Discovery of Patterns in Sleep Data

This research project aims to develop machine learning techniques for automated discovery of meaningful patterns in human sleep data. Our work to date has yielded an association mining approach for exploratory analysis of sleep data (Laxminarayan et al, 2005), including tight bounds on the false discovery rate (Laxminarayan et al, 2006), construction of a terabyte-scale database of anonymized polysomnographic recordings and health histories, and discovery of naturally occurring subgroups of sleep studies, or "sleep types", based on the stage composition of sleep (Khasawneh et al, 2010 and 2011).

Sleep Dynamics

Ongoing work includes modeling of the dynamics of sleep. An initial step in this direction is presented in (Usher et al, 2012). A major challenge to dynamical modeling of sleep is the scarcity of dynamical stage-transition events within a single all-night sleep recording for a given person. The paper (Alvarez and Ruiz, 2013) presents a general algorithmic framework for dynamical modeling of time series data (sleep in particular) that contain limited dynamical information. The approach, Collective Dynamical Modeling-Clustering (CDMC), operates on a collection of sleep recordings for a population of persons, simultaneously grouping them by dynamical similarity and developing a dynamical model for each similarity group. Selective aggregation of multiple records in this way provides an increased sample size for dynamical modeling that reduces random variation in the resulting models.

Publications (asterisks denote student co-authors)

Project Personnel


Current Students

Former group members