Novel Hybrid Feature Selection Using Binary Portia Spider Optimization Algorithm and Fast mRMR.

Area of Science:

• Oncology
• Computer Science
• Bioinformatics

Background:

• Cancer mortality rates are rising, highlighting the critical need for accurate early-stage diagnosis to improve patient outcomes.
• Machine learning algorithms applied to primary cancer datasets show promise for achieving diagnostic accuracy.
• Existing diagnostic methods require enhancement to meet the growing challenge of cancer mortality.

Purpose of the Study:

• To develop an innovative cancer classification technique utilizing machine learning.
• To improve the accuracy and efficiency of cancer diagnosis for better prognosis.
• To address the limitations of current diagnostic approaches in combating rising cancer death rates.

Main Methods:

• A novel cancer classification technique combining fast minimum redundancy-maximum relevance (mRMR) feature selection with the Binary Portia Spider Optimization Algorithm (BPSOA).
• Optimization of selected features using the fast mRMR and BPSOA.
• Validation of the optimized features using various classifiers: Support Vector Machine, Weighted Support Vector Machine, Extreme Gradient Boosting, Adaptive Boosting, and Random Forest.

Main Results:

• The proposed FmRMR-BPSOA methodology achieved a highest classification accuracy of 99.79% on six challenging cancer datasets.
• Empirical analysis confirmed the effectiveness and high performance of the developed model.
• The results demonstrate superior classification efficiency compared to existing methods.

Conclusions:

• The proposed FmRMR-BPSOA model offers a highly efficient and precise method for cancer diagnosis.
• This advanced technique holds significant promise for real-world medical applications and improving patient survival rates.
• The study underscores the importance of developing advanced computational tools for accurate and timely cancer detection.