SMART WAIST BELT FOR HEALTH MONITORING

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

SMART WAIST BELT FOR HEALTH MONITORING

1,2,3 Student, Dept. of Electronics and Communication, BNMIT, Bangalore, Karnataka, India 4 Associate Professor, Dept. of Electronics and Communication, BNMIT, Bangalore, Karnataka, India ***

Abstract Although fitness bands are the most efficient way to manage a detailed and ongoing log of a user's health parameters in the modern world, serious flaws in current fitness bands, such as inaccurate step counting due to frequent hand movements, discourage users from using them. Additionally, people's lives are being swiftly transformed by the advances in wearable technology in recent years. Employees in corporate environment nowadays are compelled to spend long periods of time in front of workstations where their posture, whether knowingly or unknowingly, is affected. Back pain is reportedly one of the most common reasons for people to visit the orthopedic, so maintaining good posture is crucial for living a healthy lifestyle. The purpose of this project is to resolve this by designing a Smart Waist Belt that actively tracks the number of steps taken, keeps track of the wearer's posture, and measures heart rate. All of the data is transmitted in real time to the mobile application and is also stored in the cloud for subsequent access. The smartphone application also categorizes the type of activities undergone, such as walking, running, or standing, and provides information on the number of calories burned, heart rate, and duration of exercises. In this paper we have implemented Random Forest Algorithm as our classifier to classify different activities. Support Vector Classifier accuracy is 94.02%, Logistic Regression accuracy is 95.19%, KNN accuracy is 90.02% and finally, Random Forest gives an accuracy of 95.51%.

Key Words: Smart Waist Belt, Posture, Step Count, Pulse Rate, Heartbeat, Random Forest, Machine Learning and IoT.

1. INTRODUCTION

The World Health Organization defines health [6] as a conditionofoverallwell being,notjustphysicalwell being, butalsomental,bodily,and social well being.People who are mentally healthy are immediately physically healthy. Man's greatest possession is good health. A healthy individual is one who can perform at their highest level without any hindrance. Numerous other bodily processes aremadeeasierbygoodhealth.Wecancopewithstressand fight off mounting demands better when we are in good health.

Our wellbeing of spine depends on having good posture, which has many advantages like fewer back pains, more energy,andmoreself assurance.Havingproperposturecan helpusavoidmusclestrain,soreness,exhaustion,andmany

other common illnesses and medical disorders, which are crucialforourgeneralhealth.It'snevertoolatetoadjustour postureormakeimprovements,especiallyifit'scausingone ormorehealthissues.

Walking is a painless way to burn calories. These days, expertsadvisetakingadaily,30 minutewalk.Fastwalkingis required, not sluggish ambling. Additionally, walking delivers a fantastic total body workout. It improves metabolism and combats fat. Diabetes patients are recommendedtoexerciseeverydaysinceitcandrastically lowertheirbloodglucoselevels.

Although wearable technology is the most efficient wayto keep track of a detailed and ongoinglog of a user's health behaviors,criticalflawsincurrentfitnesstrackers,suchas inaccurate step calculations caused by frequent hand movements,makeitdifficulttoofferpersonalizedtreatment. Asaresult,usersavoidusingthesedevices,whicheventually lowers the rate of activity tracker use over time. Working fromhomemeansthatnooneiswatchingourpostureduring thistime,andlongperiodsofsittinghavebeenshowntobe fatalformanyofus.Accordingtoresearch,smokingisnotas harmfulasprolongedsitting.Duetotheirbusyjobschedules, mostpeopleintoday'sworldforgofitnessandexercise. To overcomethisproblem,SmartWaistBeltintegrateswellwith ourdailyroutineandremindsustobephysicallyactive.

TheSmartWaistBelt'sobjectivesareto: Track theuser's step count in real time and show the number of calories burned using a mobile application. To keep track of the user's upper body posture and warn them if the right postureisn'tbeingkept. Tokeeptrackoftheuser'spulse rate. Tocategoriseactions(suchsitting,standingupright, running,andwalking)usingtheRandomForestapproach.

