Automated Sleep

Disorder Presentation By:

Detection Using • Praneeth – 204135
Machine • Sai Bharadwaj – 204170
Learning • Uday Kiran - 204174
Techniques
Final Year Major Project Presentation
B.Tech IV Year in ECE
Project Guide : Prof. B. Lakshmi
Batch : A19
Section: A
 Problem Statement

SLEEP DISORDERS ARE MEDICAL CONDITIONS THAT THE GOAL IS TO DEVELOP AN AUTOMATED SYSTEM FOR
NEGATIVELY AFFECT AN INDIVIDUAL'S REGULAR SLEEPING SLEEP DISORDER DETECTION THAT CAN OVERCOME THESE
PATTERN AND ROUTINE, LEADING TO DETRIMENTAL EFFECTS LIMITATIONS. THE SYSTEM SHOULD BE ABLE TO IDENTIFY
ON HEALTH. TRADITIONAL PROCEDURES FOR IDENTIFYING VARIOUS SLEEP DISORDERS SUCH AS INSOMNIA,
SLEEP DISORDERS INVOLVE QUESTIONNAIRES AND NOCTURNAL FRONTAL LOBE EPILEPSY (NFLE), NARCOLEPSY,
POLYSOMNOGRAPHY (PSG), WHICH ARE SUBJECTIVE, TIME- RAPID EYE MOVEMENT BEHAVIOR DISORDER (RBD),
CONSUMING, AND INCONVENIENT. MOREOVER, THESE PERIODIC LEG MOVEMENT DISORDER (PLM), AND SLEEP-
PROCEDURES ARE PRONE TO HUMAN ERRORS DUE TO THE DISORDERED BREATHING (SDB) USING
REQUIREMENT OF LONG HOURS OF CONTINUOUS ELECTROENCEPHALOGRAM (EEG) SIGNALS.
EVALUATION.
 Motivation

Sleep is a period of rest that is essential for functional learning ability, mental
health, and even the performance of normal activities. Without prompt
diagnosis, the sleep disorders can affect the quality the life significantly. Hence,
it is very critical to identify the sleep disorder in early stages to undergo proper
treatment in order to cure it. The aim is to develop a sleep disorder detection
system which has the potential to significantly improve the quality of life for
individuals suffering from sleep disorders, reduce healthcare costs associated
with undiagnosed conditions, and contribute to overall public health.
 Advantages
• The major advantage of this proposed system is that it can overcome the limitations of manual sleep
stage classification, which is a difficult, time-consuming, and subjective process that relies on visual
inspection by experts.

• It can utilize the full potential of EEG signals, which are the most important and informative signals
for sleep analysis, without losing information during feature extraction and training.

• It can achieve high accuracy, robustness, and efficiency for sleep stage classification, which can support
sleep experts and clinicians in diagnosing sleep disorders, identifying their underlying mechanisms,
selecting appropriate treatments, and predicting their outcomes.
 Disadvantages
• The proposed system requires a large amount of labeled data for training and testing, which can be
costly, time-consuming, and challenging to obtain.

• It may not be able to capture the complexity and variability of sleep patterns and disorders, especially
for patients with comorbidities or rare conditions.

• It may not be able to explain the rationale behind its predictions, which can affect the trust and
acceptance of the system by experts and users.
 Applications
• This proposed system can be used provide a practical and fast diagnostic process for patients and
doctors, by reducing the need for manual scoring and interpretation of polysomnogram (PSG)
recordings.

• It can enable self-application of sleep analysis in home settings, by using a single channel of EEG
signals acquired with an automated sleep staging device.

• It can support driver safety and alertness, by using EEG features and deep learning approaches to
detect driver drowsiness and provide timely feedback or intervention.
 Literature Survey
Application of various machine learning techniques to predict obstructive sleep apnea syndrome severity
[1]

• In this study, data used were collected from 4014 patients who visited the sleep clinic of Samsung Medical
Center between 2014 and 2021.

• For classification, the authors used gradient boost-based models such as XGBoost, LightGBM, CatBoost, and
Random Forest to predict the severity of OSAS.

• The developed model showed high performance with 88%, 88%, and 91% of classification accuracy for three
thresholds for the severity of OSAS: Apnea-Hypopnea Index (AHI) ≥ 5, AHI ≥ 15, and AHI ≥ 30,
respectively.
A Systematic Review of Literature on Automated Sleep Scoring [2]

• This study outlines a Systematic Literature Review and the results it provided to assess the present state-of-
the-art in automatic analysis of sleep data.

• The American Academy of Sleep Medicine and Rechtschaffen & Kales are the two main scoring standards
used in contemporary research, according to the report.

