Modelling An Effectual Feature Selection Approach For Predicting Down Syndrome Using Machine Learning Approaches

Document Type : Primary Research paper

Authors

1 Assistant Professor-II, Computer Science and Engineering, Bannari Amman Institute of Technology, Erode, Tamil Nadu

2 Assistant Professor, Computer Science and Engineering, Bannari Amman Institute of Technology, Erode, Tamil Nadu

3 Assistant Professor Computer Science and Engineering, Bannari Amman Institute of Technology, Erode, Tamil Nadu

Abstract

Down Syndrome (DS) is a genetic disorder which is caused due to the occurrence
of third copy of chromosome 21. For DS, pre-natal screening is the primary component
and it is suggested to be offered universally to women irrespective of their background and
age. Machine Learning (ML) plays an essential role in predicting the severity of the
disease in earlier stage with the features related to DS. It gains a considerable attention in
performing predictive analysis for various medical applications. Therefore, the effectual
and appropriate diagnosis of DS is a significant challenge for medical practitioners and
experts. The ultimate target of this work is to initiate an accurate and non-invasive
diagnostic process for predicting DS and to reduce the cost of basic prenatal diagnosis. An
effectual ML approach is developed in this work to diagnose DS. Here, L1-norm based
Support Vector Regression (L1-SVR) for feature selection is applied for selecting the
highly related and appropriate features for accurate classification of DS from normal
people. The proposed L1-SVR generates a newer feature subset from the available dataset
based on its feature weighted value. The performance metrics like sensitivity, specificity,
accuracy, F1 score, precision are evaluated for evaluation. The optimal accuracy attained
with this finest subset of chosen features is due to diverse contributions of the DS features.
The experimental outcomes of this study recommend that the anticipated model is applied
for appropriate prediction of DS and can be applied for making proper decision during the
critical condition. Recently, computer aided decision support system plays a significant role
in assisting DS prediction. The proposed L1-norm SVM pretends to fulfill the gap among
the feature selection process and classification using the available data by properly
fulfilling the experimental design. The simulation is done with MATLAB simulation
environment

Keywords

Int. J. of Aquatic Science
Volume 12, Issue 3 - Serial Number 3
ICMMNT-2021 International Virtual Conference on Materials, Manufacturing and Nanotechnology, 30th June, 2021.
June 2021
Pages 1238-1249

Files

Share

How to cite

Statistics