Blood group identification is essential for medical diagnosis, emergency treatment, and transfusion safety. Traditional serological testing requires blood samples, laboratory reagents, and trained personnel, which may not be accessible in timecritical or resource-limited environments. Recent research suggests a correlation between fingerprint ridge patterns and genetically inherited traits, including blood groups. This paper proposes a deep learning–based, non-invasive method for predicting a person's blood group using fingerprint images. Convolutional Neural Networks (CNNs) are trained on pre-processed fingerprint datasets to classify individuals into eight major blood groups (A, B, AB, O, Rh +/–). The approach aims to reduce dependency on invasive procedures and provide rapid, portable, and automated blood group identification. Experimental results show promising accuracy, demonstrating the feasibility of fingerprint-based biometric prediction.