• The current study’s major purpose was to assess state-ofthe-art in sleep scoring standards, bio-electric signal
feature extraction methodologies, and sleep data classification using machine learning approaches.

• As a result, this study indicates that using an alternate sleep scoring standard, as well as numerous feature
extraction with selection approaches, machine learning algorithms with parameter tweaking, and big data
analytics, physicians can produce a practically useful automatic sleep scoring system.
Sleep’s depth detection using electroencephalogram signal processing and neural network classification[3]

• In this model, authors suggest an automatic sleep stages classification approach based totally on a neural
community, blended with a wavelet-based function extraction method.

• In this work 4 sleep ranges will be classified, consisting of Awake, Stage 1 + rapid eye movement (REM),
Stage two and Slow Wave Stage primarily based only on the EEG signal.

• In order to validate the efficiency of this algorithm, they used PhysioNet database EEG signals for in
vitro tests.

• They used discrete wavelet transform method for EEG signal processing.

• This model has achieved an accuracy of 97.2%, 90.9%, 88.5% and 96.4% in classifying awake, stage 1, stage
2 and slow wave stage of sleep.
A review of automated sleep disorder detection [4]

• From 2010 to 2021, authors of 95 scientific papers have taken up the challenge of automating sleep disorder
detection. This paper provides an expert review of this work.

• During the review, they found eight sleep disorders (Sleep apnea, Hypopnea, INS, RBD, Narcolepsy, PLMD,
NFLE, Bruxism), of which sleep apnea and insomnia were the most studied. These disorders can be
diagnosed using several kinds of biomedical signals, such as Electrocardiogram (ECG), Polysomnography
(PSG), Electroencephalogram (EEG), Electromyogram (EMG) and snore sound.

• As per this study, 24 distinct algorithms were used to analyse 20 different signals or indeed signal
combinations.

• The authors conclude that this diversity of methods and signals clearly documents the challenge faced by
practitioners in the area of automated sleep disorder detection.
Automated sleep stage classification in sleep apnea using convolutional neural networks [5]

• This automation method publicly presents a flexible deep learning model and machine learning approach
utilizing raw electroencephalogram (EEG) signals.

• Its effectiveness is determined by using two open-source, public databases available from PhysioNet
(PhysioNet Sleep-EDF dataset, which contains 197 polysomnograms (PSG) to evaluate the performance of
the proposed model for sleep stage assignment): two recordings with 5402 epoch counts.

• The deep learning model is a Deep Convolutional Neural Network (CNN) that analyses invariant time
capacities and frequency actualities and collects assessment adaptations.

• The technique used in this approach achieves an accuracy of 90.70%, precision of 90.50%, recall of 92.70%,
and F-measure of 90.60%. The proposed method is more significant than existing models like AlexNet,
ResNet, VGGNet, and LeNet.
Classification of Brainwaves for Sleep Stages by High-Dimensional FFT Features from EEG Signals [6]

• In this paper, the authors demonstrated utilization of features extracted from EEG signals via FFT to improve
the performance of automated sleep stage classification through machine learning methods.

• They incorporated thousands of FFT features in order to classify the sleep stages into 2–6 classes. Using the
expanded version of Sleep-EDF dataset with 61 recordings, their method outperformed other state of the art
methods.

• One of the disadvantages in this method is the intensive computational requirements in memory and
processor. However, it also means that if the available resource of computing is rich, its performance can be
further improved.
Work Flow
 Work Done
• Sleep Study:

In this we have collected information about stages of sleep, sleep requirements based on age group,
effect of food intake on sleep and sleep disorders.

• Dataset Collection:

In this we have collected a suitable dataset, sleep-edf database contains 197 whole-night
PolySomnoGraphic (PSG) sleep recordings, for extraction of all required features.

• Data Pre-Processing and Feature Extraction:

In this we handled missing values and normalised the data and extracting relevant features
from the dataset.
 Sleep Study

Sleep Cycle:

Sleep is not uniform. Instead, over the

course of the night, your total sleep is
made up of several rounds of the sleep Sleep cycles can vary from person to
cycle, which is composed of four person and from night to night based on
individual stages. In a typical night, a a wide range of factors such as age,
person goes through four to six sleep recent sleep patterns, and alcohol
cycles. Not all sleep cycles are the same consumption.
length, but on average they last about
90 minutes each.
Stages of Sleep Cycle

