Original scientific paper
FROM HEART MURMUR TO ECHOCARDIOGRAPHY CONGENITAL HEART DEFECTS DIAGNOSTICS USING MACHINELEARNING ALGORITHMS
Edin Begic
orcid.org/0000-0001-6842-262X
; Department of Cardiology, General Hospital "Prim. Dr. Abdulah Nakas", Sarajevo, Bosnia and Herzegovina ; School of Medicine, Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina
Lejla Gurbeta Pokvic
; International Burch University, Sarajevo, Bosnia and Herzegovina ; Medical Device Inspection Laboratory Verlab Ltd. Sarajevo, Bosnia and Herzegovina
Zijo Begic
; Pediatric Clinic, Clinical Center University of Sarajevo, Sarajevo, Bosnia and Herzegovina
Nedim Begic
; Pediatric Clinic, Clinical Center University of Sarajevo, Sarajevo, Bosnia and Herzegovina
Mirza Dedic
; Faculty of Pharmacy, University of Sarajevo, Sarajevo, Bosnia and Herzegovina
Denis Mrsic
; Clinic for internal diseases, University Clinical Center Tuzla, Tuzla, Bosnia and Herzegovina
Mesud Jamakovic
; School of Medicine, Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina
Naim Vila
; School of Medicine, Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina
Almir Badnjevic
; International Burch University, Sarajevo, Bosnia and Herzegovina ; Medical Device Inspection Laboratory Verlab Ltd. Sarajevo, Bosnia and Herzegovina; Faculty of Pharmacy, University of Sarajevo, Sarajevo, Bosnia and Herzegovina
Abstract
The most common clinical sign in pediatric cardiology is heart murmur, which can often be uncharacteristic. The aim of this
research was to present the results of development of a classifier based on machine learning algorithms whose purpose is to classify
organic murmur that occur in congenital heart defect (CHD). The study is based on the data collected at Pediatric Clinic, Clinical
Center University of Sarajevo during three-year period. Totally, 116 children aged from 1 to 15 years were enrolled in the study.
Input parameters for classification are parameters obtained during basic physical examination and assessment of patient. First,
analysis of relevance of the feature for classification was done using InfoGain, GainRatio, Relief and Correlation method. In the
second step, classifiers based on Naive Bayes, Logistic Regression, Decision Tree, Random Forest and Support Vector Machine were
developed and compared by performance. The results of this research suggest that high accuracy (>90%) classifier for detection of
CHD based on 16 parameters can be developed. Such classifier with appropriate user interface would be valuable diagnostic aid to
doctors and pediatricians at primary healthcare level for diagnostic of heart murmurs.
Keywords
congenital heart defect; heart murmur; pediatrics; screening; machine learning; classifier
Hrčak ID:
272946
URI
Publication date:
8.2.2022.
Visits: 935 *