Title: Enhancing healthcare predictions with deep learning: insights from image datasets

Authors: Wan Aezwani Wan Abu Bakar; Muhammad Amierusyahmi Bin Zuhairi; Mustafa Bin Man; Nur Laila Najwa Bt Josdi

Addresses: Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin, Besut Campus, 22200, Besut, Terengganu, Malaysia ' Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin, Besut Campus, 22200, Besut, Terengganu, Malaysia ' Faculty of Computer Science and Mathematics, Universiti Malaysia Terengganu, 21030, Kuala Nerus, Terengganu, Malaysia ' Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin, Besut Campus, 22200, Besut, Terengganu, Malaysia

Abstract: This study builds on prior research to improve healthcare predictions using deep learning with image datasets. Unlike numerical data, image processing in deep learning faces challenges such as large data volume, storage demands, computational resource needs, manual annotation, class imbalance, overfitting, and scalability issues. Effective solutions require robust preprocessing, efficient computation, thoughtful model design, and ethical considerations. This paper presents a 3-layer deep convolutional neural network (DCNN) to integrate image datasets, achieving 99% accuracy on benchmark datasets, including the brain tumor medical dataset (BTMD). The model employs dropout regularisation and incorporates numeric data insights, showcasing adaptability across different healthcare data types. These results highlight the significant potential of DCNNs for high-accuracy predictions in medical applications.

Keywords: image dataset; DCNN; deep convolutional neural network; BTMD; brain tumour medical dataset; prediction accuracy; healthcare applications; healthcare prediction; deep learning.

DOI: 10.1504/IJDATS.2025.147518

International Journal of Data Analysis Techniques and Strategies, 2025 Vol.17 No.2, pp.107 - 120

Received: 26 Feb 2024
Accepted: 24 Jun 2024
Published online: 20 Jul 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article