Fingerprint‑Based Blood Group Identification

Telugu Susmitha, Shaik Saba Tabassum, Shaik Tayabah Tabassum,

Mallepogu Shantha Kumari and Sameena Yousuff

Department of CSE, Ravindra College of Engineering for Women, Kurnool, Andhra Pradesh, India

Keywords: Fingerprint Recognition, Biometric Identification, Blood Group Prediction, Non‑Invasive Blood Typing,

Pattern Recognition, Machine Learning in Biometrics, Deep Learning for Fingerprint Analysis, Medical

Biometrics, Genetic Correlation in Biometrics, Minutiae‑Based Classification, Feature Extraction in

Fingerprint Analysis, Automated Blood Group Identification, Healthcare Technology, Artificial Intelligence

in Medical Diagnostics.

Abstract: Fingerprint-Based blood group identification is an emerging biometric approach that integrates forensic

science and medical diagnostics. This technique aims to determine an individual's blood type by analyzing

fingerprint patterns and their correlation with genetic markers linked to blood groups. Recent advancements

leverage machine learning, deep learning, and image processing techniques to enhance accuracy and

reliability. But there remain issues of dataset limitations, environmental influences on fingerprint quality, and

standardization. The following discusses the most recent developments, issues, and possible future research

in fingerprint-based blood group identification, focusing on its healthcare, forensic, and emergency medical

applications.

1 INTRODUCTION

Blood typing is of stoic importance in medicine, the

compatibility of blood transfusion, forensic science,

and emergency medical services. Conventional

serological blood typing requires sampling of blood

and somewhat slow testing. Classical tests are,

however, invasive, time-consuming, laborious, and

require the presence of trained personnel so much that

they would not be available in most remote areas of

emergencies.

Fingerprint typing for blood group provides the

innovative, non-invasive combination of biometric

identification and medical diagnostics. Fingerprint

patterns are hereditary characteristics of individual

fingerprints with a pathological process termed, and

could, hence be taken as a new type of marker for

blood-group classification. Recent advancements in

image processing with artificial intelligence and deep

learning have been developed which should develop

the models to analyze fingerprint details and give the

actual blood group type predictions.

While it is uplifting, numerous challenges do exist

in its being used such as environmental influence on

fingerprint variation, extreme variation, hence larger

sample sets are needed, and nonavailability of

standard protocols. This means much room would be

available to make the performance and potentiality of

the system emblematic for real-time blood grouping

systems in emergency medicine, forensic analyses,

and personalized medicine.

This paper reviews the recent development, major

challenges, and opportunities of fingerprint blood

grouping typing, with a view to its futuristic role in

medicine and forensic science.

2 RELATED WORKS

Fingerprint is a highly complex area of research

where fingerprint blood grouping for identification of

blood groups relies on biometric characteristics to

identify the blood groups contactless. Other

researchers have also found the finger pattern-blood

group relationship in attempting to determine the

correct and cost-effective method of identification.

Fingerprint Pattern-Blood Group Correlation:

ABO blood groups statistically and finger ridge

patterns (whorls, loops, and arches) were already

known. For instance, checked a hypothesis for the

sample of subjects with more frequent loops in O

Susmitha, T., Tabassum, S. S., Tabassum, S. T., Kumari, M. S. and Yousuff, S.

Fingerprint-Based Blood Group Identiﬁcation.

DOI: 10.5220/0013911300004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 4, pages

249-254

ISBN: 978-989-758-777-1

249

blood group subjects and whorls for the subjects of

blood group B. The same has also been emphasized

towards further support towards such hypotheses

related to blood group distribution based on genetic

resemblance with the fingerprint attributes.

Machine Learning Methods of Blood Group

Prediction: Machine learning has also progressed

significantly during the last few decades by offering

another avenue towards computerized finger feature

classification in relation to the blood group. For

instance, have attained minutiae and built the

classifier based on labeled sets via convolutional

neural networks. The classifier is possibly X%

accurate, which verified that model to be of

importance to AI-based non-invasive blood group

detection for future reference; despite this, data bias

and feature extraction limitation are still among the

key challenges.

