Fingerprint based Blood Group Detection Using Machine Learning

This research investigates a new method of blood group prediction based on fingerprint patterns processed by Support Vector Classifiers (SVC). Conventional serological techniques are equipment-intensive and time-consuming, while finger prints provide a non-invasive, convenient biometric solution. The system extracts spatial features such as ridge frequency, minutiae points, and texture by applying pre-processing techniques such as Gabor filtering and normalization. These characteristics are subsequently utilized to train an SVC model on a blood group A, B, AB, O-labelled dataset. The outcomes reveal strong blood group correlations with fingerprint patterns (such as whorls and loops), which suggest very high predictive accuracy. This approach is a cost-saving, effective alternative to blood typing, particularly in emergency or resource-constrained environments. Future enhancements can involve greater datasets, integrating deep learning, and broader analysis to take into account gender and geographical fingerprint differences.