A New Approach for The Prediction of Obstructive Sleep Apnea Using a Designed Device

International Journal of Advancements in Research & Technology, Volume 3, Issue 11, November-2014ISSN 2278-7763

Copyright © 2014 SciResPub.

A New Approach for The Prediction of Obstructive Sleep ApneaUsing a Designed Device

Abdulkader Helwan1, Nafez Haddad2, Ahmed Said Ahmed Ashit3

1,3Near East University, Near east boulevard, Mersin 10, Turkey; 1Faculty of Engineering, Biomedical engineering Department3Faculty of Engineering, Computer engineering department

2Al Manar University of Tripoli, Biomedical Department, Tripoli, Lebanon

Email: [email protected]@[email protected]

ABSTRACTSome diseases are still ambiguous for scientists and researchers, despite of studies and researches that have been done to studyand diagnose them. The effects of these diseases may be known, but the unknown is their causes, times, and signs. These unob-vious causes and signs make scientists unable to detect, nor diagnose these diseases. One of these diseases is the obstructivesleep apnea, which is defined as breathing difficulties and pauses in breathing during sleep. This research project aims to pro-duce a diagnostic device for early prediction of sleep apnea through the analysis and classification of sleep levels. The idea is toestablish an algorithm linking sleep apnea and snoring by the implementation of a device that measures the chest expansionduring inspiration and expiration of subjects. Then, collected data from which snorers and apnea goers are pointed out so thatwe can provide the patient with quantitative measurements and statistics that end up with a classification of sleep disorderslevels in terms of voltages interval of each level. Thus, it becomes easy to find a voltage interval just prior to obstructive sleepapnea occurrence. In such case, the machine is programmed to early predict and prevent the obstructive sleep apnea occurrenceby providing a buzzing sound in order to wake the patient up right before OSA.

Keywords : ambiguous, obstructive sleep apnea, early prediction, classification (keywords)

1 INTRODUCTIONiomedical Engineering is the combination engineeringwith life sciences; thus, this makes it an interdisciplinaryfield. The importance of that combination of fields can be

seen through our everyday lives, specifically when we go tohospital, receive medical treatment or even when we buyhealth products such as an heart rate monitor device. This ex-citing field has had a graet technological development inhealth care over the past years and this is due to the coopera-tive work of engineers, medical doctors, mathematicians, andphysicians [1].

When the person sleeps and wakes up in the morning,he/she definitely doesn’t know what problems occur withinhis/her internal body structure and organs. However, he/sheshould be careful, because these problems (if exist) may resultin big and dangerous consequences. The sleep apnea is one ofthese dangerous and ambiguous problems that may face eve-ryone without knowing and feeling because it happens whilesleep. In other words, breathing difficulties and pauses inbreath may happen to us during sleep and they may result ina high blood pressure, diabetes and even sudden death [2]. Tostudy and diagnose this disease, some researches had to bedone. Some researchers suggest that a snoring may be a signfor a sleep apnea and it is the first condition before getting intoapnea, and that the excessive snoring may lead to a sleep ap-

nea. On the other hand, other researchers suggest that not eve-ryone who snores has sleep apnea and not everyone has sleepapnea snores [3]. Therefore, one of these two studies should beconfirmed. The solution then, is to find the relationship be-tween sleep apnea and snoring, if there is, and attempt to de-tect sleep apnea by monitoring and analyzing the snoring lev-els while sleep.

Snoring can be defined as a consequence of changes in theconformation and properties of the upper airway that mayoccur during sleep. About 20 years ago some authors notedthat in way or another, habitual snoring might be related toobstructive sleep apnea (OSA) [2].

This project is divided into three essential parts:

1. Engineering work which led us to construct an electronicdevice that is simple, easy to use and low cost to measurethe person’s chest expansion using a Force SensitiveResistor (FSR) sensor. This device also gives off a warningalert when apnea is about to occur.

2. Research work that is based on a set of hypothetical values.

B

IJoART

62



Collected data were further studied statistically using SPSSsoftware by finding the mean, standard deviation, andcorrelation coefficients; and plotting some graphs in orderto analyze and find the voltages variation of each sleeplevel.

