Insomnia is the most prevalent sleep disorder in the US. In-home insomnia monitoring is important for both diagnosis and treatment. Existing solutions, however, require the user to either maintain a sleep diary or wear a sensor while sleeping. Both can be quite cumbersome. This paper introduces EZ-Sleep, a new approach for monitoring insomnia and sleep.
We focus on predicting sleep stages from radio measurements without any attached sensors on subjects. We introduce a new predictive model that combines convolutional and recurrent neural networks to extract sleep-specific subjectinvariant features from RF signals and capture the temporal progression of sleep. A key innovation underlying our approach is a modified adversarial training regime that discards extraneous information specific to individuals or measurement conditions, while retaining all information relevant to the predictive task. We analyze our game theoretic setup and empirically demonstrate that our model achieves significant improvements over state-of-the-art solutions.