SPARSE OPTIMIZATION TECHNIQUES FOR HIGH-DIMENSIONAL GENOMIC DATA ANALYSIS

The explosion of genomic data worldwide poses a daunting challenge for scientists seeking meaningful insights, especially in resource-constrained contexts like Ghana. Between 2020 and 2024, sparse optimization techniques such as Lasso Regression, Elastic Net, and Principal Component Analysis became indispensable tools for high-dimensional genomic data analysis in Ghana, addressing limitations of traditional methods. This study aimed to investigate how sparse optimization improves genomic data analysis performance by focusing on classification accuracy, robustness, biological interpretability, and computational efficiency, while considering moderating data characteristics like sample size and noise level. Employing a quantitative secondary data analysis design, the study reviewed 23 Ghanaian genomic studies, using correlation and regression analyses to validate the relationships. Major findings revealed that sparse optimization techniques significantly enhanced genomic analysis, with mean classification accuracy reaching 81.9% and average AUROC at 0.83, while data characteristics negatively impacted performance (correlation coefficient r = -0.445). Regression results showed feature selection algorithms had the strongest positive effect (β = 0.368), with an overall model R² of 0.712. These results demonstrate that integrating sparse optimization leads to substantial improvements in genomic research outputs even in low-resource settings. Consequently, the study recommends the broader institutional adoption of sparse methods, investment in computational infrastructure, and continuous training to fully unlock the benefits of genomic analytics in Ghana and similar contexts.