Scholarlink

Modern research relies on computer-aided diagnostic (CAD) tools for efficient utilisation of time and resources. These technologies help medical practitioners make decisions by swiftly identifying anomalies and providing reliable information. The effective selection of features (characteristics) is crucial for these systems to appropriately categorise high-dimensional biological data. These traits may help diagnose linked medical issues. Biomedical data has many unnecessary or redundant information, making pattern identification difficult. Therefore, a feature selection (FS) technique should be implemented and used to remove these characteristics from the training data. Eliminating irrelevant or uninformative characteristics is the main objective of FS techniques to enhance classification accuracy along with reduction of training and testing time of machine learning (ML) classifiers. In ML-driven CAD systems, the FS phase is crucial to success. Adding efficient features during training may boost ML model performance. This empirical research proposes an FS-based biological data categorisation approach based on integration of Water Wave Optimization (WWO) soft-computing algorithm with efficient ML classifiers. Using the publicly accessible Wisconsin Diagnostic Breast Cancer (WDBC) standard dataset, this study classifies benign and malignant tumours. The 70:30 train-test split approach is implemented and the suggested strategy’s competency is assessed using accuracy, sensitivity (Recall), specificity, precision, F1-Score, Kappa Score, Matthews Correlation Coefficient (MCC), and ROC AUC Score. Best accuracy calculated using the proposed method is 97.96%. The suggested clinical decision support system has a good classification performance, making it a useful secondary opinion tool and lowering the workload of experienced medical physicians.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Efficient Approach for Feature Selection for Enhanced Performance in Breast Cancer Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Arin Bansal,
  • Vishal Kumar Lohan,
  • Munish Khanna,
  • Shantanu Agnihotri

摘要

Modern research relies on computer-aided diagnostic (CAD) tools for efficient utilisation of time and resources. These technologies help medical practitioners make decisions by swiftly identifying anomalies and providing reliable information. The effective selection of features (characteristics) is crucial for these systems to appropriately categorise high-dimensional biological data. These traits may help diagnose linked medical issues. Biomedical data has many unnecessary or redundant information, making pattern identification difficult. Therefore, a feature selection (FS) technique should be implemented and used to remove these characteristics from the training data. Eliminating irrelevant or uninformative characteristics is the main objective of FS techniques to enhance classification accuracy along with reduction of training and testing time of machine learning (ML) classifiers. In ML-driven CAD systems, the FS phase is crucial to success. Adding efficient features during training may boost ML model performance. This empirical research proposes an FS-based biological data categorisation approach based on integration of Water Wave Optimization (WWO) soft-computing algorithm with efficient ML classifiers. Using the publicly accessible Wisconsin Diagnostic Breast Cancer (WDBC) standard dataset, this study classifies benign and malignant tumours. The 70:30 train-test split approach is implemented and the suggested strategy’s competency is assessed using accuracy, sensitivity (Recall), specificity, precision, F1-Score, Kappa Score, Matthews Correlation Coefficient (MCC), and ROC AUC Score. Best accuracy calculated using the proposed method is 97.96%. The suggested clinical decision support system has a good classification performance, making it a useful secondary opinion tool and lowering the workload of experienced medical physicians.