Fingerprint-Based Biomarker Recognition for Non-Invasive Blood Group Identification

Abstract

In emergency medical situations, the rapid and accurate identification of an individual\'s blood group can be life-saving. Traditional methods for blood group testing, although reliable, require physical samples, laboratory processing, and time. This research investigates a novel, non-invasive approach to blood group classification using fingerprint patterns. The hypothesis is based on a potential correlation between dermatoglyphic features such as ridges and minutiae and blood group types. In this study, fingerprint images are captured and processed through image preprocessing techniques, followed by feature extraction using Convolutional Neural Networks (CNNs). A supervised learning classifier is then trained to categorize each fingerprint into its corresponding blood group (A, B, AB, or O). The proposed model demonstrates promising accuracy, indicating that biometric traits like fingerprints can be effectively utilized for blood group prediction. This approach holds significant potential to transform healthcare diagnostics by enabling faster, contactless blood group identification especially valuable in rural settings, accident scenarios, and emergency medical camps

Introduction

This study introduces a non-invasive, AI-driven method to predict an individual's blood group using fingerprint images, leveraging Convolutional Neural Networks (CNNs) and image processing techniques. The goal is to offer a fast, cost-effective, and accessible alternative to traditional blood tests, particularly valuable in emergency, rural, or resource-limited settings.

Motivation

• Traditional blood typing via serological methods is accurate but time-consuming and impractical in urgent situations.

• Fingerprint recognition is already used in security and identity systems; this research explores its novel medical application: correlating fingerprint features with blood group types.

Related Research

Past studies applied machine learning and image processing to blood group classification using blood sample images. Key methods:

• SVM, Decision Trees, Neural Networks, CNNs for prediction.

• Image preprocessing (e.g., contrast enhancement, noise reduction).

• Automated systems for fast and accurate blood group identification.

• Growing interest in combining AI with diagnostic imaging for real-time and privacy-aware healthcare solutions.

Proposed Methodology

1. Dataset

• Fingerprint images labeled with blood groups (A, B, AB, O) and Rh factors (+/−).

• Data sourced from public biometric repositories.

• Applied data augmentation (rotation, scaling, flipping) to enhance diversity and combat class imbalance.

2. Preprocessing

• Image resizing to 128×128 pixels.

• Noise reduction using Gaussian filtering.

• Contrast enhancement with histogram equalization.

• Normalization to scale pixel values [0,1].

3. CNN Architecture

• Convolutional layers detect fingerprint ridges and minutiae.

• ReLU activation for non-linearity.

• Pooling layers for dimensionality reduction and translation invariance.

• Dropout layers to prevent overfitting.

• Fully connected layers classify the fingerprint into eight blood group categories (A+, A−, B+, B−, AB+, AB−, O+, O−).

• Softmax activation in the output layer for multi-class classification.

4. Training

• Trained using TensorFlow/Keras on Google Colab with GPU support.

• Configuration:

  • Optimizer: Adam

  • Learning rate: 0.001

  • Batch size: 32

  • Epochs: 25

• Evaluation metrics: Accuracy, Precision, Recall, F1 Score, and Confusion Matrix.

• Early stopping used to avoid overfitting.

Results & Discussion

• Achieved 93.5% accuracy on test data.

• Precision (92.8%), Recall (91.9%), F1 Score (92.3%) show strong model reliability.

• Confusion matrix indicates accurate prediction across all 8 blood groups.

• Preprocessing and augmentation improved model robustness to fingerprint image variability.

• CNN proved effective in learning discriminative features from ridge patterns.

Limitations

• Dataset size was limited (800 samples), affecting generalizability.

• Fingerprints alone may not fully capture blood group-linked features; integrating other biometrics (e.g., palm veins) could boost accuracy.

• Performance may drop with low-quality or distorted fingerprints.

• Model interpretability and real-world applicability need further validation.

Practical Implications

• Can be deployed in emergency response, rural clinics, mobile health units.

• Enables faster triage and transfusion decisions.

• Non-invasive, low-cost, and scalable alternative to lab testing.

Ethical Considerations

• Biometric data must be anonymized, encrypted, and handled in compliance with privacy regulations.

• Future systems must include security protocols to safeguard sensitive information.