2. RELATED WORK

This chapter has the available literature that we have referredto.Wehavereferredmanyresearchpapers,someof therelevantresearchpapersarelistedbelow:

“IoT based Smart Posture Detector” Greeshma Karanth, NiharikaPentapati,ShivangiGuptaandRoopaRavish[1].In thispaper,afeasiblesolutiontothebadpostureproblemis presented using a wearable device that recognizes the posture of the person and sends live data on the phone throughanapp.

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal |

Neeraj H Gowda1 , Rahul M2 , Shashank Simha B K3, Dr. Subodh Kumar Panda4

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

“Microcontroller based fitness analysis using IoT” M Gunasekaran and Dr. G Nallavan [2]. This paper, aims to measurethenumberofstepsandthedistancewalkedbythe personusingasmallmechanicalpedometerofsize50×50×25 mm. The device counts the number of steps taken by the personandmultipliesitbytheaveragesteplengthfedtogive thedistancewalkedbytheperson.

“Posture Detection and Correction System using IoT”: A Survey NiteshSahani,PrafullPatilandSiddheshChikane[3] This paper provides a survey on detecting posture of a personandsuitablemeanstocorrectitwiththeaidofIoT.

Averycompactdevicewhereusercanattachiteasilyonthe backsideoftheirbody.Thedevicealertstheuserthrough buzzer alarm in case of a bad posture and sends data to androidapplicationviaBluetooth.

“Health Monitoring Systems using IoT and Raspberry Pi” Vivek Pardeshi, Saurabh Sagar, Swapnil Murmurwar and PankajHage[4] Inthispaper,ananalysisonRaspberry Pi basedhealthmonitoringsystemusingIoT hasbeenmade. Any abnormalities in the health conditions can be known directlyandareinformedtotheparticularpersonthrough GSM technology or via internet. An accessible database is createdaboutpatient’shealthhistorywhichcanbefurther monitored&analysedbythedoctorifnecessary.

“AnActivityRecognitionFrameworkDeployingtheRandom ForestClassifierandASingleOpticalHeartRateMonitoring and Triaxial Accelerometer Wrist Band” Saeed Mehrang, Julia Pietila and Ilkka Korhonen [5] In this paper, wrist wornsensorshavebettercomplianceforactivitymonitoring comparedtohip,waist,ankleorchestpositions.Arandom forest(RF)classifierwasexploitedtodetecttheseactivities basedonthewristmotionsandopticalHR.

3. METHODOLOGY

MPU6050(accelerometer)isintegratedwiththeRaspberry Pi which is used to monitor the posture. MAX30100 Heartbeat sensor keeps track of Heart Rate and Oxygen Saturationleveloftheuser. Vibration sensor is used to sense thevibrationswhilewalking.Hencecountingthenumberofsteps taken by the user. Datafromsensorsissenttothecloudover theWi Fi TheSmartWaistBeltispoweredbyabatteryor power bank. The data stored in the cloud is displayed throughthemobileapplication.Dataanalyticsisdoneusing thedatacollectedfromcloud

Fig 2:FlowChartofSmartWaistBelt

The Flow Chart of Smart Waist Belt is as shown in Fig 2. UsingRaspberryPi,theaccelerometerandheartratesensors areintegratedandcalibrated.Thenumberofstepstakenare countedusingavibrationsensor,whichismountedonthe backofthebodyontheSmartWaistBelt Wecanassessour body's postureusingan accelerometer. We can record the user’s heartbeat whenever necessary by employing a heartbeatsensor.WemakeuseofBlynktocreateanappwith datafromhardwaresensors(i.e.,storedincloud).

