Electrodermal Activity Akane Sano and Rosalind W Picard Massachusetts Institute of Technology Media Lab Affective Computing Group akanesmitedu What is Electrodermal Activity Electrical measures of sweat gland ID: 293944
Download Presentation The PPT/PDF document "Toward a Taxonomy of Autonomic Sleep Pat..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Toward a Taxonomy of Autonomic Sleep Patterns with Electrodermal Activity
Akane
Sano and Rosalind W. Picard,
Massachusetts Institute of Technology
Media Lab
Affective Computing Group
akanes@mit.edu Slide2
What is
Electrodermal Activity?
Electrical measures of sweat gland
activity
Index of sympathetic nervous activationClassically, has been measured with wired and gelled electrodes on the skinOur research group developed a dry electrode, wearable sensor for long-term ambulatory measurementSlide3
Electrodermal Activity (EDA) during sleep
Q:Sympathetic nervous activity goes up during a day and goes down and get silent during sleep?
A: No!!!
High frequency “storm” patterns during sleep
Why these storm patterns happen?Slide4
Measurement of Sleep
Polysomnography (PSG)
+ measures EEG and more, provides 30 s epochs labeled as: Wake,
NonREM (stage 1-3), and REM - expensive and obtrusiveActigraphy + less invasive than PSG, low cost - only measures movementOur EDA sensor + comfortable, same or lower cost than
actigraphy
+ measures EDA, skin temperature and
actigraphy
-+ measures different patterns than traditionalSlide5
ObjectivesEvaluate
EDA sleep
patterns quantitatively from
healthy
groupsUnderstand what the changing patterns of EDA mean in terms of traditional PSG.
Experiments
Collected
EDA+motion
during
sleep from healthy adults
Total:
168
nightsSlide6
Analysis: sleep vs. wake
Sleep and wake are discriminated from accelerometer data with standard
zero-crossing and
Cole’s function
wakeSlide7
Analysis: EDA storms during sleep
After
low-pass filtering (0.4
Hz, 32nd order FIR filter
), we detected “storm” regions during sleep, regions of EDA with a burst of peaksStorm epoch: > 3
peaks /
30-sec with
the slope of each peak
>
0.09 micro Siemens/s Storm
: Storm epochs that are adjacent or within 5 minutes of each other Example: 6 storms in one night of sleep
*
* wake
EDA Storm
Raw EDASlide8
EDA vs. sleep stages from PSG
EDA raw data
Motion data
EDA peaks
Sleep Stage
Wake is redSlide9
More than 90 % of EDA Storms occurred in SWS and NREM2 (N=7, one night each)
One subject had storms below the
threshold
Portion of storm epochs in each category of sleep.Slide10
Histogram of # of storms over night (168 nights)
2/3 of nights had >= 1 storm
1/2 of nights had >= 2 stormsSlide11
SummaryWe analyzed electrodermal activity
from healthy subjects over
150 nights
More than 90 % of EDA storms occurred in SWS and NREM2 (N=7, one night each)
2/3 of nights showed more than 1 EDA storm 1/2 of nights had more than 2 storms Next StepsNeeds more detail analysis with EEG and heart ratesAre they related with sleep quality/ Sleep disorders or Memory consolidation?