This paper presents a new binary version of the Spider Wasp Optimizer (BSWO), called Binary Boosted SWO (BBSWO), to address the challenge of high-dimensional feature selection (HFS). The original BSWO had several shortcomings, including early convergence, local optima, limited exploration and exploitation, and lack of population diversity. To overcome these issues, BBSWO incorporates chaos optimization, a new convergence parameter, multiple exploration mechanisms, and quantum-based optimization to enhance the search process. The proposed BBSWO not only finds the most suitable feature subset but also reduces data redundancy. It was evaluated on 23 HFS problems from the biomedical domain using the k-Nearest Neighbor (k-NN) classifier. The results showed that BBSWO outperformed traditional BSWO and other meta-heuristics in terms of classification accuracy. The paper also discusses the challenges of HFS, including the high dimensionality of data, the complexity of feature interactions, and the need for efficient feature selection methods. It highlights the importance of FS in improving learning performance by reducing dimensionality and removing redundant features. The paper reviews various FS approaches, including filtering, wrapper, and embedding methods, and discusses the limitations of traditional FS techniques. It also introduces meta-heuristic algorithms, which are effective for solving complex optimization problems. The paper concludes that BBSWO is a promising approach for HFS due to its ability to balance exploration and exploitation, enhance search diversity, and achieve high classification accuracy.This paper presents a new binary version of the Spider Wasp Optimizer (BSWO), called Binary Boosted SWO (BBSWO), to address the challenge of high-dimensional feature selection (HFS). The original BSWO had several shortcomings, including early convergence, local optima, limited exploration and exploitation, and lack of population diversity. To overcome these issues, BBSWO incorporates chaos optimization, a new convergence parameter, multiple exploration mechanisms, and quantum-based optimization to enhance the search process. The proposed BBSWO not only finds the most suitable feature subset but also reduces data redundancy. It was evaluated on 23 HFS problems from the biomedical domain using the k-Nearest Neighbor (k-NN) classifier. The results showed that BBSWO outperformed traditional BSWO and other meta-heuristics in terms of classification accuracy. The paper also discusses the challenges of HFS, including the high dimensionality of data, the complexity of feature interactions, and the need for efficient feature selection methods. It highlights the importance of FS in improving learning performance by reducing dimensionality and removing redundant features. The paper reviews various FS approaches, including filtering, wrapper, and embedding methods, and discusses the limitations of traditional FS techniques. It also introduces meta-heuristic algorithms, which are effective for solving complex optimization problems. The paper concludes that BBSWO is a promising approach for HFS due to its ability to balance exploration and exploitation, enhance search diversity, and achieve high classification accuracy.