Cross-Omics correlation and genomic pattern recognition for accurate tumor detection in oncological surgery decision-making

Abstract

Abstract A genomic sequence analysis is necessary to determine the cancer-specific genomic patterns for oncological surgery decision-making. However, the existing works overlooked the cross-omics interaction regarding Genomic Pattern Recognition (GPR). Therefore, Chebyshev-Excoffier Heun Density-Based Spatial Clustering of Application with Noise (CEHDBSCAN)-based GPR and Cross-Omics Correlation Analysis (COCA) using Heaviside Function-based Pearson Correlation Coefficient (HeFPCC) is proposed. Primarily, the patient registers on the hospital’s server. Then, the multi-omics data is collected and then encrypted using Legendre Chi Function-based Advanced Encryption Standard (LCFAES). Further, in the hospital server, the secured multi-omics data is decrypted and pre-processed. Afterward, the CEHDBSCAN-based GPR is performed. Then, the features are extracted, and the hidden mutation progress is analyzed using Jacobi Elliptic Function-Hidden Markov Model (JEFHMM). Also, the HeFPCC-based COCA is carried out. Meanwhile, the extracted features are selected using Hawk Fish Walsh Function Optimization Algorithm (HFWaFOA). Finally, the gradients are secured using Differential Privacy (DP); then, the tumor is detected using Deep Shifted and Scaled Sigmoid Alpha Dropout Learning Neural Network (DSSSADLNN). The classified output is further given for decision-making regarding oncological surgery. Thus, the proposed framework effectively recognized the genomic pattern and classified the tumor with an accuracy of 99.05%, showing better performance than existing works.