SleepSensei: An Automated Sleep Quality Monitor and Sleep Duration Estimator

Abstract

SleepSensei is an automated sleep quality monitor and estimates the sleep duration for a user. It essentially builds a personalized model for determining the apt sleep duration for a given user. In our model, with the help of multiple inter-connected devices, we combine features related to user surroundings and those related to user movements during sleep. We use regression models in our system to calculate "sleep share" of a user, and report an MSE as low as 0.0041 when tested on 7 users, with more than a week's data. We also develop a smart alarm based on the notion of "sleep quota" from the model. The alarm has an average precision of 0.87 from the survey conducted. We also present an in-depth feature analysis to give further insights of individual sleep behavior.