Integrating gene selection and deep learning for enhanced Autisms' disease prediction: a comparative study using microarray data

In this article, Autism Spectrum Disorder (ASD) is discussed, with an emphasis placed on the multidimensional nature of the disorder, which is anchored in genetic and neurological components. Identifying genes related to ASD is essential to comprehend the mechanisms that underlie the illness, yet th...

Full description

Saved in:
Bibliographic Details
Published in:AIMS mathematics Vol. 9; no. 7; pp. 17827 - 17846
Main Authors: Abdelwahab, Mahmoud M., Al-Karawi, Khamis A., Semary, H. E.
Format: Journal Article
Language:English
Published: AIMS Press 01.01.2024
Subjects:
autism spectrum disorder
convolutional neural networks
gene selection
microarray technology
ISSN:2473-6988, 2473-6988
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this article, Autism Spectrum Disorder (ASD) is discussed, with an emphasis placed on the multidimensional nature of the disorder, which is anchored in genetic and neurological components. Identifying genes related to ASD is essential to comprehend the mechanisms that underlie the illness, yet the condition's complexity has impeded precise information in this field. In ASD research, the analysis of gene expression data helps choose and categorize significant genes. The study used microarray data to provide a novel approach that integrated gene selection techniques with deep learning models to improve the accuracy of ASD prediction. It offered a detailed comparative examination of gene selection approaches and deep learning architectures, including singular value decompositions (SVD), principal component analyses (PCA), and convolutional neural networks (CNNs). This paper combines gene selection methods (PCA and SVD) with deep learning models (CNN) to improve ASD prediction. Compared to more traditional approaches, the study revealed that its integrated methodology was more effective in improving the accuracy of ASD prediction results through experimentation. There was a difference in the accuracy between the PCA-CNN model, which achieved 94.33% with a loss of 0.4312, and the SVD-CNN model, which achieved 92.21% with a loss less than or equal to 0.3354. These discoveries help in the development of more accurate diagnostic and prognostic tools for ASD, which is a complicated neurodevelopmental disorder. Additionally, they provide insights into the molecular pathways that underlie ASD.
ISSN:2473-6988
2473-6988
DOI:10.3934/math.2024867