Enhancing Genetic Marker Identification for Disease Prediction: Deep Learning vs. Traditional Machine Learning Approaches

Accurate identification of genetic markers plays a vital role in early disease diagnosis and personalized healthcare. Genomic data disease prediction has seen extensive use of traditional machine learning methods like Logistic Regression, which are characterized by their straightforwardness and ease of interpretation. Nonetheless, these models often require manual feature engineering and may struggle to capture the complex, nonlinear patterns that are characteristic of high-dimensional genetic data. In contrast, methods based on deep learning (DL) like Convolutional Neural Networks (CNNs) offer a robust alternative by automatically acquiring feature representations from raw input data. This study provides a comparative analysis of CNN and Logistic Regression in the context of identifying genetic markers and predicting diseases. The CNN model demonstrates a significant advantage over Logistic Regression regarding predictive accuracy and robustness against noisy genetic inputs, as evidenced by experimental results. Even though CNN has a greater computational complexity, it shows considerable promise for revealing complex patterns in genomic datasets, underscoring its importance for the progress of precision medicine.