Tuberculous pleural effusion prediction using ant colony optimizer with grade-based search assisted support vector...

Area of Science:

• Medical diagnostics
• Computational biology
• Machine learning applications

Background:

• Tuberculous pleural effusion (TBPE) is an inflammatory condition resulting from tuberculosis infection.
• TBPE can lead to pleural adhesions and thickening, potentially impairing chest cavity mobility.

Purpose of the Study:

• To propose and evaluate the bGACO-SVM model for the adjunctive diagnosis of TBPE.
• To enhance diagnostic accuracy and classification capabilities for TBPE.

Main Methods:

• Development of the bGACO-SVM model, integrating an enhanced continuous ant colony optimization (ACOR) with a grade-based search technique (GACO) and support vector machine (SVM).
• Utilizing GACO for wrapped feature selection to improve classification performance and avoid local optimization.
• Comparative experiments were conducted between GACO and other algorithms to validate its effectiveness.

Main Results:

• The bGACO-SVM model demonstrated good diagnostic power in predicting TBPE.
• Experimental results confirmed the core advantages of GACO, particularly in classification capability.
• Evaluation on a TBPE dataset of 147 patients showed the bGACO-SVM method to be effective for accurate TBPE prediction.

Conclusions:

• The bGACO-SVM model is an effective technique for the accurate prediction of TBPE.
• The grade-based search in GACO significantly improves convergence and avoids local optima, enhancing diagnostic accuracy.
• This model offers a promising approach for the adjunctive diagnosis of tuberculous pleural effusion.