SlideShare une entreprise Scribd logo
1  sur  18
Télécharger pour lire hors ligne
Automatic Sleep Staging Using
State Machine-controlled Decision Trees
Syed Anas Imtiaz
Esther Rodriguez-Villegas
IEEE Engineering in Medicine and Biology Conference 2015
2
INTRODUCTION
 Sleep staging has been an active research area for over four decades
 Involves analysis of physiological signals and classification into one of
Wake, N1, N2, N3 or REM stages
 Typically uses ExG (EEG/EOG/EMG) signals
 Actigraphy, HRV and respiratory increasingly used now
 Recent consumer focus with sleep analysis integrated with wearables and
fitness trackers
IEEE Engineering in Medicine and Biology Conference 2015
3
…for sleep tracking
Wearable Devices
Actigraphy, accelerometry, heart rate
Difficult to distinguish between all stages
Over/under estimation of sleep time
Medical usage still questionable
May improve with time
Not as accurate as EEG
IEEE Engineering in Medicine and Biology Conference 2015
4
SLEEP STAGING RESEARCH
0
5
10
15
20
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Number of papers indexed in IEEEXplore over the last 25 years
IEEE Engineering in Medicine and Biology Conference 2015
5
SLEEP STAGING RESEARCH
 Most algorithms have very good classification accuracy
 However, this performance comes at a cost
 COMPUTATIONAL COMPLEXITY: ANN, SVM, Large number of features, etc.
 On hardware this translates to higher POWER CONSUMPTION
 Limited power budget in wearable systems
IEEE Engineering in Medicine and Biology Conference 2015
6
OUR AIM
To develop an automatic sleep staging algorithm that is
suitable for use in resource-constrained wearable systems“
 Extract spectral features and classify using a decision tree (DT)
 DTs are amongst the simplest of classifiers
 Use small contextually-aware DTs rather than one long tree
 Order of execution controlled by a state machine
IEEE Engineering in Medicine and Biology Conference 2015
7
DATABASE
 PhysioNet Sleep EDF Expanded Database
 61 PSG recordings
 Almost 60,000 epochs of 30-second duration
 31 recordings in training set and 30 in test set
 Two EEG channels: Fpz-Cz and Pz-Oz
IEEE Engineering in Medicine and Biology Conference 2015
8
FEATURES
Channel Features
Fpz-Cz
sigma/beta, beta/delta, delta/alpha, beta/alpha, SEFd(8-16 Hz),
SEFd(0.5-8 Hz), SEF95(0.5-30 Hz), SEF50(0.5-8 Hz), line length
(11-16 Hz), rel. delta2, rel. beta, rel. gamma, abs. delta, abs.
delta1, abs. delta2, abs. alpha2
Pz-Oz
sigma/beta, beta/delta, theta/alpha, beta/alpha, SEF95(0.5-30
Hz), SEF50(0.5-8 Hz), rel. beta, rel. gamma, rel. alpha, rel. theta,
abs. delta, abs. delta1, abs. alpha1
 66 features initially
 Each feature for an epoch is an average for fifteen 2-second subepochs
 Top 30 features using sequential feature selection (SFS)
IEEE Engineering in Medicine and Biology Conference 2015
9
DESIGN OF DECISION TREES
Decision trees, in comparison to more complex classifiers,
are only beneficial when the number of nodes in tree is
relatively small
“
 Use features with a decision tree
 But limiting the number of nodes can directly affect the performance
 Trade-offs needed
 Alternative approach?
IEEE Engineering in Medicine and Biology Conference 2015
10
OUR APPROACH
When a certain sleep stage is prevalent, it normally stays
for a few epochs before transitioning to the next stage.“
 An epoch only needs to be tested to check whether it is of the same sleep stage or not
 One-vs-all decision tree
 If transition needed, use a series of one-versus-one decision trees
 Order of execution based on the likelihood of next sleep stage
 Combination of decision trees contextually driven by a state machine
 Activated based on the current sleep stage
 Approach commonly used in AI for game development
IEEE Engineering in Medicine and Biology Conference 2015
11
HOW IT WORKS?
Initial State
Check if the new epoch
is also Wake or not
If Wake, maintain
state and return
Assign Wake
if all tests fail
Test for other sleep
stages and exit when
different stage found
IEEE Engineering in Medicine and Biology Conference 2015
12
HOW IT WORKS?
New State
Check if the new epoch
is also N2 or not
If N2, maintain
state and return
Assign N2 if
all tests fail
Test for other sleep
stages and exit when
different stage found
13
RESULTS
IEEE Engineering in Medicine and Biology Conference 2015
14
Using 30 PSG recordings for PhysioNet EDF Expanded Database
TRAINING RESULTS
WAKE 84.3%
N1 29.8%
N2 88.5%
N3 81.8%
Of the 29499 epochs in this set,
24255 are correctly classified
resulting in an overall accuracy of
82.22%
All stages show sensitivity of over
80% except N1
REM 87.4%
IEEE Engineering in Medicine and Biology Conference 2015
15
Using 31 PSG recordings for PhysioNet EDF Expanded Database
TEST RESULTS
WAKE 72.1%
N1 22.0%
N2 88.2%
N3 79.6%
Of the 29817epochs in this set,
23512 are correctly classified
resulting in an overall accuracy of
78.85%
Similar results in N2, N3 and REM
and a noticeable reduction in
sensitivity for Wake an N1
REM 84.1%
IEEE Engineering in Medicine and Biology Conference 2015
16
DISCUSSION
 About 80% classification performed by the Core decision trees.
 Average number of comparisons for a decision is 3.4 (longest 12, shortest 2).
 Total number of comparisons reduced by over 40% compared to a traditional decision
tree for similar classification performance.
 Only a subset of features and trees needed in each state resulting in better usage of
processor resources consequently reducing power consumption.
IEEE Engineering in Medicine and Biology Conference 2015
17
FUTURE WORK
Search for more characteristic features
with better discriminatory ability and
potentially reduce number of features.
BETTER FEATURES
Implement the algorithm on low-power
microcontroller and/or integrated circuit
to measure actual power consumption.
HARDWARE IMPLEMENTATION
The peripheral decision trees to discriminate
between two stages can be designed with
different weights assigned.
IMPROVED DESIGN OF TREES
Check to see if algorithm is useful with
recordings from others sources and
ways of improving performance.
MORE TESTING
Q u e s t i o n s / C o m m e n t s
Thank You