Source: www.ezsleeptime.com
 • Sleep occurs in four stages: NREM1, NREM2,
NREM3 and REM.
• Stages NREM1 to NREM3 are considered non-
rapid eye movement (NREM) sleep, with each
stage a progressively deeper sleep.
Approximately 75% of sleep is spent in the
Stages of NREM stages, with the majority spent in the
NREM2 stage.
Sleep Cycle • Last stage REM is considered rapid eye
movement sleep during which even though the
eyes are closed, they can be seen moving
quickly.
• REM stages make up around 25% of sleep in
adults.
 • NREM1 stage is essentially when a person first
falls asleep. This stage normally lasts just one to
seven minutes.
• During N1 sleep, the body has not fully relaxed,
though the body and brain activities start to slow
NREM1 with periods of brief movements.
Stage (Pre- • It is easy to wake someone up during this sleep
stage, but if a person is not disturbed, they can
sleep) move quickly into stage 2.
• The feeling of falling (officially named hypnic
myoclonia) is sometimes experienced.
• EEG recording: theta waves - low voltage.
 • During stage 2, or N2, the body enters a more
subdued state including a drop in temperature,
relaxed muscles, and slowed breathing and heart
rate.
NREM2 • Eye movement stops and brain activity slows,
but there are short bursts of activity that actually
Stage (Light help resist being woken up by external stimuli.
Sleep) • Stage 2 sleep can last for 10 to 25 minutes
during the first sleep cycle, and each N2 stage
can become longer during the night.
• EEG recording: Sleep spindles(brief, powerful
bursts) and K complexes(long delta waves).
 • During stage 3, muscle tone, pulse, and breathing
rate decrease in N3 sleep as the body relaxes even
further.
• It is harder to wake someone up if they are in this
phase.

NREM3 • The brain activity during this period has an

identifiable pattern of what are known as delta
Stage (Deep waves. For this reason, stage 3 may also be called
delta sleep or slow-wave sleep (SWS).
Sleep) • Body begins rebuilding processes: tissue
regeneration, building of bone and muscle and,
strengthening of immune system.
• EEG recording: delta waves - lowest frequency,
highest amplitude.
 • During REM sleep, brain activity picks up, nearing
levels seen when you are awake.
• At the same time, the body experiences atonia,
which is a temporary paralysis of the muscles, with
two exceptions: the eyes and the muscles that
control breathing.
REM Stage • REM sleep is believed to be essential to cognitive
(Dream functions like memory, learning, and creativity.
• REM is associated with dreaming and is not
Sleep) considered a restful sleep stage.
• Breathing, heart rate and blood pressure are irregular
during REM stage.
• EEG recording: beta waves - similar to brain waves
during wakefulness.
Stages of Sleep Cycle
Sleep Average
Stages Other names Length Description Percentage

NREM1 Pre-sleep 1-7 minutes Brain activity decreases from its stage of 5-10% of total sleep
wakefulness. The feeling of falling (officially in adults.
named hypnic myoclonia) is sometimes
experienced.
NREM2 True sleep or light sleep 10-25 minutes Periods of muscle tone mix with periods of 45-55% of total
muscle relaxation. Body temperature decreases sleep in adults.
and heart rate slows in preparation for deep
sleep.
NREM3 Deep sleep, Slow wave 20-40 minutes Your body begins rebuilding processes: tissue 15-25% of total
sleep or Delta sleep regeneration, building of bone and muscle and, sleep in adults.
strengthening of immune system.

REM Dream sleep 10-60 minutes Intense dreaming with periods of muscle 20-25% of total
paralysis. First period lasts around ten minutes sleep in adults.
and each REM cycle is longer than the last.

Source: www.sleepfoundation.org
Duration of sleep stage as sleep progresses

Source: www.sleepfoundation.org
 Sleep Cycle NREM1 NREM2 NREM3 REM Total time of
Round Duration Duration Duration Duration sleep cycle

First 5-15 minutes 15-25 minutes 30-40 minutes 1-10 minutes 70-80 minutes

Second 1-3 minutes 25-35 minutes 30-40 minutes 10-20 80-90 minutes
minutes
Third 1-3 minutes 35-45 minutes 20-30 minutes 20-30 90-100 minutes
minutes
Fourth 1-3 minutes 45-50 minutes 15-25 minutes 40-50 90-110 minutes
minutes
Fifth 1-3 minutes 45-50 minutes 10-20 minutes 40-60 100-120 minutes
minutes

