Synergistic feature selection and distributed classification framework for high-dimensional medical data analysis

D Dhinakaran ¹, L Srinivasan ², S Edwin Raja ¹

⁰Department of Computer Science and Engineering, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology, Chennai, India.

Methodsx +

|

March 14, 2025

View abstract on PubMed

Summary

A new algorithm, Synergistic Kruskal-RFE Selector and Distributed Multi-Kernel Classification Framework (SKR-DMKCF), enhances medical data analysis by significantly reducing features and improving classification accuracy. This method offers better efficiency and scalability for complex datasets.

Area Of Science

Medical Data Analysis
Machine Learning
Computational Biology

Background

Medical datasets are large and complex, leading to computational challenges, memory limitations, and reduced classification accuracy.
Effective feature selection and classification are crucial for accurate medical data interpretation and decision-making.

Purpose Of The Study

To introduce an integrated algorithm, Synergistic Kruskal-RFE Selector and Distributed Multi-Kernel Classification Framework (SKR-DMKCF), to address limitations in medical data analysis.
To improve dimensionality reduction, feature preservation, and classification performance in complex medical datasets.

Main Methods

Developed the Synergistic Kruskal-RFE Selector and Distributed Multi-Kernel Classification Framework (SKR-DMKCF).
Utilized recursive feature elimination and multi-kernel classification in a distributed environment.
Evaluated the algorithm on four diverse medical datasets.

Main Results

Achieved an average feature reduction ratio of 89% with SKR-DMKCF.
Obtained an average classification accuracy of 85.3%, precision of 81.5%, and recall of 84.7%.
Demonstrated a 25% reduction in memory usage and significant speed-up compared to existing methods.

Conclusions

SKR-DMKCF effectively reduces dimensionality while preserving critical data characteristics.
The proposed framework offers superior classification accuracy and computational efficiency, ensuring scalability for resource-limited environments.

Keywords:

Distributed computing Feature selection Medical data analysis Recursive feature elimination Synergistic Kruskal-RFE Selector and Distributed Multi-Kernel Classification Framework (SKR-DMKCF) and Classification