Contenu connexe

Tendances

Classification of Pre-ictal and Interictal Periods Based
Classification of Pre-ictal and Interictal Periods BasedClassification of Pre-ictal and Interictal Periods Based
Classification of Pre-ictal and Interictal Periods Based
Bharat Kumar Reddy Karumuri
 
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
Victor Asanza
 
Noise analysis & qrs detection in ecg signals
Noise analysis & qrs detection in ecg signalsNoise analysis & qrs detection in ecg signals
Noise analysis & qrs detection in ecg signals
Harshal Ladhe
 

Tendances (15)

Classification of Pre-ictal and Interictal Periods Based
Classification of Pre-ictal and Interictal Periods BasedClassification of Pre-ictal and Interictal Periods Based
Classification of Pre-ictal and Interictal Periods Based
 
Telemetry 101: Exploring the New ADInstruments’ Small Animal Telemetry Systems
Telemetry 101: Exploring the New ADInstruments’ Small Animal Telemetry SystemsTelemetry 101: Exploring the New ADInstruments’ Small Animal Telemetry Systems
Telemetry 101: Exploring the New ADInstruments’ Small Animal Telemetry Systems
 
⭐⭐⭐⭐⭐ SSVEP-EEG Signal Classification based on Emotiv EPOC BCI and Raspberry Pi
⭐⭐⭐⭐⭐ SSVEP-EEG Signal Classification based on Emotiv EPOC BCI and Raspberry Pi⭐⭐⭐⭐⭐ SSVEP-EEG Signal Classification based on Emotiv EPOC BCI and Raspberry Pi
⭐⭐⭐⭐⭐ SSVEP-EEG Signal Classification based on Emotiv EPOC BCI and Raspberry Pi
 
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
⭐⭐⭐⭐⭐ 2020 #IEEE #IES #UPS #Cuenca: Clasificación de señales de Electroencefa...
 