Thedatadeliveredtothemobileapp(application)hastobe cleansed because it contains erroneous calculations and sensor generatedmistakes.Pandas(PanelData)isaPython modulethatisusedfordatacleansing Dataisread,written, cleaned,andmodifiedusingPandas.Themobileappdisplays thenumberofstepstakenandcaloriesburned,andinforms the user anytime poor posture is detected. To create an activityclassifierforactivitieslikesitting,walking,jogging, etc., we employ a Machine Learning algorithm (i.e., the Random Forest classifier). Both Classification and Regression models use the Random Forest classifier. The collectivedataisusedfordataanalyticswhichisstoredin theformofgraphs.Thesegraphscanbeobtainedondaily, weeklyandmonthlybasis.Ifthereareanyirregularitiesin theuser,thiscanbehelpfulforthedoctorforanalyzingthe problem.

Fig 1: BlockDiagram

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

A well known Machine Learning Algorithm that uses Supervised Learning is called Random Forest . The Flow DiagramoftheRandomForestAlgorithmisshowninFig 3. Random Forest Algorithm can be applied to Machine LearningissuesinvolvingbothClassificationandRegression. Itisbuiltontheideaofensemblelearning,whichisamethod ofintegratingvariousclassifierstoaddressdifficultissues andenhancemodelperformance.

Fig 4showstheprototype of Smart WaistBelt whichhas beenimplementedbyinterfacingRaspberryPiwithdifferent sensors like the Accelerometer [IMU MPU6050], the Heartbeatsensor[MAX30100]andtheVibrationsensor.The Blynk Platform is used to build the mobile application as showninFig 5,whichdisplaystheHeartRate(inBPM)and SpO2 readings(in%),stepstaken,caloriesburned,number ofunfavorablepostures,aswell asgraphsofbothcurrent andhistoricaldata.

Fig 3: FlowDiagramofRandomForestAlgorithm

RandomForestisaclassifierthatusesmanydecisiontrees on different subsets of the input dataset and averages the resultstoincreasethedataset'spredictedaccuracy.Instead ofdependingonasingledecisiontree,theRandomForest usesforecastsfromeachtreeandpredictstheresultbased onthevotesofthemajorityofpredictions.Higheraccuracy andoverfittingarepreventedbythelargernumberoftrees intheforest.

4. RESULTS AND DISCUSSION

Fig 5: SmartWaistBeltMobileApp

Fig 6illustrateshowthemobileapplicationcountssteps.It isnotedthat42stepsweretakenand1.6kilocalorieswere burned.Additionally,thegraphhasbeensetuptodisplay thetotalamountofstepstakenovertheprevioushour.The graphcanalsobeshownforvarioustimeperiods,suchas onehour,sixhours,oneday,oneweek,onemonth,andthree months.

Fig 4: PrototypeofSmartWaistBeltforHealthMonitoring

Fig 6: Countingstepsinrealtime

The user's pulse rate (in BPM) and saturation level (in percent)aredisplayedinFig 7.Ithasbeennotedthatthe user's BPM, or beats per minute, is 96. Additionally, the user'sSpO2,oroxygensaturation,isobservedtobe100%.

Fig 8: LinegraphofPulseRate(inBPM)v/stime

The notification when an improper posture is found is shown in Fig 9. A notification is displayed in the mobile applicationasshowninFig 9whentheusermaintainsthe improperpostureformorethan5secondstherebywarning theuserofhisbadposture.

Fig 10showsthedemonstrationofcorrectpostureusingthe Smart Waist Belt and Fig 11 shows the demonstration of incorrectposturewhereinthepostureofthepersonis not right,leadingtoanincorrectposturenotificationasshownin Fig 9.

Fig 7: DisplayofPulseRate(inBPM)andSpO2 (in%)

Thelinegraphofpulseratev/stimeisdisplayedinFig 8.It canbeseenthatthegraphisdrawnusingBPMdatafromthe previoushour.Thegraphcanalsobeshownforvarioustime periods,suchasonehour,sixhours,oneday,oneweek,one month, and three months. Similar graphs can also be observedforSpO2(in%),stepstakenandthenumberofbad postures.

Fig -9: Notificationwhenincorrectpostureisdetected

Fig -12: Activityclassification

Fig 12 classifies the data into different activities like sitting, walking and standing. The values in pie chart indicatethepercentageoftheactivitiesperformed.

