Deep Learning Based Blood Group Prediction Using Fingerprint

Abstract

Blood group identification is an essential factor in medical diagnostics, transfusions, and emergency healthcare management. Conventional blood type involves invasive serological testing. This paper introduces a deep learning-based approach for a non-invasive method of blood group identification through statistical correlations using images of fingerprints. A Convolutional Neural Network (CNN) architecture is employed for this purpose. The system is based on a dataset of 6000 images of fingerprints categorized into eight classes of blood groups: A+, A-, B+, B-, AB+, AB-, O+, and O-. The system is based on a combination of image processing, ridge pattern detection, and SoftMax classifier within a web application framework. The feasibility of this approach is demonstrated through experimental evaluation using deep learning architectures for blood group identification through images of fingerprints. Although this approach does not replace conventional blood type, it is useful for exploratory diagnostics for research purposes.

Introduction

Blood groups are important in healthcare for safe blood transfusions, organ transplants, and prenatal care. They are determined mainly by the ABO and Rh systems, and incorrect transfusions can cause serious health complications. Traditional blood group detection methods use serological tests such as slide tests and tube tests, which require blood samples and controlled laboratory conditions. Although accurate, these methods are invasive, time-consuming, and sometimes difficult to perform in emergency situations. Molecular techniques like PCR provide high precision but are expensive and require advanced equipment.

To overcome these limitations, researchers are exploring non-invasive diagnostic methods. One such approach is using fingerprint patterns, which are genetically formed during fetal development and remain unchanged throughout life. Some studies suggest a statistical relationship between fingerprint patterns (loops, whorls, arches) and blood groups, although the biological link is not fully proven.

This study proposes a deep learning approach using Convolutional Neural Networks (CNNs) to predict blood groups from fingerprint images. A dataset of 6000 labeled fingerprint images was collected and preprocessed by converting them to grayscale, resizing them, and applying data augmentation. The CNN model extracts ridge features from fingerprints and classifies them into eight blood group categories (A, B, AB, O with Rh ±).

The system architecture includes fingerprint collection, image preprocessing, feature extraction using CNN, model training, evaluation, and deployment through a web application for real-time prediction. The proposed algorithm processes fingerprint images, extracts features, classifies them, and outputs the predicted blood group with confidence.

Experimental results show that the proposed CNN-based model achieved an accuracy of 91.4%, which is higher than many existing techniques such as image processing, spectrophotometric detection, and statistical fingerprint analysis. The method offers a non-invasive, cost-effective, and automated solution for blood group prediction, though further research with larger datasets is needed to improve reliability.

Conclusion

In this paper, a deep learning-based framework has been proposed for blood group prediction based on fingerprints using Convolutional Neural Networks (CNNs). In the proposed framework, all the aspects related to fingerprint-based blood group prediction are included in a single framework. Unlike classical blood group prediction techniques that involve invasive serological tests, the proposed framework has focused on exploring a non-invasive blood group prediction framework. In addition, the experimental results have also been included to prove that the proposed framework using CNNs is better in terms of accuracy when compared to classical image processing and statistical correlation-based blood group prediction techniques. Moreover, the automatic feature learning capability of the proposed framework using CNNs will help in avoiding the feature extraction process, thus providing better scalability to the proposed framework. In addition, the proposed framework can be easily implemented as a web interface. While the results of the suggested method are promising, the area of fingerprint-based blood group prediction is in the research stage and needs to be validated further. Some of the future research directions may be the application of attention mechanisms and the application of diverse and explainable data. Overall, the research paper demonstrates the potential application of deep learning techniques in the development of medical prediction systems with the help of biometric information. Apart from this, this study also has the potential to utilize the integration of biometric and artificial intelligence to assess the possibilities of non-traditional medical predictive methods. Even though the relationship between fingerprint patterns and blood groups needs further scientific investigation and research, the suggested system has the potential to utilize the predictive analysis with the help of dermatoglyphic features through deep learning.

References

[1] Ravindran G et al., “Determination and classification of blood types using image processing techniques”, Int J Comput Appl, 2017. [2] Fernandes J et al., “A complete blood typing device for automatic agglutination detection”, IEEE Trans InstrMeasur, 2015. [3] Pimenta S et al., “Development of an automatic electronic system to human blood typing”, IEEE EMBC, 2012. [4] Alshafie SEM et al., \"Types of Fingerprints Characteristics and their Association with Gender and Blood Groups in Sudan\", Integrative Journal of Medical Science. [5] Ramrekh Dhaker et al., \"Study of Fingerprint Pattern in Relation to Gender and Blood Group\", IJMBS, 2021. [6] Khalifa et al., \"Assessment of Fingerprint Pattern and Density in Relation to Blood Groups\", Egypt J Forensic Sci, 2021. [7] Kumari P et al., “Correlation between Fingerprint and Blood Group”, Ann Int Med Den Res, 2021. [8] Keerthana D, Ranganathan L, “Blood sample analyzer using intelligent machine vision techniques”, Int J Eng Sci Comput, 2017. [9] Dalvi A et al., “Determination of blood group using Image processing”, Int J Sci Eng Res, 2018. [10] N. Zaeri, \"Minutiae-based Fingerprint Extraction and Recognition\", IntechOpen, 2011. [11] A.K. Jain et al., “An identity authentication system using fingerprints,” Proc. IEEE, 1997. [12] W. Zafar et al., \"Minutiae based fingerprint matching techniques,\" IEEE INMIC, 2014. [13] D.SivaSundhara Raja &J.Abinaya, “Cost-Effective Blood Group Detection Using Fingerprints”, 2019. [14] Khan et al., “Survey of deep convolutional neural networks”, ArtifIntell Rev, 2020. [15] Alzubaidi et al., “Review of deep learning: concepts and CNN architectures”, J Big Data, 2021. [16] Singh et al., “Performance evaluation of deep learning models in medical imaging”, 2022. [17] Nur-A-Alam et al., “Automatic fingerprint identification using deep learning”, 2021.

Copyright

Copyright © 2026 Agastin S, Charles Y, Mr. S. Kandeeban, Dr. S. Akila. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.