Applications of Machine Learning in SARS-COV-2 Genomics for Varient Detection and Disease Prediction

SARS-CoV-2 variant detection and disease prediction have become critical areas of research, driven by the need for accurate diagnostics and outbreak management. Machine learning (ML) addresses these challenges by enabling robust genomic analysis and predictive modeling. This study presents a comprehensive machine learning framework, starting with advanced preprocessing techniques, including K-mers, one-hot encoding, and Principal Component Analysis (PCA), to ensure high-quality input data and effective dimensionality reduction. For feature extraction, the framework integrates a Dual-Crossed Squeeze Net and a Self-Adaptive Harris Hawk Optimization (SA-HHO) algorithm with a lookahead optimizer to enhance feature learning and representation. The classification stage employs Support Vector Machine (SVM) and Random Forest models, combined using a majority voting mechanism to achieve robust and accurate predictions. The proposed framework demonstrates exceptional performance with Accuracy (98.5432), Sensitivity (97.3210), and also other metrics with superior performance compared to existing methods.