Deep Genomic Signature Classifier for Covid-19 Strains

Abstract

The rapid growth of biomedical data and infectious disease monitoring has created a strong demand for intelligent systems capable of accurately predicting viral status based on patient and sequencing-related attributes. Early and precise detection of viral infections plays a crucial role in disease control, treatment planning, and public health management. However, traditional diagnostic approaches and conventional data analysis techniques often struggle to efficiently handle high-dimensional and heterogeneous datasets. Existing traditional systems primarily rely on statistical methods or single machine learning models for classification. While these approaches provide baseline performance, they often lack robustness, fail to capture complex non-linear relationships in data, and are sensitive to feature variations. Additionally, standalone models may not generalize well across diverse datasets, leading to reduced prediction accuracy and reliability. These limitations highlight the need for a more advanced, flexible, and data-driven approach. To address these challenges, the proposed system introduces a multimodel intelligent framework that integrates several machine learning (ML) algorithms such as Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Logistic Regression (LR), Decision Tree Classifier (DTC), Gradient Boosting Classifier (GBC), and Adaptive Boosting Classifier (ABC), alongside a Proposed Hybrid Deep Learning (PHDL) model based on a Dense Batch Normalization Serial Neural Network (DBN-SNN). In addition, an Optimal Rule List Classifier (ORLC), implemented as a RF based interpretable model, is incorporated to enhance decision-making. The system adopts a performance-based selection strategy, where both the PHDL model and ORLC are trained and evaluated, and the best-performing model is selected dynamically for final prediction. Furthermore, the system is deployed as a web-based application using Flask, enabling data upload, preprocessing, exploratory data analysis, model training, performance comparison, and batch prediction. This framework improves accuracy, scalability, and supports data-driven healthcare analytics.