3. Once statistical data and computations were obtained, analgorithm was established depending on the predeterminedvalues of voltages and conditions found throughout thedata analysis of voltages prior to get apnea.

[4] Fig. 1. Force sensing resistor (FSR)

Force Sensing Resistor (FSR) is a polymer thick film (PTF) de-vice, in which it exhibits resistance change when an increaseor decrease in the force applied to its active surface. It is opti-mized to the force of human touch control of electronic devic-es [4].

2 MACHINE DESIGN AND SIMULATIONThe main purpose of this research project is to develop a lowcost, low power, reliable, and non-invasive Obstructive SleepApnea Detector that processes the data acquired from the FSRsensor to determine if the apnea is imminent to occur, accord-ing to an accurate protocol and algorithm based on a detailedand predetermined research about sleep apnea and the sleeplevels before it happens. This table below represents the elec-trical items used in design modeling.

Table 1: Electrical items used in design modeling

The system was programmed using Picbasic language onProton plus (Proton IDE) software. After debugging andcreating a PICmicro interface with LCD and FSR output usingProton IDE, the design was simulated using Proteus afteradding all required features such as the buzzer and some leds

to the design (see fig. 19).

Fig.2. Design simuation using proteus.

The circuit was implemented and tested on a breadboard; itwas then implemented on a PCB as a final step. In order not tomake mistakes, the implementation steps were done succes-sively and the circuit was tested after each step.

IJoART

63



Fig.3. Implemented circuit

As this step was accomplished, the machine was tested and itwas almost ready for starting tests. Whereas, certainreadjustment to the microcontroller program can be made.Also some additional features to the circuit may be addedsuch as additional push buttons.

2.1 Belt and sensor replacement

Since the FSR sensor is composed of a two conductive plates[4], it requires double side push to have a change in itsresistance; that’s why it was not easy to find the appropriatemethod and location to place the sensor; in which it is pushedup by the chest (Expiration) and at the same time pushed byan opposite direction by the belt.

Fig.4. Doule pushes side on FSR

After many trials, we found out that the thorax (chest region)is the most appropriate region for the belt to be placed and forthe sensor to be built in. Beside, we found out that one belt isnot enough because two opposite forces should be applied tosensor to have variation in its resistance value. Therefore thesensor should be placed between two belts in order to bepushed from its both sides (fig.4.b). The first lower belt shouldbe a rubber one (fig.4.a), while the upper belt should be a rigidone. In the expiration phase, the chest contracts; which pushesthe sensor from its lower side and at the same time the rigidnon rubber belt produces an opposite force that pushes theother side of the sensor results in resistance decrease.

a bFig. 4. (a) Belt on abdominal region, (b) two belts on thoracicregion

The device was first tested using one rubber belt placed on thepatient chest and the results were as following:Table 2: Output voltages obtained when one rubber belt is-placed on the chest

As noticed, the output voltage is not exceeding 1.8V whichmeans that the FSR sensor is not pushed against on its bothsides; data was not being precise. To solve that problem, an-other non rubber belt is placed just above the rubber one andthe result is as following:

Table 3: Output voltages obtained when two rubber belt are

placed on the chest

IJoART

64



3 DATA COLLECTION3.1 Candidates SelectionsA general criterion is considered to make selection of candi-dates; this criterion is a simple questionnaire to be filled out byeach individual candidate in order to speed up the process ofselection. After careful screening of the questionnaire, a policyhas been placed to supervise the whole process from candi-date selections to performing tests. The protocol of this re-search study says to include candidates from different agegroups, so that wider data can be obtained. Therefore candi-dates of different categories were selected. Those categoriesare [5, 6]:

• Normal Respiratory Rhythm (NRR) without snoringto establish standard voltage values

• Simple snoring (SS) to detect voltage values variationduring snoring

• Hypopnea (Heavy Snoring) (Hypo)2 Data CollectionThe following flowchart depicts the procedure of testing for

the three different categories (NRR, SS, and HS) using OSADetector.

