findings

Support vector machines and ensemble methods show high accuracy.

Deep learning models like LSTMs outperform traditional ML techniques.

The proposed methods enable scalable sleep annotation in children.

Abstract

Sleep detection and annotation are crucial for researchers to understand sleep patterns, especially in children. With modern wrist-worn watches comprising built-in accelerometers, sleep logs can be collected. However, the annotation of these logs into distinct sleep events: onset and wakeup, proves to be challenging. These annotations must be automated, precise, and scalable. We propose to model the accelerometer data using different machine learning (ML) techniques such as support vectors, boosting, ensemble methods, and more complex approaches involving LSTMs and Region-based CNNs. Later, we aim to evaluate these approaches using the Event Detection Average Precision (EDAP) score (similar to the IOU metric) to eventually compare the predictive power and model performance.