Blood Group Identification from Fingerprints Using Convolutional Neural Network

Abstract

Identifying blood groups usually means needles, labs, and waiting for biochemical results-resources that are not always available in emergencies or remote areas. This project introduces a non-invasive, intelligent alternative: predicting human blood groups directly from fingerprint ridge patterns and textures using a Convolutional Neural Network (CNN). By shifting the process from chemical analysis to biometric deep learning, we can classify fingerprints into the eight primary blood groups (A+, A-, B+, B-, O+, O-, AB+, AB-) almost instantly. We trained the CNN model on labeled datasets and integrated it into a responsive web interface built with Flask, HTML, and CSS. Users simply upload a fingerprint scan to receive an immediate prediction and confidence score. This system demonstrates how merging biometric analysis with deep learning can provide a rapid, preliminary screening tool, potentially saving critical time in resource-limited environments.

Introduction

The text discusses a deep learning-based system that predicts a person’s blood group using fingerprint analysis instead of traditional blood testing methods. Conventional blood typing requires blood samples, laboratory equipment, trained professionals, and time-consuming procedures. To provide a faster, non-invasive, and cost-effective alternative, researchers have explored the relationship between fingerprint patterns and genetically inherited blood groups such as ABO and Rh types.

Fingerprints are unique, stable throughout life, and formed under genetic and environmental influences during fetal development. Since blood groups are also genetically determined, researchers believe subtle relationships may exist between fingerprint ridge patterns and blood types. Modern deep learning techniques, particularly Convolutional Neural Networks (CNNs), can detect hidden and complex patterns in fingerprint images that are difficult for humans to identify manually.

The proposed system uses CNNs to automatically extract fingerprint features such as ridge flow, texture, orientation, and minutiae points. A fingerprint image is uploaded through a Flask-based web application, where preprocessing steps like resizing, noise removal, ridge enhancement, and segmentation are performed. The CNN model then analyzes the processed image and predicts the most likely blood group along with a confidence score. This provides a quick, automated, and needle-free preliminary blood group identification method.

The study highlights several advantages of this approach, including reduced cost, faster results, non-invasive operation, and the use of existing fingerprint scanners already integrated into many devices. Such systems could support emergency healthcare, crime investigation, and identity verification. However, the research also acknowledges challenges such as limited scientific certainty regarding fingerprint-blood group correlations, dependence on large and diverse datasets, image quality issues, and ethical concerns related to biometric privacy and misuse of personal health information.

The literature survey traces the evolution of fingerprint-based blood group prediction from early dermatoglyphic studies to advanced machine learning and deep learning methods. Earlier rule-based and statistical approaches had limited accuracy and depended heavily on handcrafted features. Recent studies using CNNs, transfer learning models like VGG16 and ResNet50, hybrid architectures, and GAN-based image enhancement have significantly improved prediction performance, with some studies reporting over 90% accuracy.

The system architecture consists of a web interface, server, CNN-based analysis module, and database support. Users upload fingerprint images through a website, while the backend server preprocesses the images, runs the CNN model, and returns the prediction results. The methodology includes fingerprint image acquisition, preprocessing, feature extraction, classification, and blood group prediction. The CNN model learns automatically from labeled fingerprint datasets and improves its ability to classify blood groups through training.

Conclusion

The project demonstrates fingerprint images as a non-invasive accessible method for blood group prediction through the implementation of Convolutional Neural Networks and current web technologies. The system built by developers combines image preprocessing and deep learning classification with a Flask web interface to deliver complete end-to-end functionality. The CNN model developed through a comprehensive training and evaluation process demonstrates its ability to learn distinct fingerprint patterns which it uses to accurately determine blood group categories. The system demonstrates artificial intelligence applications which improve medical diagnostics through its ability to quickly and contact-free assess blood groups. The model serves as an additional testing solution which proves useful for emergency medical situations and remote healthcare environments and resource-limited facilities despite its design not to replace laboratory serological testing. The project establishes multiple pathways for future development because it supports dataset growth and deep learning model development and mobile platform deployment for immediate field applications. The project work demonstrates that using biometric imaging together with deep learning results in effective diagnostic solutions. The project establishes a vital foundation which will enable medical support systems to become more intelligent and faster and easier to access.

References

[1] P. Vaidya (2025). Fingerprint analysis for blood group determination. ScienceDirect. [2] S. Paudel (2025). A Dermatoglyphic Study of Primary Fingerprint Patterns in Relation to Blood Groups and Gender. IET Biomedical Engineering Journal (or similar). [3] T. Nihar, K. Yeswanth & K. Prabhakar. “Blood Group Determination Using Fingerprints.” MATEC Web of Conferences, 2024. [4] S. Manikandan, S. Sumathi, S. Thangadurai, et al. “Dermatoglyphics and Their Relationship with Blood Group.” Journal of Pharmacy &Bioallied Sciences, 2019 [5] “Fingerprint analysis for blood group determination.” P. Vaidya, 2025. ScienceDirect / (journal article). [6] Blood Group Detection from Fingerprint Using Machine Learning and Deep Learning Methodologies.” IJSREM, 2025. [7] Blood Group Classification from Fingerprints Using a Custom Convolutional Neural Network (CNN).” JETIR, 2025. [8] U. B. Ghaffar, et al. “Palmar Dermatoglyphic Pattern in Relation to Blood Group and Gender.” Journal of … 2024 [9] A Dermatoglyphic Study: Association of Fingerprint Patterns Among ABO Blood Groups.” T. P. Susmiarsih, M. S. Mustofa &Mirfat. Biosaintifika, 2016. [10] Study of Fingerprint Patterns in Relation to Different Blood Groups.” Devi et al., International Journal of Current Pharmaceutical Review & Research, 2024. [11] Surendra Kumar Sah, Samyog Mahat et al. (2023). In their study “Blood Group Determination Using Fingerprint” [12] “Blood Group Detection Using Fingerprint Images.” (2024). IJRASET. (Research article describing preprocessing and ML pipeline for fingerprint-to-blood-group). IJRASET [13] Yashas, D. R., Vinutha, H. N., Merlin, B., Soundarya, R., Chethan, V. (2025). Identification of an existent\'s blood group is pivotal in exigency situations, for identity authentication, and in population analysis. [14] M. Mondal, U. K. Suma, M. Katun, R. Biswas, Md. R. Islam (2019) This research, titled “Blood Group Identification Based on Fingerprint”, utilized 2D Discrete WaveletTransform (DWT) and Binary Transform techniques to extract key fingerprint features for blood group classification. [15] Patil, V., & colleagues. (2020). An association between fingerprint patterns with blood groups and lifestyle-based diseases. Journal of Pharmacy &Bio allied Sciences.

Copyright

Copyright © 2026 Mr. G. Ramesh, J. Rithika, M. Vamshi Krishna, P. Surya Vishal. 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.