Prediction of heart disease using neural network

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

Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072

Prediction of heart disease using neural network

1M. Tech Student, Dept of Computer Engineering, and IT, VJTI College, Mumbai, Maharashtra, India

2Associate Professor, Dept of Computer Engineering, and IT, VJTI College, Mumbai, Maharashtra, India ***

Abstract - Healthcare involves an vital portion in human lives. The healthcare industry contain huge sum of psychiatric data hence machine learning models were utilized to supply conclusion effectively within the heart illness forecast. The classification of healthy individual and non-healthy individual can be done dependably by using machine learning strategies. We created a system in this investigation that can get it the standards of predicting the chance profile of patients with the clinical data parameters. The proposed model is built utilizing Deep Neural Network gives better results on both the testing and preparing information. DNN and ANN were utilized to dissect the effectiveness of the demonstrate which accurately predicts the nearness or nonappearance of heart infection. This paper gives the web based framework for prediction of heart disease utilizing Deep Learning (DL) calculations with a great exactness compared to other works

Keywords Deep Learning, Machine learning, Artificial Neural network, health care services, heart disease.

1. INTRODUCTION

One of the discernible disease that impact numerous individuals in old age is coronary illness, and numerous times, it in the long run leads to dangerous issues. Cardiovasculardiseasesaregettingtobemoreincommonin India. It is considered that suitable health care administrations should be accessible for a customary checkupofone'swellbeing.Nearly31%ofallpassingsare due to heart-related malady in all over the world. Early discovery [1] and treatment of a few heart infections is exceptionally complex, particularly in developing nations, since of the need of diagnostic centers and qualified specialistsandotherassetsthatinfluencetheexactguessof heart malady. With this concern, in later times computer innovationanddeeplearningstrategiesarebeingutilizedto formrestorativehelpprogramasabackframeworkforearly conclusion of heart infection. Identification of any heart related sickness at essential organize can diminish the passingchance.DifferentDLstrategiesareutilizedinmedical informationtogetitthedesignofinformationandmaking expectationfromthem.Healthcareinformationareforthe mostpartenormousinvolumesandcomplexinstructure.DL calculationsarecapabletohandletheenormousinformation and mine them to discover the significant data. Deep Learning calculations learn from past information and do forecastongenuinetimeinformation.ThissortofDLsystem for coronary sickness desire can energize cardiologists in

takingfasteractionssomorepatientscangetmedications inside a shorter time period, hence sparing expansive numberoflives.

DeepLearningisadepartmentofAIinvestigate[2]and has gotten to be an awfully well known viewpoint of data science. The Deep Learning calculations are outlined to perform a large number of errands such as prediction, classification,choicemakingetc.TolearntheDLcalculations, preparinginformationisrequired.Afterthelearningstage,a show is delivered which is considered as an output of DL calculation.Thismodelisthentestedandapprovedonaset ofunseengenuinetimetestdataset.Thefinalprecisionofthe model isatthatpointcomparedwiththegenuineesteem, which legitimize the general rightness of the anticipated result.

2. RELATED WORK

Lotsofresearchworkhavebeendoneforappraisalofthe classification correctnesses of diverse deep learning and machine learning calculations by utilizing the Cleveland heartillnessdatabasewhichisuninhibitedlyavailableatan internet information mining store of the UCI. Various systems for coronary disorder forecast utilizing AI calculationwascreatedearly.Algorithmlikesupportvector machine,K-nearestneighborandArtificialNeuralNetwork wascreatedearlytoexpecttheexistenceornonappearance of cardiac disorder. Study demonstrates that ANN-based modelsaregenerallyutilizedinheartdiseaseforecastand theprecisionofpastworkswaslessindifferentiatewiththe created model. A self-operating demonstrate for cardiac disease discovery employing a Deep neural network and ANN. The proposed model had 2 covered up layers, ConvolutionalDNNislessprecisethantheproposedmodel i.e2-DNN,andtheexactnessof2-ANNislessascomparedto 2-DNN.[5].

systemwasbasedonthedatasetqualityclassificationout ofeighteenqualitiestheychoseeightqualitiesasthemost quality are Age, sexual orientation, smoking propensity, diabetes, cholesterol, chest torment, hypertension, family history.ThecreatorhasutilizeddifferentMLstrategiesand SVM gave 91% exactness utilizing eight properties and 86.03% utilizing eighteen qualities [3]. proposed several models for anticipating heart illness. Decision tree, naïve Bayes,multilayerrecognition,andEnsembleclassificationin whichRF,NBarecombinedandinproposedmodelauthor

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

Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072

connectedtherandomforestclassifierforfeatureextraction and k-means at that point connected the decision tree strategyforclassification,Theproposeddemonstrategave themostelevatedexactnessrateof94.44%[4].