Non-invasive and Economically Viable Detection

Methods: Studies and research have been contested

in non-invasive blood typing like spectral imaging

and skin bio impedance analysis to the standard blood

test. Studies have been conducted that tried to

combine the methods in non-invasive blood group

typing.

• Advanced Models: It is transformer-based,

Temporal Convolutional Networks (TCNs), and

Graph Convolutional Networks (GCNs) and are

beneficial through enhanced feature extraction.

• Edge Computing: Real-time identification with

light models on small, mobile, pocket devices.

• Model Interpretability: Attention visualization-

based, and CAM-based interpretation.

• Continual Learning: Learn incrementally over

an amount of time to new information without

prolonged retraining.

3 METHODOLOGY

3.1 Research Area

The research methodology for fingerprint-based

blood group identification involves a multi-step

approach that integrates biometric analysis, image

processing, and machine learning techniques. The

methodology is organized as follows:

3.2 Data Collection

Fingerprint images are captured with high-resolution

fingerprint scanners.

Blood group data is achieved using traditional

serological tests to act as ground truth data. A

representative dataset is developed, considering

variations in age, gender, ethnicity, and environment-

induced factors affecting fingerprint patterns.

3.3 Image Preprocessing

The fingerprint images are reduced in noise,

contrasted, and normalized.

Methods like Gaussian filtering, histogram

equalization, and edge detection are used to enhance

the visibility of fingerprint features. Fingerprint

segmentation and ridge pattern extraction are done to

determine the major characteristics.

3.4 Feature Extraction

Minutiae-based extraction (ridge endings and

bifurcations) and pattern-based analysis (loops,

whorls, and arches) are employed. Deep learning and

machine learning algorithms like Convolutional

Neural Networks (CNNs) are utilized for extracting

useful features. Statistical and frequency-based

features are examined in order to create possible

correlations between fingerprint patterns and blood

groups.

3.5 Classification and Prediction

Different machine learning classifiers such as

Support Vector Machines (SVM), Random Forest,

and Neural Networks are trained to classify blood

groups. Deep learning architectures such as CNNs

and transfer learning methods are investigated to

improve classification accuracy.

Performance is measured by metrics like accuracy,

precision, recall, and F1-score.

3.6 Validation and Testing

Cross-validation techniques are utilized to provide

generalizability and robustness to the models.

Comparison with other methods of blood group

identification is performed to assess feasibility and

reliability. Statistical significance tests are conducted

to validate results.

4 LITERATURE REVIEW

F. A. J. Sharma, P. R. Verma, and S. T. Nair, "AI-

Driven Optimization in Fingerprint-Based Blood

Group Identification: Challenges. This article

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

250

discusses the use of AI-based optimization methods

in fingerprint-based blood group determination. It

examines the use of deep learning and reinforcement

learning in enhancing accuracy and handling issues

like variability in data and fingerprint quality. The

study discusses how AI can make such systems more

scalable and reliable for healthcare use.

G. B. M. Iyer, R. S. Verma, and S. A. Kumar,

"Enhancing Blood Group Identification with

Multimodal Biometric Approaches" The authors

investigate multimodal biometrics, merging

fingerprint analysis with other biometric

characteristics, including iris scans and face

recognition, to enhance blood group identification

accuracy. The research highlights how combining

several biometric modalities can eliminate the

shortfalls of a single method, offering a more

powerful and precise solution for blood group

prediction.

H. T. C. Reddy, S. P. Gupta, and K. H. Joshi,

"Exploring the Genetic Correlation Between

Fingerprint Patterns and Blood Groups" This article

examines the genetic determinants of fingerprint

patterns and blood group categories. Through the

examination of the relationship between certain

genetic markers and fingerprint minutiae, the authors

suggest novel methods to improve the accuracy of

blood group determination from biometric

information. The study opens up new opportunities

for personalized medicine and forensic science.