Wireless Recording and Stimulation Technologies for in vivo Electrophysiology...
Wireless Recording and Stimulation Technologies for in vivo Electrophysiology...Wireless Recording and Stimulation Technologies for in vivo Electrophysiology...
Wireless Recording and Stimulation Technologies for in vivo Electrophysiology...
 
Rigel 288
Rigel 288Rigel 288
Rigel 288
 
IRJET - FPGA based Electrocardiogram (ECG) Signal Analysis using Linear Phase...
IRJET - FPGA based Electrocardiogram (ECG) Signal Analysis using Linear Phase...IRJET - FPGA based Electrocardiogram (ECG) Signal Analysis using Linear Phase...
IRJET - FPGA based Electrocardiogram (ECG) Signal Analysis using Linear Phase...
 
Rigel 377
Rigel 377Rigel 377
Rigel 377
 
⭐⭐⭐⭐⭐ IX Jornadas Académicas y I Congreso Científico de Ciencias e Ingeniería...
⭐⭐⭐⭐⭐ IX Jornadas Académicas y I Congreso Científico de Ciencias e Ingeniería...⭐⭐⭐⭐⭐ IX Jornadas Académicas y I Congreso Científico de Ciencias e Ingeniería...
⭐⭐⭐⭐⭐ IX Jornadas Académicas y I Congreso Científico de Ciencias e Ingeniería...
 
⭐⭐⭐⭐⭐ #FPGA Based Meteorological Monitoring Station
⭐⭐⭐⭐⭐ #FPGA Based Meteorological Monitoring Station⭐⭐⭐⭐⭐ #FPGA Based Meteorological Monitoring Station
⭐⭐⭐⭐⭐ #FPGA Based Meteorological Monitoring Station
 
IRJET-Advanced Method of Epileptic detection using EEG by Wavelet Decomposition
IRJET-Advanced Method of Epileptic detection using EEG by Wavelet DecompositionIRJET-Advanced Method of Epileptic detection using EEG by Wavelet Decomposition
IRJET-Advanced Method of Epileptic detection using EEG by Wavelet Decomposition
 
⭐⭐⭐⭐⭐ LECCIÓN SISTEMAS EMBEBIDOS, 2do Parcial (2020 PAO 1) C6
⭐⭐⭐⭐⭐ LECCIÓN SISTEMAS EMBEBIDOS, 2do Parcial (2020 PAO 1)  C6⭐⭐⭐⭐⭐ LECCIÓN SISTEMAS EMBEBIDOS, 2do Parcial (2020 PAO 1)  C6
⭐⭐⭐⭐⭐ LECCIÓN SISTEMAS EMBEBIDOS, 2do Parcial (2020 PAO 1) C6
 
Classification and Detection of ECG-signals using Artificial Neural Networks
Classification and Detection of ECG-signals using Artificial Neural NetworksClassification and Detection of ECG-signals using Artificial Neural Networks
Classification and Detection of ECG-signals using Artificial Neural Networks
 
Epileptic Seizure Detection using An EEG Sensor
Epileptic Seizure Detection using An EEG SensorEpileptic Seizure Detection using An EEG Sensor
Epileptic Seizure Detection using An EEG Sensor
 
Noise analysis & qrs detection in ecg signals
Noise analysis & qrs detection in ecg signalsNoise analysis & qrs detection in ecg signals
Noise analysis & qrs detection in ecg signals
 

Similaire à Automatic Sleep Staging Using State Machine-controlled Decision Trees

Final presentation dwi riyono
Final presentation dwi riyonoFinal presentation dwi riyono
Final presentation dwi riyono
Dwi Riyono
 
The second seminar
The second seminarThe second seminar
The second seminar
AhmedMahany
 
Matthew Gray Summer 2015 Presentation
Matthew Gray Summer 2015 PresentationMatthew Gray Summer 2015 Presentation
Matthew Gray Summer 2015 Presentation
Matthew Gray
 
Services for preclinical_and_clinical_trials
Services for preclinical_and_clinical_trialsServices for preclinical_and_clinical_trials
Services for preclinical_and_clinical_trials
TheSiestaGroup
 