Fig 11: Demonstrationofincorrectposture

Fig 13: Accuracyofdifferent classifiers

Fig 13showstheaccuracyoftheMachineLearningmodels of Support vector, Logistic Regression, K Nearest and RandomForestusingthedataset.

TheRandomForestgivesthebestaccuracyforthedataset used. Support Vector Classifier accuracy is 94.02%, Logistic Regressionaccuracy is 95.19%,KNN accuracyis 90.02% and finally, Random Forest gives an accuracyof 95.51%. In this project we have implemented Random Forestasourclassifiertoclassifydifferentactivities.

Fig 10: Demonstrationofcorrectposture

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p ISSN: 2395 0072

5. CONCLUSION

Inthis paper, a prototypeof Smart WaistBelt for Health Monitoringhasbeenimplementedtomonitorthehealthof apersonandclassifytheactivitiesperformedaccordingly using Random Forest Algorithm which gives greater accuracy as compared to other algorithms. This Smart WaistBelttracksthenumberofstepstakenbytheuserin real time and displays the calories burnt through the mobileapplication.ThisSmartWaistBeltcanbeusedto caterwiderangeofuserswhichhelptheminmonitoring thestepcount,caloriesburnt,posturedetectionandHeart Rate.

6. REFERENCES

[1]GreeshmaKaranth,NiharikaPentapati,ShivangiGupta, and Roopa Ravish, "IoT based Smart Posture Detector", ConferencePaper,September2019,pp1 10.

[2] M Gunasekaran and Dr. G Nallavan, “Microcontroller basedfitnessanalysisusingIoT”,InternationalJournalof PhysicalEducation,SportsandHealth,April2017,pp1 4.

[3]NiteshSahani,PrafullPatil,SiddheshChikane,Posture Detection and Correction System using IOT: A Survey, International Research Journal of Engineering and Technology (IRJET), Volume:07, Issue: 12, Dec 2020, pp 1165 1169.

[4] Vivek Pardeshi, Saurabh Sagar, Swapnil Murmurwar and Pankaj Hage, “Health Monitoring Systems using IoT andRaspberryPi”,InternationalConferenceonInnovative MechanismsforIndustryApplications(ICIMIA),February 2017,pp1 4.

[5] Saeed Mehrang, Julia Pietilä and Ilkka Korhonen. “An Activity Recognition Framework Deploying the Random ForestClassifierandASingleOpticalHeartRateMonitoring andTriaxialAccelerometerWrist Band”,EuropeanMedical andBiologicalEngineeringConference(EMBEC),February 2018,pp1 11.

[6]https://www.who.int/about/governance/constitution

[7]Prof.BharatiPandhare,DarshitAkbari,FenilVaghani, AkashGupta,"SMARTWAISTBELTUSINGINTERNETOF THINGS", International Journal of Creative Research Thoughts(IJCRT),Volume9,Issue3,March2021,pp60 64.

[8] Wann Yun Shieh, Tyng Tyng Guu, An Peng Liu, "A Portable Smart Belt Design for Home Based Gait Parameter Collection", Healthy Aging Research Center, ICCP2013Proceedings,2013IEEE,pp16 19.

[9] Miss. Pradnya Ravi Nakhale and Dr. U. A. Kshirsagar, “IoTbasedPedometerusingRaspberry Pi”,International ResearchJournalofEngineeringandTechnology(IRJET),

Volume:07Issue:03,Mar2020,pp1 3.

[10] Ms. Rutuja Anil Shinde, Dr. S. L. Nalbalwar and Dr. Sachin Singh, “Smart Shoes: Walking Towards a Better Future”,InternationalJournalofEngineeringResearch& Technology(IJERT),Vol.8Issue07,July2019,pp1 3.

[11]ChelseaG.Bender,JasonC.Hoffstot,BrianT.Combs, SaraHooshangiandJustinCappos,“MeasuringtheFitness of Fitness Trackers”, 2017 IEEE Sensors Applications Symposium(SAS),March2017,pp1 5.

2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal