OPTIMIZED DEEP NEURAL NETWORKS FOR HIGH DIMENSIONAL RNA SEQ GENE EXPRESSION ANALYSIS IN CANCER DIAGNOSIS

Abstract

Cancer remains one of the leading causes of death worldwide, requiring accurate and early diagnostic techniques for effective treatment and improved patient survival. Recent advancements in Ribonucleic Acid sequencing technology have enabled the generation of large scale gene expression datasets that provide critical insights into cancer biology and molecular mechanisms. The high dimensionality of Ribonucleic Acid Sequencing data, which often contains thousands of gene features with relatively few samples, presents significant challenges such as overfitting, increased computational complexity, and reduced model performance when using traditional analytical methods. This study focused on the development of an optimized Deep Neural Network for high dimensional Ribonucleic Acid Sequencing gene expression analysis in cancer diagnosis. The approach integrates data preprocessing, feature selection, dimensionality reduction, and hyperparameter optimization to enhance classification accuracy and computational efficiency. Techniques such as Principal Component Analysis and Recursive Feature Elimination were applied to extract the most relevant gene features, while optimization strategies improved model convergence and generalization. The optimized Deep Neural Network was evaluated using standard performance metrics including accuracy, precision, recall, F1-score, and Receiver Operating Characteristic Area Under the Curve. Experimental results demonstrate that the proposed model achieves high classification performance and outperforms conventional machine learning algorithms in identifying cancer related gene expression patterns. The findings confirmed that optimized deep learning approaches are highly effective for analyzing complex genomic data and have strong potential for improving cancer diagnosis and supporting precision medicine.