Intelligent Controller Design for a Chemical Process
Intelligent Controller Design for a Chemical ProcessIntelligent Controller Design for a Chemical Process
Intelligent Controller Design for a Chemical Process
CSCJournals
 

Similaire à Automatic Sleep Staging Using State Machine-controlled Decision Trees (20)

Final presentation dwi riyono
Final presentation dwi riyonoFinal presentation dwi riyono
Final presentation dwi riyono
 
PMED Transition Workshop - Machine Learning Methods to Learn Improved Electro...
PMED Transition Workshop - Machine Learning Methods to Learn Improved Electro...PMED Transition Workshop - Machine Learning Methods to Learn Improved Electro...
PMED Transition Workshop - Machine Learning Methods to Learn Improved Electro...
 
Wavelet-Based Approach for Automatic Seizure Detection Using EEG Signals
Wavelet-Based Approach for Automatic Seizure Detection Using EEG SignalsWavelet-Based Approach for Automatic Seizure Detection Using EEG Signals
Wavelet-Based Approach for Automatic Seizure Detection Using EEG Signals
 
Ijetcas14 323
Ijetcas14 323Ijetcas14 323
Ijetcas14 323
 
The second seminar
The second seminarThe second seminar
The second seminar
 
A hybrid classification model for eeg signals
A hybrid classification model for  eeg signals A hybrid classification model for  eeg signals
A hybrid classification model for eeg signals
 
EEG course.pptx
EEG course.pptxEEG course.pptx
EEG course.pptx
 
Matthew Gray Summer 2015 Presentation
Matthew Gray Summer 2015 PresentationMatthew Gray Summer 2015 Presentation
Matthew Gray Summer 2015 Presentation
 
Services for preclinical_and_clinical_trials
Services for preclinical_and_clinical_trialsServices for preclinical_and_clinical_trials
Services for preclinical_and_clinical_trials
 
Survey analysis for optimization algorithms applied to electroencephalogram
Survey analysis for optimization algorithms applied to electroencephalogramSurvey analysis for optimization algorithms applied to electroencephalogram
Survey analysis for optimization algorithms applied to electroencephalogram
 
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
 
Quantification and Comparison of Different Non-Linear Techniques of EEG Signals
Quantification and Comparison of Different Non-Linear Techniques of EEG SignalsQuantification and Comparison of Different Non-Linear Techniques of EEG Signals
Quantification and Comparison of Different Non-Linear Techniques of EEG Signals
 
Ns presentation 2016.7 v2 d (eeg, emg, nbetter)
Ns presentation 2016.7 v2 d (eeg, emg, nbetter)Ns presentation 2016.7 v2 d (eeg, emg, nbetter)
Ns presentation 2016.7 v2 d (eeg, emg, nbetter)
 
Intelligent Controller Design for a Chemical Process
Intelligent Controller Design for a Chemical ProcessIntelligent Controller Design for a Chemical Process
Intelligent Controller Design for a Chemical Process
 
IRJET-A Survey on Effect of Meditation on Attention Level Using EEG
IRJET-A Survey on Effect of Meditation on Attention Level Using EEGIRJET-A Survey on Effect of Meditation on Attention Level Using EEG
IRJET-A Survey on Effect of Meditation on Attention Level Using EEG
 
Smart Home for Paralyzed Aid
Smart Home for Paralyzed AidSmart Home for Paralyzed Aid
Smart Home for Paralyzed Aid
 
Taugagreining hf
Taugagreining hfTaugagreining hf
Taugagreining hf
 
Taugagreining hf 2003
Taugagreining hf 2003Taugagreining hf 2003
Taugagreining hf 2003
 
40220130405010 2-3
40220130405010 2-340220130405010 2-3
40220130405010 2-3
 
Ffeature extraction of epilepsy eeg using discrete wavelet transform
Ffeature extraction of epilepsy eeg  using discrete wavelet transformFfeature extraction of epilepsy eeg  using discrete wavelet transform
Ffeature extraction of epilepsy eeg using discrete wavelet transform
 

Dernier

Integrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - NeometrixIntegrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - Neometrix
Neometrix_Engineering_Pvt_Ltd
 