3. PROPOSED SYSTEM

3.1 Problem Statement Elaboration

“ToPredicttheheartdiseaseusingtheNeuralnetwork.” Thestudyanalyzedbytheworldhealthorganization(WHO) gagesthat24%ofpeoplepassedoninIndiaduetocardiac disorder.Analystshaverecordedthedistinctivecomponents that increment the possibility of cardiac disorder and coronarysupplycoursedisease[12].WepointtoapplyDeep Learning strategies to the dataset to anticipate heart infection.

3.2 Proposed System Methodology

In practical applicationhighaccuracycomes aboutare generatedusinganeuralnetwork.Theproposedframework increments the classification exactness. The dataset is separated into the testing data and training dataset. The training dataset was given to the neural network. Neural systemsaresetofalgorithmsthatareutilizedtorecognize designs.Thelayerswithintheneuralnetworkaremadeupof activationfunction.

Thetrainingfeaturesaregiventothenetworkthrough theinputlayer.Thehighlightsaregiventothehiddenlayer wheregenuineprocessinghappenswiththeassistanceofa weightedconnection.Theoutputlayeroftheorganisationis joined with the hidden layer. Era of theory through deep learning models was the point of the predictive model. Hypothesisistherelationshipbetweeninformationwhich canbetriedbycollectingdataandmakingobservations[12]. Ready to produce the hypothesis by limiting the blunder withinthetrainingoccasions.

The execution of the network is dependent on the numberofrulesutilizedwhichdecidethebehaviourofthe network.Amodelwithlessparameterleadstolowcapacity which comes about in under fitting. Model with more numberofparametersthanrequiredleadstohighcapacity whichresultsinoverfittingsubsequentlythemodeloughtto be in such a way that it produces a hypothesis with ideal capacity. The hypothesis is defined using forward propagation. The input is given to the neurons which performafewoperationstocreatetheyieldthishandleis called the activation function. The activation work characterizestheoutputofanode.

3.3 Dataset Description

TheClevelandcoronarydisorderdatasetisutilizedwhich wastakenfromanonlineAIdocument.Thisdatasetisused forresearchstudies.Thedatasethas303occasionsand14 attributes[5].

TableIgivesthedescriptionofdataset

Sr. no Attribute Description Range

1.

AGE Patientage 29-77

2. SEX Patientgender 1=male 0= female

3. CP chestpain type 0=Atypical angina, 1=typical angina, 2= asymptotic, 3=non anginapain

4. TRE STBPS Resting Blood pressure 94-200 5. CHOL Serum cholesterollevel 126-564 6. FBS Fasting blood sugar 1>= 120, 0<=120

0=false 1= true 7. RESTEC G Resting electrocardiograp hic result

0=normal 1 = ST – T wave abnormaliti es 2 = left ventricular hypertroph y 8. THALAC H Maximum heart rateAchived 71-202 9.

EXANG Exercise Indused Angina 0=no 1=yes 10. OLD PEAK ST depression induced by exerciserelatedto rest

0.0–6.2

11. SLOPE Slope of the peak exercise ST segment

12. CA Count of major vesselscoloredby Fluoroscopy

0=un sloping 1=flat 2 = down sloping

0-3

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

Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072

13. THAL ThalliumScan 3=normal 6=fixed 7 = reversible effect

14. Target ClassAttribute 0=no 1=yes

4. IMPLEMENTATION

4.1 Data Preprocessing

One of the most crucial milestones in the process of putting deep learning models into practice is this one. To makethedatasetmoreappealingandpracticalforthemodel training phase, we deliberately applied all data cleaning strategiestoourdataset[9].Weremovedalltheunnecessary and irrelevant data from our dataset throughout the data cleaningprocess.

