Authors:
Chiying Wang
1
;
Sergio A. Alvarez
2
;
Carolina Ruiz
1
and
Majaz Moonis
3
Affiliations:
1
Worcester Polytechnic Institute, United States
;
2
Boston College, United States
;
3
U. of Massachusetts Medical School, United States
Keyword(s):
Markov Chain, Semi-Markov Chain, Sleep Stage Dynamics, Weibull Distribution.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Cardiovascular Technologies
;
Computing and Telecommunications in Cardiology
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Enterprise Information Systems
;
Health Engineering and Technology Applications
;
Health Information Systems
;
Medical and Nursing Informatics
;
Pattern Recognition and Machine Learning
;
Physiological Modeling
;
Sensor Networks
;
Signal Processing
;
Soft Computing
Abstract:
In this paper, a semi-Markov chain of sleep stages is considered as a model of human sleep dynamics. Both sleep stage transitions and the durations of continuous bouts in each stage are taken into account. The semi-Markov chain comprises an underlying Markov chain that models the temporal sequence of sleep stages but not the timing details, together with a separate statistical model of the bout durations in each stage. The stage bout durations are modeled explicitly, by the Weibull parametric family of probability distributions. This family is found to provide good fits for the durations of waking bouts and of bouts in the NREM and REM sleep stages. A collection of 244 all-night hypnograms is used for parameter optimization of the Weibull bout duration distributions for specific stages. The Weibull semi-Markov chain model proposed in this paper improves considerably on standard Markov chain models, which force geometrically distributed (discrete
exponential) stage bout durations for
all stages, contradicting known experimental observations. Our results provide more realistic dynamical modeling of sleep stage dynamics that can be expected to facilitate the discovery of interesting and useful dynamical patterns in human sleep data in future work.
(More)