Note that these values are given based on an average adult of age 18 – 64
years, who in general experiences 4-6 sleep cycles in entire sleep. So these
values can vary based on age, health status and other factors.
• The first sleep cycle is often the shortest, ranging from 70
to 100 minutes, while later cycles tend to fall between 90
and 120 minutes.
• N1 stage normally lasts just one to seven minutes.
Usually a person spends most of the N1 stage in first
sleep cycle. As the night unfolds, an uninterrupted sleeper
may not spend much more time in stage 1 as they move
through further sleep cycles.
• So, stage 1 sleep is considered to be only 5% of total
sleep in adults.
• Stage 2 sleep can last for 10 to 25 minutes during the first
sleep cycle, and each N2 stage can become longer during
the night. Collectively, a person typically spends about
half their sleep time in N2 sleep.
• In total, NREM2 stage of sleep make up around 50% of
total sleep in adults.
• Usually a person spends most time in deep sleep during
the first half of the night. During the early sleep cycles,
N3 stages commonly last for 20 to 40 minutes. As you
continue sleeping, these stages get shorter, and more time
gets spent in REM sleep instead.
• In total, stage 3 or deep sleep make up around 20% of
total sleep in adults.
• Usually REM stage first occurs after 90 minutes falling
asleep.
• As the night goes on, REM stages get longer, especially
in the second half of the night. While the first REM stage
may last only a few minutes, later stages can last for
around an hour.
• In total, REM stages make up around 25% of sleep in
adults.
Sleep requirements based on age group

• Sleep requirements by age group as per the National Sleep Foundation.

Results of Polysomnography on
different stages of sleep

Source: www.slideshare.net
 • Sleep apnea is a potentially serious sleep disorder in which
breathing repeatedly stops and starts. If you snore loudly and feel
tired even after a full night's sleep, you might have sleep apnea.
• Insomnia (INS) is defined as having trouble falling asleep or staying
asleep, or as having unrefreshing sleep despite having ample
opportunity to sleep.

Sleep • Restless Leg Syndrome (RLS) is a condition that causes an

uncontrollable urge to move the legs, usually because of an
Disorders uncomfortable sensation.
• Periodic limb movement (PLM) disorder is a asleep disorder that
causes frequent and rhythmic movements of the legs and
sometimes arms while sleeping.
• REM sleep behavior disorder (RBD) is a sleep disorder in which you
physically act out vivid, often unpleasant dreams with vocal sounds
and sudden, often violent arm and leg movements during REM
sleep — sometimes called dream-enacting behavior.
• Nocturnal frontal-lobe epilepsy (NFE) previously known as
nocturnal frontal lobe epilepsy, is a form of focal
epilepsy characterized by seizures which arise during sleep. The
seizures are most typically characterized by complex motor
behaviors.
• Narcolepsy is a sleep disorder that makes people very drowsy during
the day. People with narcolepsy find it hard to stay awake for long
periods of time. They fall asleep suddenly.
• Hypopnea is when you take in shallow breathes for 10 seconds or
longer while asleep and your air flow is at least 30% lower than
normal. This can lead to abrupt reduction in blood oxygen saturation
with oxygen levels falling as much as 40% or more in severe cases.
• Bruxism is involuntary teeth grinding. This condition can pose
serious risks to tooth and jaw health and may require treatment to
reduce its impact.
 • A factor that may have an impact on sleep is
dietary intake, as both play an important role in
maintaining long-term health and wellbeing.
• The relationship between diet and sleep is
complex and multifaceted.