Fig.5. Data collection process

After collecting the data for each sleep level, data is directlyfilled out in tables using the SPSS software in order to be plot-ted as graphs and compared to other normal graphs to checkfor the accuracy and reliability for the designed machine (OSADetector).

3.2 Data analysisThe collected data were analyzed and interpreted on the

same Software (SPSS) by plotting graphs represent the voltagevariation with respect to the number of voltages read by theOSA Detector. The following are the analyzed data for eachstudied level of sleep:

• Data analysis For NRR Fig.6 represents Normal Breathing Rhythm obtained di-

rectly from a candidate as compared to a Normal RespiratoryWaveform. It is is a waveform drawing of voltages that areproduced by full respiration (inspiration-expiration) versustime in 35 seconds. Normal respiratory rate of a healthy sub-ject is between 15 Bpm which is one complete breath (cycle)per 4 seconds. The microcontroller is programmed to have adelay of 0.6 seconds between two consecutive readings due tosensitivity issues of the electronic circuit. The X-axis representsthe number of voltages read by the device every 0.6s and it isautomatically updated. One complete respiratory cycle repeatsitself almost every seven voltage readings that is proportionalto the normal respiratory timing cycle (4s per 1 cycle), this isderived as

Respiratory Rate = Readings*Delay = 7*0.6 = 4s

and by generating some descriptive statistics tables that calcu-late the mean, the standard deviation and the skew

Fig.6. NRR pattern using OSA detector

• SS AnalysisAfter finishing the NRR category and getting the required re-sults, the next category to be tested was the SS category. Thetests were performed for three minutes in order to monitor allthe sleep levels happened during sleep, and then the samplingof the simple snoring pattern took place. The SS data weresampled, analyzed then compared to other normal snoringdata and graphs to check the accuracy of the designed deviceand to determine the voltage interval of the SS. Fig.7 repre-sents Simple Snoring pattern obtained directly from a candi-date using the OSA Detector.

IJoART

65



Fig.7. SS pattern using OSA detector

• HS analysisAs far as the candidate is tested by the device (OSA Detec-

tor), he/she may pass through different sleep disorders andlevels. One of the most dangerous disorders is the Hypopnea orHeavy Snoring. Hypopnea is a medical term for a disorderwhich involves episodes of overly shallow breathing or an ab-normally low respiratory rate. It is a partial blockage of the air-way that results in an airflow reduction of greater than 50% for10 seconds or more. It is less severe than apnea (which is a morecomplete loss of airflow). It may likewise result in a decreasedamount of air movement into the lungs and can cause oxygenlevels in the blood to drop. It more commonly is due to partialobstruction of the upper airway. Figure 33 represents HeavySnoring pattern directly from a obtained candidate using theOSA Detector (fig. 8).

Fig. 8. HS pattern using OSA detector

3.3 Statistical analysisThe application of statistics to medical data is used to design

experiments and clinical studies; to summarize, explore, analyze,and present data; to draw interference from data by estimation orhypothesis testing; to evaluate diagnostic; and to assist clinical

decision making. Quantitative data are measured on a continu-ous or discrete numerical scale with some precision [7]. Distribu-tions of data reflect the values of a variable or characteristic andthe frequency of occurrence of those values. In this statisticalstudy we used the following quantitative parameters: the mean,median, range, standard deviation.

The collected data were analyzed and interpreted individuallyfor each candidate and collectively for each category of candi-dates. In this regard, I have performed the followings:

• NRR data interpretationTable 4: NRR Descriptive Statistics

Fig.9. NRR Normal Distribution

SS Interpretation

Table 5: SS Descriptive Statistics

Fig.10. SS Normal Distribution

Hypo (HS) Interpretation

IJoART

66



Table 5: HS Descriptive Statistics

Fig.11. HS Normal Distribution

4 RESULTS DISCUSSIONA quick assessment of the above tables and recorded data isthat each table contains voltage range (low/high) for eachtype of sleep level. The voltage interval of each sleep level isthe range of voltages in between these two values (Low/high)including them. NRR voltage distribution is a symmetricaldistribution which that the right and left side has equal dis-tance to the mean (standard deviation). SS voltage distributionis a left skewed distribution since most of the voltages aregreater than the mean. Also the voltages are getting closer tothe mean since the standard deviation is 0.7. HS voltage dis-tribution is also left skewed; however, the voltages are gettingcloser to the mean than the SS voltages which results in a lowstandard deviation (0.38).

