Deciphering Key Genes in Colon Cancer Through Deep Learning Techniques

Identification and prognostication of colon cancer hold pivotal significance for biomedical research studies. Cancer is a genetically related disease in which defective genes are prone to making changes in expression. So, detecting colon cancer at an early stage can contribute to increasing patient survival rates. In recent years, computer-aided diagnosis systems utilizing Deep Learning techniques have emerged for the accurate diagnosis of colon cancer in the healthcare sector. In this methodology, microarray analysis is utilized to reveal the characteristics of normal and diseased genes. In the research work, high-dimensional microarray colon cancer data are utilized to detect the cancer disease. The primary function of the research work is to develop a model using Deep Learning techniques to identify key genes associated with colon cancer and predict the disease. An unsupervised Autoassociator Dimensionality Reduction Technique (ADRT), is applied to identify significant biomarker genes while filtering out redundant and noisy genes present in the gene data. To predict colon cancer, Deep Neural Network classifier is utilized on the selected genes and the classifier’s performance is calculated using evaluation metrics. Finally, comparative analyses are performed before and after applying the dimensionality reduction technique. The experimental outcomes demonstrate that the research model after applying ADRT, significantly improves the detection analysis in terms of minimizing the dimensionality of the gene data and increasing the accuracy value. Moreover, the research model can be smoothly integrated into medical healthcare systems, facilitating the precise diagnosis of Colon cancer.