Effect of • Certain foods and drinks make it easier or harder

to get the sleep that you need.
food intake • While there is no single diet that is best for sleep,
a balanced diet made up largely of a variety of
on Sleep vegetables and fruits is able to provide a broad
range of vitamins and minerals, contributing to
better sleep while promoting a healthy weight.
• Food intake within 30 to 60 minutes before
bedtime was associated with delayed sleep onset
and decreased sleep efficiency.
 It is important to note that the effects of food on sleep can vary based on age,
health status and other factors.
 • Caffeine is thought to block the cell receptors that adenosine (a
substance in the brain) uses to trigger its sleep-inducing
signals. It can take as long as 6–8 hours for the effects of
caffeine to wear off completely. Thus, drinking a cup of coffee
in the late afternoon may prevent your falling asleep at night.
• Nicotine is another stimulant that can keep you awake.
Nicotine also leads to lighter than normal sleep, and heavy
Reasons for smokers tend to wake up too early because of nicotine
withdrawal.
sleep • Although alcohol is a sedative that makes it easier to fall
asleep, it prevents deep sleep and REM sleep, allowing only
disruption the lighter stages of sleep. People who drink alcohol also tend
to wake up in the middle of the night when the effects of an
alcoholic “nightcap” wear off.
• Heart and blood pressure medications known as beta blockers
can make it difficult to fall asleep and cause more awakenings
during the night.
• People who have chronic asthma or bronchitis also have more
problems falling asleep and staying asleep than healthy people, either
because of their breathing difficulties or because of the medicines they
take.
• A number of psychological disorders including schizophrenia, bipolar
disorder, and anxiety disorders are well known for disrupting sleep.
• Depression often leads to insomnia, and insomnia can cause
depression. Some of these psychological disorders are more likely to
disrupt REM sleep.
• People who feel stressed also tend to spend less time in deep sleep and
REM sleep. Many people report having difficulties sleeping if, for
example, they have recently lost a loved one, are going through a
divorce, or are under stress at work.
• Certain lifestyle factors also may deprive a person of needed sleep.
Large meals or vigorous exercise just before bedtime can make it
harder to fall asleep.
• There can be many other factors which can disrupt sleep like noises,
bright lights and temperature of the room.
 The sleep-edf database contains 197 whole-
night PolySomnoGraphic (PSG) sleep
recordings, containing EEG, EOG, chin EMG,
and event markers. Some records also contain
Dataset respiration and body temperature.
Corresponding hypnograms (sleep patterns)
Collection were manually scored by well-trained
technicians according to the Rechtschaffen and
Kales manual, and are also available.
 • The extraction of informative statistical features
from the EEG signal is necessary to perform
sleep stage classification efficiently. In general,
the EEG signal is highly complex and non-
linear, so it would be better to use a non-linear
Data Pre- model.
processing • The values of a given time-series data as a
and Feature numeric sequence data are converted into a finite
set of the frequency domain.
Extraction • Then, to deconstruct signals into segmented
EEG signal sequences, we divided them into
equal time intervals called epochs. The length of
each epoch was set to every 30 s of EEG signal.
The time –domain and frequency-domain representation of EEG signal
 References
[1] Han H, Oh J. Application of various machine learning techniques to predict obstructive sleep apnea syndrome
severity. Scientific Reports. 2023 Apr 19;13(1):6379.

[2] Alsolai H, Qureshi S, Iqbal SM, Vanichayobon S, Henesey LE, Lindley C, Karrila S. A Systematic Review of
Literature on Automated Sleep Scoring. IEEE Access. 2022 Jul 27.

[3] Touil M, Bahatti L, El Magri A. Sleep’s depth detection using electroencephalogram signal processing and neural
network classification. Journal of Medical Artificial Intelligence. 2022 Sep 30;5.

[4] Xu S, Faust O, Seoni S, Chakraborty S, Barua PD, Loh HW, Elphick H, Molinari F, Acharya UR. A review of
automated sleep disorder detection. Computers in Biology and Medicine. 2022 Nov 1;150:106100.

[5] Sundar GN, Narmadha D, Jone AA, Sagayam KM, Dang H, Pomplun M. Automated sleep stage classification in
sleep apnoea using convolutional neural networks. Informatics in Medicine Unlocked. 2021 Jan 1;26:100724.
[6] Delimayanti MK, Purnama B, Nguyen NG, Faisal MR, Mahmudah KR, Indriani F, Kubo M, Satou K.
Classification of brainwaves for sleep stages by high-dimensional FFT features from EEG signals. Applied
Sciences. 2020 Mar 5;10(5):1797.
[7] Yildirim O, Baloglu UB, Acharya UR. A deep learning model for automated sleep stages classification using
PSG signals. International journal of environmental research and public health. 2019 Feb;16(4):599.
[8] Ravan M. Machine learning approach to measure sleep quality using EEG signals. In2019 IEEE Signal
Processing in Medicine and Biology Symposium (SPMB) 2019 Dec 7 (pp. 1-6). IEEE.
[9] Ghasemzadeh P, Kalbkhani H, Shayesteh MG. Sleep stages classification from EEG signal based on
Stockwell transform. IET signal processing. 2019 Apr;13(2):242-52.
[10] Zhang Y, Wang B, Jing J, Zhang J, Zou J, Nakamura M. A comparison study on multidomain EEG features
for sleep stage classification. Computational Intelligence and Neuroscience. 2017 Nov 5;2017.
[11] Diykh M, Li Y. Complex networks approach for EEG signal sleep stages classification. Expert Systems
with Applications. 2016 Nov 30;63:241-8.
[12] Troxel WM, Buysse DJ, Matthews KA, Kip KE, Strollo PJ, Hall M, Drumheller O, Reis SE. Sleep
symptoms predict the development of the metabolic syndrome. Sleep. 2010 Dec 1;33(12):1633-40.
[13] Vural C, Yildiz M. Determination of sleep stage separation ability of features extracted from EEG signals
using principle component analysis. Journal of medical systems. 2010 Feb;34:83-9.
Thank
You