- Datacleaninghadthefollowinggoalsinmind-

- Removalofmissingdata

- Removalofduplicateentries

- RemoverowswithNaNvalues

4.2 Exploratory Data Analysis

Exploratorydataanalysisisastrategyforexaminingdata sets to highlight their key properties, frequently utilising statisticaltoolsandothertechniquesfordatavisualisation.it aids

webettercomprehendourdataset[11].ExecutingEDAon ourdatasetassistedusin:

- RecognizeandhandleNULLvalues.

- Recognizeandeliminateoutliers.

- Identifytheunderlyingrelationshipsandstructure. Additionally,wecreatedawordcloudandseveralgraphs. tolearnmoreaboutthedata.

5. Training and Testing

After preprocessing and EDA, we had the final dataset thathadbeenthoroughlycleanedandanalysed.The80-20 train-testvalidationmethodwasused,whichspecifiesthat 80%oftheinformationisusedforplanningand20%isused fortesting.Thesklearnlibraryisusedtodividethedatainto training and testing portions. Out of 303 samples, 242 examples or instances are chosen and used to create the model[5].Theremaining61samplesareusedastestingdata tojudgehowwelltheconstructedmodelperforms.

To implement the model, we proceeded forward. We carried out two distinct sorts of experiments during the implementation.Boththeconsidereddeepneuralnetwork and the artificial neural network were implemented individually. As a result, we were able to determine how accuratelyeachofthesemodelsperformed.

A Deepneural network has multiple hidden layers. WhereastheArtificialNeuralnetworkhasoneortwohidden layersinit.

Theactivationofneuronsispresentattheoutputlayer.

In the output layer, the sigmoid activation function is applied.Thedataset'sredundantfeaturesareremovedusing featureselection.Featureextractionandfeatureselectionare different.Findingrelevantcomponentsfromtheexistingdata is called feature extraction. By removing unnecessary featuresthroughfeatureselection,theneuralnetworkisfed withpertinentinformation.

Fig1:SystemArchitecture

6. RESULT

Theoutcomesthatwefoundaftertestingandputtingthe suggested algorithms into practice will be covered in this part.

The Deep neural network with the more hidden layer performs better than that of Artificial neural network. Accuracyofartificialneuralnetworkis70%andaccuracyof deepneuralnetworkis95%.SotheDNNismoreaccurate thanthatofANN.

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

Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072

patientswon'tundergoincorrectdiagnosesduetoinaccurate forecastsanditaidsthepatientinanemergency.Patients' lives are thereby saved since it aids in early disease prediction.Futureiterationsoftheapplicationcouldinclude ahugearrayofpatientrecordsandmoreattributes.

REFERENCES

[1] Md. Istiaq Habib Khan, M. Rubaiyat Hossain Mondal “Effectiveness of Data-Driven Diagnosis of Heart Disease” 2020pp:978-1DOI:10.1109/ICECE51571.2020.9393055

[2]VijetaSharma,ShrinkhalaYadav,ManjariGupta“Heart Disease Prediction using Machine Learning Techniques” 2020 2nd International Conference on Advances in Computing, Communication Control and Networking (ICACCCN)DOI:10.1109/ICACCCN51052.2020.9362842

[3]MohammedNowshadRuhaniChowdhury,EzazAhmed, Md. Abu Dayan Siddik, Akhlak Uz Zaman “Heart Disease Prognosis Using Machine Learning Classification Techniques” 2021 6th International Conference for Convergence in Technology (I2CT) DOI: 10.1109/12CT51068.2021.941818

[4] Ufaq Jeelani khan, Dr.Ashish Oberoi, Dr.Jasmeen Gill,” Hybrid Classification for Heart Disease Prediction using ArtificialIntelligence”20215thInternationalConferenceon ComputingMethodologiesandCommunication(ICCMC)pp: 978-1DOI:10.1109/ICCMC51019.2021.9418345

Fig.2.Predictionresultoftestrecord

Themodelistestedusingpatientreal-timedata.When patientinformationisenteredintothesystem,itdetermines thelikelihoodthatthepatienthasaheartproblem.

