4:00PM Integrated Sleep Stability: Dynamic Mapping
of Sleep Oscillations in Health and Disease
Robert Thomas, MD, Assistant Professor of
Medicine at Harvard Medical School and Beth Israel Deaconess Medical Center,
rthomas1@bidmc.harvard.edu
Moderator: Ary L. Goldberger, MD, Director, The Margret & H. A. Rey
Institute for Nonlinear Dynamics in Medicine (ReyLab), BIDMC; Program Director,
Research Resource for Complex Physiologic Signals,
agoldber@caregroup.harvard.edu
For a long
time, researchers have used electroencephalogram (EEG) data to measure sleep on
a discrete scale. The traditional
technique is to determine whether a person is experiencing REM (rapid eye
motion) sleep and to then fit non-REM sleep into one of various stages. Although this method works well for research
purposes, it doesn’t work very well in clinical practice because the vast
majority of natural sleep is classified as non-REM stage 2. The current classification system is also
imperfect because it does not capture the effects that sleep produces outside
the brain. In the lab of Robert Thomas,
researchers are attempting to address this problem by developing a new
sleep-measurement rubric that incorporates heart and respiration rate information.
Their method is based on the fact
that many signals from the sleeping body oscillate in a coupled fashion. This coupling can be grouped into three
categories: high-frequency coupling (HFC), low-frequency coupling (LFC), and
very-low-frequency coupling (VLFC). The
frequency of HFC is the same as that of respiration. Over the course of a night, the amount of
each type of coupling can be measured, and this information can be depicted on
a sleep spectrum. Dr. Thomas’s team
produced sleep spectra for a number of people with sleep apnea, and they found
that continuous positive airway pressure (CPAP), a treatment for sleep apnea,
caused these people to experience increased HFC. They also analyzed data from patients with
heart failure, and they found that these subjects experienced a tremendous
amount of LFC but very little HFC. Thus,
it seems that HFC can be used as a biomarker for healthy sleep. The sleep spectra produced by Dr. Thomas’s
lab seem to be “sleep thermometers” that might someday allow doctors to gain
insight into many aspects of a patient’s physiology. This approach challenges the current
understanding of sleep because it incorporates more than brain waves and
because it breaks from the discrete stages that are currently used to describe
sleep. View
this presentation
5:00PM Modeling the
Dynamics of Sleep Using State Space Analysis
Tom Scammell, MD, Associate Professor of
Neurology at Harvard Medical School and Beth Israel Deaconess Medical Center,
tscammel@bidmc.harvard.edu
Moderator: Clifford B. Saper, MD, PhD, James Jackson Putnam Professor of
Neurology, HMS; Chairman, Department of Neurology, BIDMC,
csaper@bidmc.harvard.edu
Narcolepsy is a common
disease that causes daytime sleepiness often in conjunction with hypnagogic
hallucinations (vivid dreams as one falls asleep), sleep paralysis (paralysis
upon waking up), and cataplexy (muscle weakness triggered by strong
emotions). In patients with the disease,
the brain has lost almost all the neurons responsible for producing the peptide
neurotransmitter orexin, also known as hypocretin. Mice engineered to lack orexin display many
of the symptoms of narcolepsy, transitioning very frequently between
wakefulness and sleepiness. Using discrete
state variables to describe the sleeping and waking states of these
“narcoleptic” mice does not capture the breadth of their behavior, so
researchers in the lab of Tom Scammell have come up with a continuous
two-dimensional state space upon which the sleepiness of the mice can be
measured.
In their
state space, the relative number of low-frequency waves is measured on one
axis, and the relative number of middle-frequency (theta) waves is depicted on
the other. One-second
electroencephalogram (EEG) recordings can be plotted in this state space, and
by sampling the sleep state of a mouse every second, researchers can trace the
mouse’s trajectory through the state space.
In the state space, REM sleep, non-REM sleep, and wakefulness occupied
distinct areas. Dr. Scammell’s team
found that the “wakefulness” and “non-REM” areas were closer for “narcoleptic”
mice than for regular mice. They also
found that the “narcoleptic” mice transitioned from one area to another more
quickly than the normal mice. These
observations suggest intriguing ideas about narcolepsy, and the “state space”
concept offers a promising new way to look at sleep.