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
dollysharma2066
 
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
dharasingh5698
 
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Kandungan 087776558899
 

Dernier (20)

University management System project report..pdf
University management System project report..pdfUniversity management System project report..pdf
University management System project report..pdf
 
Hostel management system project report..pdf
Hostel management system project report..pdfHostel management system project report..pdf
Hostel management system project report..pdf
 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torque
 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
 
Block diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.pptBlock diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.ppt
 
Minimum and Maximum Modes of microprocessor 8086
Minimum and Maximum Modes of microprocessor 8086Minimum and Maximum Modes of microprocessor 8086
Minimum and Maximum Modes of microprocessor 8086
 
Integrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - NeometrixIntegrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - Neometrix
 
22-prompt engineering noted slide shown.pdf
22-prompt engineering noted slide shown.pdf22-prompt engineering noted slide shown.pdf
22-prompt engineering noted slide shown.pdf
 
Thermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - VThermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - V
 
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Ramesh Nagar Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
 
Introduction to Serverless with AWS Lambda
Introduction to Serverless with AWS LambdaIntroduction to Serverless with AWS Lambda
Introduction to Serverless with AWS Lambda
 
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
 
Thermal Engineering Unit - I & II . ppt
Thermal Engineering  Unit - I & II . pptThermal Engineering  Unit - I & II . ppt
Thermal Engineering Unit - I & II . ppt
 
Employee leave management system project.
Employee leave management system project.Employee leave management system project.
Employee leave management system project.
 
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
 
Hazard Identification (HAZID) vs. Hazard and Operability (HAZOP): A Comparati...
Hazard Identification (HAZID) vs. Hazard and Operability (HAZOP): A Comparati...Hazard Identification (HAZID) vs. Hazard and Operability (HAZOP): A Comparati...
Hazard Identification (HAZID) vs. Hazard and Operability (HAZOP): A Comparati...
 
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
 
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort ServiceCall Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
Call Girls in Netaji Nagar, Delhi 💯 Call Us 🔝9953056974 🔝 Escort Service
 
Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
 

Automatic Sleep Staging Using State Machine-controlled Decision Trees

  • 1. Automatic Sleep Staging Using State Machine-controlled Decision Trees Syed Anas Imtiaz Esther Rodriguez-Villegas
  • 2. IEEE Engineering in Medicine and Biology Conference 2015 2 INTRODUCTION  Sleep staging has been an active research area for over four decades  Involves analysis of physiological signals and classification into one of Wake, N1, N2, N3 or REM stages  Typically uses ExG (EEG/EOG/EMG) signals  Actigraphy, HRV and respiratory increasingly used now  Recent consumer focus with sleep analysis integrated with wearables and fitness trackers
  • 3. IEEE Engineering in Medicine and Biology Conference 2015 3 …for sleep tracking Wearable Devices Actigraphy, accelerometry, heart rate Difficult to distinguish between all stages Over/under estimation of sleep time Medical usage still questionable May improve with time Not as accurate as EEG
  • 4. IEEE Engineering in Medicine and Biology Conference 2015 4 SLEEP STAGING RESEARCH 0 5 10 15 20 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Number of papers indexed in IEEEXplore over the last 25 years
  • 5. IEEE Engineering in Medicine and Biology Conference 2015 5 SLEEP STAGING RESEARCH  Most algorithms have very good classification accuracy  However, this performance comes at a cost  COMPUTATIONAL COMPLEXITY: ANN, SVM, Large number of features, etc.  On hardware this translates to higher POWER CONSUMPTION  Limited power budget in wearable systems
  • 6. IEEE Engineering in Medicine and Biology Conference 2015 6 OUR AIM To develop an automatic sleep staging algorithm that is suitable for use in resource-constrained wearable systems“  Extract spectral features and classify using a decision tree (DT)  DTs are amongst the simplest of classifiers  Use small contextually-aware DTs rather than one long tree  Order of execution controlled by a state machine
  • 7. IEEE Engineering in Medicine and Biology Conference 2015 7 DATABASE  PhysioNet Sleep EDF Expanded Database  61 PSG recordings  Almost 60,000 epochs of 30-second duration  31 recordings in training set and 30 in test set  Two EEG channels: Fpz-Cz and Pz-Oz
  • 8. IEEE Engineering in Medicine and Biology Conference 2015 8 FEATURES Channel Features Fpz-Cz sigma/beta, beta/delta, delta/alpha, beta/alpha, SEFd(8-16 Hz), SEFd(0.5-8 Hz), SEF95(0.5-30 Hz), SEF50(0.5-8 Hz), line length (11-16 Hz), rel. delta2, rel. beta, rel. gamma, abs. delta, abs. delta1, abs. delta2, abs. alpha2 Pz-Oz sigma/beta, beta/delta, theta/alpha, beta/alpha, SEF95(0.5-30 Hz), SEF50(0.5-8 Hz), rel. beta, rel. gamma, rel. alpha, rel. theta, abs. delta, abs. delta1, abs. alpha1  66 features initially  Each feature for an epoch is an average for fifteen 2-second subepochs  Top 30 features using sequential feature selection (SFS)
  • 9. IEEE Engineering in Medicine and Biology Conference 2015 9 DESIGN OF DECISION TREES Decision trees, in comparison to more complex classifiers, are only beneficial when the number of nodes in tree is relatively small “  Use features with a decision tree  But limiting the number of nodes can directly affect the performance  Trade-offs needed  Alternative approach?
  • 10. IEEE Engineering in Medicine and Biology Conference 2015 10 OUR APPROACH When a certain sleep stage is prevalent, it normally stays for a few epochs before transitioning to the next stage.“  An epoch only needs to be tested to check whether it is of the same sleep stage or not  One-vs-all decision tree  If transition needed, use a series of one-versus-one decision trees  Order of execution based on the likelihood of next sleep stage  Combination of decision trees contextually driven by a state machine  Activated based on the current sleep stage  Approach commonly used in AI for game development
  • 11. IEEE Engineering in Medicine and Biology Conference 2015 11 HOW IT WORKS? Initial State Check if the new epoch is also Wake or not If Wake, maintain state and return Assign Wake if all tests fail Test for other sleep stages and exit when different stage found
  • 12. IEEE Engineering in Medicine and Biology Conference 2015 12 HOW IT WORKS? New State Check if the new epoch is also N2 or not If N2, maintain state and return Assign N2 if all tests fail Test for other sleep stages and exit when different stage found
  • 14. IEEE Engineering in Medicine and Biology Conference 2015 14 Using 30 PSG recordings for PhysioNet EDF Expanded Database TRAINING RESULTS WAKE 84.3% N1 29.8% N2 88.5% N3 81.8% Of the 29499 epochs in this set, 24255 are correctly classified resulting in an overall accuracy of 82.22% All stages show sensitivity of over 80% except N1 REM 87.4%
  • 15. IEEE Engineering in Medicine and Biology Conference 2015 15 Using 31 PSG recordings for PhysioNet EDF Expanded Database TEST RESULTS WAKE 72.1% N1 22.0% N2 88.2% N3 79.6% Of the 29817epochs in this set, 23512 are correctly classified resulting in an overall accuracy of 78.85% Similar results in N2, N3 and REM and a noticeable reduction in sensitivity for Wake an N1 REM 84.1%
  • 16. IEEE Engineering in Medicine and Biology Conference 2015 16 DISCUSSION  About 80% classification performed by the Core decision trees.  Average number of comparisons for a decision is 3.4 (longest 12, shortest 2).  Total number of comparisons reduced by over 40% compared to a traditional decision tree for similar classification performance.  Only a subset of features and trees needed in each state resulting in better usage of processor resources consequently reducing power consumption.
  • 17. IEEE Engineering in Medicine and Biology Conference 2015 17 FUTURE WORK Search for more characteristic features with better discriminatory ability and potentially reduce number of features. BETTER FEATURES Implement the algorithm on low-power microcontroller and/or integrated circuit to measure actual power consumption. HARDWARE IMPLEMENTATION The peripheral decision trees to discriminate between two stages can be designed with different weights assigned. IMPROVED DESIGN OF TREES Check to see if algorithm is useful with recordings from others sources and ways of improving performance. MORE TESTING
  • 18. Q u e s t i o n s / C o m m e n t s Thank You