Table II gives the Precision,Recall andF1-Score of the model.

precision recall F1-score

Without heartdisease 0.94 0.97 0.96

Withheart disease 0.96 0.92 0.94

7. CONCLUSION

The proposed work develops a web application employing deep neural networks and artificial neural networks to discover heart problems. The major goal of model construction is to create a system or model that provideshighaccuracythroughwhichonecanassurethat

[5] P.Ramprakash, R.Sarumathi, R.Mowriya, S.Nithyavishnupriya“HeartDiseasePredictionUsingDeep Neural Network” 2020 International Conference on Inventive Computation Technologies(ICICT) DOI: 10.1109/ICICT48043.2020.9112443

[6] M.Snehith Raja, M.Anurag, Ch,Prachetan Reddy, NageshwaraRao Sriisala “Machine Learning-Based Heart DiseasePredictionSystem”2021InternationalConference on Computer Communication and Informatics(ICCI) DOI: 10.1109/ICCC|50826.2021.9402653

[7] Savitha Kamalapurkar, Samyama Gunjal G H “Online portal for prediction of Heart Disease using Machine Learning Ensemble Method” 2020 IEEE Bangalore HumanitarianTechnologyConference(B-HTC)

[8] Shaik Farzana, Duggineni Veeraiah “Dynamic Heart Disease Prediction using Multi-Machine Learning Techniques” 2020 5th International Conference on Computing, Communication and Security (ICCCS) DOI: 10.1109/ICCCS49678.2020.9277165.

[9] Rakibul Islam; Abhijit Reddy Beeravolu; Md. Al Habib Islam “A Performance Based Study on Deep Learning AlgorithmsintheEfficientPredictionofHeartDisease”2021

2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page974

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

Volume: 09 Issue: 09 | Sep 2022 www.irjet.net p-ISSN: 2395-0072

2nd International Informatics and Software Engineering Conference (IISEC) DOI: 10.1109/IISEC54230.2021.9672415.

[10] Ufaq Jeelani khan, Dr.Ashish Oberoi, Dr.Jasmeen Gill “Hybrid Classfication for Heart Disease Prediction using Artificial Intelligence” , Fifth International Conference on Computing Methodologies and Communication (ICCMC 2021).

[11] Surai Shinde; Juan Carlos Martinez-Ovando, “Heart DiseaseDetectionwithDeepLearningUsingaCombination of Multiple Input Sources” 2021 IEEE Fifth Ecuador Technical Chapters Meeting (ETCM) , DOI: 10.1109/ETCM53643.2021.9590672.

[12]SnehalB.Gavande,PramilM.Chawan,“HeartDisease Prediction using Deep Learning Techniques”, Volume: 08 Issue:11|Nov2021.

BIOGRAPHIES

Snehal B. Gavande, MTech Computer Engineering. VJTIMumbai.

which has played a keyrole in improving the teachinglearning process at VJTI.Awarded by SIESRP withInnovative & Dedicated Educationalist Award Specialization : Computer Engineering & I.T. in 2020 AD Scientific Index Ranking (World Scientist and University Ranking2022)– 2ndRank-BestScientist,VJTIComputer Science domain1138th Rank- Best Scientist, Computer Science,India.

Prof.PramilaM.Chawan,isworkinhg as an Associate Professor in the ComputerEngineeringDepartmentof VJTI, Mumbai. She has done her B.E.(ComputerEngineering) and M.E.(ComputerEngineering) from VJTICollegeofEngineering,Mumbai University.She has 28 years of teachingexperienceandhasguided85+M.Tech.projects and130+B.Tech.projects.Shehaspublished143papersin the InternationalJournals, 20 papers in theNational/InternationalConferences/ Symposiums. ShehasworkedasanOrganizingCommitteememberfor 25International Conferences and 5 AICTE/MHRD sponsoredWorkshops/STTPs/FDPs.Shehasparticipated in 16National/InternationalConferences. Worked as Consulting Editor on – JEECER, JETR,JETMS, Technology Today, JAM&AER Engg. Today, The Tech. World Editor –Journals of ADRReviewer -IJEF, Inderscience She has workedasNBACoordinatoroftheComputerEngineering DepartmentofVJTIfor5years.Shehadwrittenaproposal underTEQIP-IinJune2004for‘CreatingCentralComputing Facility at VJTI’. Rs.Eight Crore were sanctioned bythe World Bank under TEQIP-I onthis proposal. Central ComputingFacility was set up at VJTI throughthis fund

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