Deep Learning-based Blood Group Classification from Fingerprint Biometrics

The precise classification of blood types is essential for ensuring safe transfusions and improving clinical diagnostics. Traditional methods, including those that rely on manual testing or mathematical-based simulations, are often time-consuming and susceptible to human error. To overcome these limitations, we propose an automated classification system based on deep learning architectures. This system leverages convolutional neural networks (CNN) to capture spatial features, residual neural networks (ResNet) to enhance feature extraction through residual connections and densely connected convolutional networks (DenseNet) to ensure efficient information flow. The model also incorporates visual geometry group network 19 (VGG19) for hierarchical feature learning. Our approach accurately classifies blood types based on the ABO system using fingerprint biometric patterns, thereby reducing the need for manual interpretation. This solution minimizes diagnostic errors, streamlines laboratory workflows, and enhances medical efficiency, ultimately supporting safer patient outcomes.