Article: Enhancing healthcare data classification with deep edited nearest neighbours ensemble learning systems Journal: International Journal of Internet Manufacturing and Services (IJIMS) 2025 Vol.11 No.4 pp.357 - 381 Abstract: The prevalence of unbalanced datasets in medical diagnostics requires strong methods to improve prediction model accuracy and generalisability. The revolutionary deep edited closest neighbours ensemble (DENNE) framework addresses these difficulties by integrating deep learning and augmented closest neighbour approaches. This study tested DENNE against SMOTE and other conventional resampling methods. The approaches are tested using heart disease, missed abortion, and diabetes datasets. A hybrid edited closest neighbours approach refines input for an ensemble of classifiers after an AutoEncoder extracts features efficiently. This balances class disparities. The new strategy of DENNE combines complex feature transformation with improved noise reduction to increase prediction accuracy. DENNE significantly surpasses existing approaches in accuracy, precision, recall, and F1-scores across all datasets, according to comprehensive testing. These findings imply that DENNE effectively tackles medical dataset imbalance and sets a new threshold for medical diagnostic reliability, which has major significance for predictive healthcare analytics research. Inderscience Publishers - linking academia, business and industry through research

Title: Enhancing healthcare data classification with deep edited nearest neighbours ensemble learning systems

Authors: M. Kavitha; M. Kasthuri

Addresses: Department of Computer Science, Bishop Heber College, Tiruchirappalli – 620017, Tamil Nadu, India; Affiliated to: Bharathidasan University, India ' Department of Computer Science, Bishop Heber College, Tiruchirappalli – 620017, Tamil Nadu, India; Affiliated to: Bharathidasan University, India

Abstract: The prevalence of unbalanced datasets in medical diagnostics requires strong methods to improve prediction model accuracy and generalisability. The revolutionary deep edited closest neighbours ensemble (DENNE) framework addresses these difficulties by integrating deep learning and augmented closest neighbour approaches. This study tested DENNE against SMOTE and other conventional resampling methods. The approaches are tested using heart disease, missed abortion, and diabetes datasets. A hybrid edited closest neighbours approach refines input for an ensemble of classifiers after an AutoEncoder extracts features efficiently. This balances class disparities. The new strategy of DENNE combines complex feature transformation with improved noise reduction to increase prediction accuracy. DENNE significantly surpasses existing approaches in accuracy, precision, recall, and F1-scores across all datasets, according to comprehensive testing. These findings imply that DENNE effectively tackles medical dataset imbalance and sets a new threshold for medical diagnostic reliability, which has major significance for predictive healthcare analytics research.

Keywords: imbalanced medical data classification; deep learning; edited nearest neighbours; ENN; ensemble learning and AutoEncoders; medical diagnostics; treatment planning.

DOI: 10.1504/IJIMS.2025.150846

International Journal of Internet Manufacturing and Services, 2025 Vol.11 No.4, pp.357 - 381

Received: 30 Dec 2024
Accepted: 15 Mar 2025
Published online: 24 Dec 2025 *

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

Title: Enhancing healthcare data classification with deep edited nearest neighbours ensemble learning systems

Keep up-to-date