Andrew Phillips

Sleep behavior significantly impacts health and acts as an indicator of physical and mental well-being. Monitoring and predicting sleep behavior with ubiquitous sensors may therefore assist in both sleep management and tracking of related health conditions. While sleep behavior depends on, and is reflected in the physiology of a person, it is also impacted by external factors such as digital media usage, social network contagion, and the surrounding weather. In this work, we propose SleepNet, a system that exploits social contagion in sleep behavior through graph networks and integrates it with physiological and phone data extracted from ubiquitous mobile and wearable devices for predicting next-day sleep labels about sleep duration. Our architecture overcomes the limitations of large-scale graphs containing connections irrelevant to sleep behavior by devising an attention mechanism. The extensive experimental evaluation highlights the improvement provided by incorporating social networks in the model. Additionally, we conduct robustness analysis to demonstrate the system's performance in real-life conditions. The outcomes affirm the stability of SleepNet against perturbations in input data. Further analyses emphasize the significance of network topology in prediction performance revealing that users with higher eigenvalue centrality are more vulnerable to data perturbations.

In humans, the nocturnal secretion of melatonin by the pineal gland is suppressed by ocular exposure to light. In the laboratory, melatonin suppression is a biomarker for this neuroendocrine pathway. Recent work has found that individuals differ substantially in their melatonin-suppressive response to light, with the most sensitive individuals being up to 60 times more sensitive than the least sensitive individuals. Planning experiments with melatonin suppression as an outcome needs to incorporate these individual differences, particularly in common resource-limited scenarios where running within-subjects studies at multiple light levels is costly and resource-intensive and may not be feasible with respect to participant compliance. Here, we present a novel framework for virtual laboratory melatonin suppression experiments, incorporating a Bayesian statistical model. We provide a Shiny web app for power analyses that allows users to modify various experimental parameters (sample size, individual-level heterogeneity, statistical significance threshold, light levels), and simulate a systematic shift in sensitivity (e.g., due to a pharmacological or other intervention). Our framework helps experimenters to design compelling and robust studies, offering novel insights into the underlying biological variability in melatonin suppression relevant for practical applications.