EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Balancing dimensionality reduction and information retention in medical datasets using adaptive relevance scoring and trifold evolutionary optimization

Asha Shanmugham, Parvathy Meenakshi Sundaram

Journal of Biological Regulators and Homeostatic Agents · 27.07.2026 · 40 (3) · 126

DOI: 10.65746/jbrha126 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Biological Regulators and Homeostatic Agents, online erschienen am 27.07.2026. Verfasst von Asha Shanmugham und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

High-dimensional medical datasets often contain redundant, irrelevant, and noisy features that increase computational complexity. Existing approaches for dimensionality face several critical challenges: (i) aggressive feature reduction leading to loss of critical clinical information, (ii) selection unfairness in model-dependent methods such as Recursive Feature Elimination (RFE), (iii) reliance on variance in Principal Component Analysis (PCA) (iv) poor generalization of existing feature selection methods across diverse medical datasets and (v) uncertainty about whether retained features truly capture essential clinical information after DR. To address these limitations, this study proposes an adaptive feature selection framework that balances DR with effective information retention. The proposed method introduces a Gene Noise Filtering Layer along with entropy-based filtering, Mutual Information, and variance filtering to remove low-informative features. An Adaptive Relevance Score (ARS) is developed to identify clinically significant features. A Trifold Evolutionary Optimization (TEO) approach, integrating Particle Swarm Optimization, Quantum-inspired search, and genetic mutation, is employed to determine optimal feature subsets. Experimental evaluation achieves a Feature Stability Index (FSI) of 0.9819, indicating excellent preservation of informative features and high stability of the selected feature subsets. The proposed method improved classification accuracy, reduced dimensionality, and attained better preservation of clinically meaningful information across multiple medical datasets.

Zitieren

APA

Shanmugham, A., & Sundaram, P. M. (2026). Balancing dimensionality reduction and information retention in medical datasets using adaptive relevance scoring and trifold evolutionary optimization. Journal of Biological Regulators and Homeostatic Agents, 40(3), 126. https://doi.org/10.65746/jbrha126

Vancouver

Shanmugham A, Sundaram PM. Balancing dimensionality reduction and information retention in medical datasets using adaptive relevance scoring and trifold evolutionary optimization. Journal of Biological Regulators and Homeostatic Agents. 2026;40(3):126. doi:10.65746/jbrha126

BibTeX
@article{Shanmugham2026,
  author = {Shanmugham, Asha and Sundaram, Parvathy Meenakshi},
  title = {Balancing dimensionality reduction and information retention in medical datasets using adaptive relevance scoring and trifold evolutionary optimization},
  journal = {Journal of Biological Regulators and Homeostatic Agents},
  year = {2026},
  volume = {40},
  number = {3},
  pages = {126},
  doi = {10.65746/jbrha126}
}
Kostenlos starten →