Conclusion

This study presents a novel deep learning approach for blood group identification using fingerprint images.The proposed method leverages Convolutional Neural Networks (CNNs) to automatically extract discriminative features from fingerprint patterns, enabling accurate and non-invasive blood group prediction. This approach holds significant potential for applications in healthcare diagnostics, biometric identification, and forensic investigations. This study makes the following key contributions: ? Development of a robust dataset containing fingerprint images labeled with confirmed blood group information. ? Design and implementation of a CNN-based classification model achieving high predictive accuracy. ? Integration of preprocessing and data augmentation techniques to improve model robustness and generalization. The results highlight the potential of using fingerprint biometrics for blood group prediction, though certain limitations still exist. The accuracy and generalizability of the model can be further enhanced with the inclusion of a more diverse dataset. Future research directions include the integration of additional biometric modalities such as palm prints or vein patterns and the application of advanced image preprocessing and feature extraction techniques to reduce the effects of low-quality or distorted fingerprints.. In conclusion, the proposed model offers a promising, non-invasive solution for blood group detection by combining biometric data with deep learning, thereby contributing to both healthcare technology and biometric security systems.

References

[1] Ahemad, M. T., Hameed, M. A., & Vankdothu, R. (2022). \"COVID-19 detection and classification for machine learning methods using human genomic data.\" Measurement: Sensors, 24(October), 100537. doi: 10.1016/j.measen.2022.100537 [2] Ayyalraj, M. K., Balamurugan, A. A., & Misganaw, W. (2022). \"Health Monitoring of Employees for Industry 4.0.\" Journal of Nanomaterials, 2022. doi: 10.1155/2022/1909364 [3] Balakrishnan, A., et al. (2021). \"A Personalized Eccentric Cyber-Physical System Architecture for Smart Healthcare.\" Security and Communication Networks, 2021. doi: 10.1155/2021/1747077 [4] Caubet, M., et al. (2022). \"Bayesian joint modeling for causal mediation analysis with a binary outcome and a binary mediator: Exploring the role of obesity in the association between cranial radiation therapy for childhood acute lymphoblastic leukemia treatment and the long-term risk.\" Computational Statistics and Data Analysis, 177, 107586. doi: 10.1016/j.csda.2022.107586 [5] Das, D., Ghosal, S., & Mohanty, S. P. (2022). \"CoviLearn: A Machine Learning Integrated Smart X-Ray Device in Healthcare Cyber-Physical System for Automatic Initial Screening of COVID-19.\" SN Computer Science, 3(2), 1–11. doi: 10.1007/s42979-022-01035-x [6] Gahane, A., & Kotadi, C. (2022). \"An Analytical Review of Heart Failure Detection based on IoT and Machine Learning.\" 2nd International Conference on Artificial Intelligence and Smart Energy, ICAIS 2022, 1308–1314. doi: 10.1109/ICAIS53314.2022.9742913 [7] Kanna, R. K., Gomalavalli, R., Devi, Y., & Ambikapathy, A. (2023). \"Computational Cognitive Analysis for Intelligent Engineering Using EEG Applications.\" In B. Mishra (Ed.), Intelligent Engineering Applications and Applied Sciences for Sustainability (pp. 309–350). IGI Global. [8] Kasat, K., Rani, D. L., Khan, B., J, A., Kirubakaran, M. K., & Malathi, P. (2022). \"A novel security framework for healthcare data through IoT sensors.\" Measurement: Sensors, 24(October), 100535. doi: 10.1016/j.measen.2022.100535 [9] Kishore, P., Dash, A. K., Pragallapati, A., Mugunthan, D., Ramesh, A., & Kumar, K. D. (2021). \"A Tripod-Type Walking Assistance for the Stroke Patient.\" Congress on Intelligent Systems, 151–160. [10] Muhammad, Y., Alshehri, M. D., Alenazy, W. M., Hoang, T. V., & Alturki, R. (2021). \"Identification of Pneumonia Disease Applying an Intelligent Computational Framework Based on Deep Learning and Machine Learning Techniques.\" Mobile Information Systems, 2021. doi: 10.1155/2021/9989237 [11] Sharathchandra, D., & Ram, M. R. (2022). \"ML Based Interactive Disease Prediction Model.\" IEEE Delhi Section Conference, DELCON 2022, doi: 10.1109/DELCON54057.2022.9752947 [12] Sherwood, J., et al. (2022). \"Automated insulin delivery with the iLet bionic pancreas for the management of cystic fibrosis–related diabetes.\" Journal of Cystic Fibrosis, 21, S6–S7. doi: 10.1016/s1569-1993(22)00700-7

Copyright

Copyright © 2025 Arya A S. 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.