Table 6: Sleep Levels Voltages Interval

5 OSA PREDICTION ALGORITHMAfter analysis of the data collected using the designed ma-chine (OSA Detector), and depending on the accurate resultsand determination of sleep levels voltage, we created an algo-rithm based on the analysis of voltages interval and standarddeviation of the last sleep level (Heavy Snoring) before OSA.The Hypopnea (HS) is a sleep disorder in which a reduction ofairflow of 50% and greater, and a 3% desaturation and lastingfor at least 10 seconds. After the 10 seconds the patient mayeither return to the Normal Respiratory Rhythm (NRR) or get

into OSA. The algorithm consists of many conditions:1. Check if the voltage interval is still in the range of HS

interval or lower at the tenth and eleventh second.2. Check if the standard deviation is still in the range of

HS standard deviation (0.38-0.40) or lower, and also at thetenth and eleventh second.

3. If these two conditions were satisfied, this means thatthe patient is going into OSA; therefore the device will give offan alarm to wake the patient up. Otherwise, the patient is get-ting back to the NRR.

5 CONCLUSION

This research study has not passed without difficulties, butperseverance and upbeat determination to a successful con-clusion have overcome such nuisance. An outcome of this re-search study has been revealed, and that is the prediction thesleep dangerous disorder: the Obstructive Sleep Apnea. Thisnew method of prediction provides patients with quantitativemeasurements to uphold this and statistical analysis that hasbeen employed to solidify the data obtained. These three di-agnosed levels (NRR, SS, and HS) were studied and analyzedthrough measuring the chest voltages collected using a beltwrapped onto candidate chest with a built in sensor (FSR) anda simple circuit design to display voltages. The early predic-tion of OSA was a result of setting a voltage interval of eachsleep level, and also a result of determining a standard devia-tion and a mean of each level depending on the voltages col-lected directly from candidates. Furthermore, snoring data (SS,HS) is sampled and extracted form a complete sleep pattern,then it is analyzed and plotted in order to reduce biases anderrors. Finally, we can conclude that this research study hassucceeded in fulfilling the objectives set forth in the introduc-tion and our data is confirmed by earlier research studies doneby other machines.

ACKNOWLEDGMENT

I wish to thank my father for his help throughout my life

REFERENCES[1] Imperial College London, :Definition of Biomedical Engineerig”,

available:http://www3.imperial.ac.uk/pls/portallive/docs/1/51182.PDF

[2] National Sleep Foundation, “Sleep Apnea”, available:http://sleepfoundation.org/sleep-disorders-problems/sleep-apnea

[3] Medscape, “Obstructive Sleep Apnea”, available:http://emedicine.medscape.com/article/295807-overview

[4] Adafruit, “Force Sensitive Resistor”, available:https://learn.adafruit.com/force-sensitive-resistor-fsr/overview

[5] Fiz JA1, Abad J, Jané R, Riera M, Mañanas MA, Caminal P, Ro-denstein D, Morera J., “Acoustic analysis of snoring sound in patientswith simple snoring and obstructive sleep apnoea”. European Res-

IJoART

67



piratory Journal, 1996 Nov, p.p.2365-70.[6] Liistro G, Stanescu DC, Vertier C, Rodenstein DA, Aubert-Tulkens

G., “Pattern of snoring in obstructive sleep apnea patients and inheavy snorers”. Sleep, 1991; 2014, p.p.517–525.

[7] John G. Webster, Medical Instrumentation Application and Design. 4thedition, p.p.40.150.

IJoART

68

A New Approach for The Prediction of Obstructive Sleep Apnea Using a Designed Device

Documents

Transcript of A New Approach for The Prediction of Obstructive Sleep Apnea Using a Designed Device