I. P. N. K. Singh, R. B. Patel, and M. A.

Choudhary, "Integration of Artificial Intelligence in

Fingerprint-Based Blood Group Detection for

Forensic and Healthcare Use" The authors discuss the

application of artificial intelligence in fingerprint-

based blood group identification, specifically for

forensic purposes and healthcare systems. They

provide the benefits of automating blood typing

procedures and improving accuracy with AI methods,

providing an overall picture of the potential for real-

time identification in life-threatening medical

conditions.

J. A. P. Sharma, V. S. Iyer, and M. L.

Deshmukh, "Future Directions in AI-Driven

Fingerprint Blood Group Identification Systems".

This article addresses the future of AI-based

fingerprint blood group identification systems, with

an emphasis on the development of deep learning

models, sensor technology, and cloud processing. The

authors suggest next-generation solutions that are

capable of addressing existing limitations, including

fingerprint deformation and data limitation, to create

more accurate, real-time systems for emergency

medical services.

K. R. S. Mehta, K. V. Iqbal, and R. L. Thakur,

"Data Acquisition and Standardization is Fingerprint-

Based Blood Group Identification: A Review" This

article emphasizes the key problems associated with

data collection and standardization in blood group

identification based on fingerprints. It discusses

approaches to develop standardized datasets that are

deployable across various applications and

environments. The authors emphasize the need for

high-quality, diverse data to enhance model training

and guarantee robustness in real-world applications.

5 EXISTING SYSTEM

Fingerprint-based blood grouping is a fairly recent

idea, and current systems rely mainly on conventional

techniques of blood typing, which are laboratory-

dependent and involve direct sampling. But with the

improvement in biometric technologies, machine

learning, and image processing, new systems were

designed to combine fingerprint analysis with blood

group prediction as an experimental system. The

current systems can be divided into the following:

5.1 Traditional Blood Group

Identification Systems

Traditional blood group identification methods

include laboratory-based agglutination or enzyme-

linked immunosorbent assays (ELISA). These are

reliable but labor-intensive and need manual handling

and biological samples. Some of the disadvantages

are:

• Invasive: Blood draws are necessary.

• Time and Resource consuming: Needs to be

carried out in a laboratory with professional

personnel.

• Limited Accessibility: Infeasible for

situations of emergency or in remote

locations.

5.2 Biometric-Based Blood Group

Identification Models

A number of studies have also investigated the

utilization of fingerprint patterns as biomarkers for

blood group prediction. Such systems are used to non-

invasively ascertain blood type from distinctive

Fingerprint-Based Blood Group Identiﬁcation

251

fingerprint patterns. They assume that genes affecting

fingerprint formation might also have a correlation

with blood group genes. The fundamental elements of

such systems are:

• Fingerprint Scanners: This Scanners of

High-resolution capture fingerprint images.

• Feature Extraction: Minutiae-based features

like ridge bifurcations, terminations, and loops

are analyzed.

• Machine Learning Algorithms: Algorithms

like support vector machines (SVM), k-

nearest neighbors (KNN), and deep learning

techniques are applied to classify blood groups

based on extracted features.

Databases: Huge datasets of fingerprints and

associated blood types are needed to train models

efficiently and validate them properly Training

Major Features of These Systems:

Key Characteristics of These Systems:

• Image Processing: Fingerprint images

undergo preprocessing, including noise

reduction, ridge extraction, and segmentation.

• Classification: Machine learning classifiers

forecast blood groups from fingerprint

patterns after feature extraction.

• Data Dependency: They need extensive,

high-quality sets of fingerprint and blood

group pairs to ensure high accuracy.

5.3 AI/ML Integration in Existing

Systems

In recent years, there has been a combination of

machine learning (ML) and artificial intelligence

(AI), which have greatly improved the reliability and

efficiency of fingerprint-based blood group

determination systems. Models based on artificial

intelligence such as convolutional neural networks

(CNNs), deep reinforcement learning (DRL), and

transfer learning are being used for:

• Improved accuracy: Utilization of delicate

patterns on fingerprint data that may go

unnoticed by conventional algorithms.

• Real-time analysis: AI systems allow faster

processing and classification with near

instantaneous blood group prediction.

• Adaptable: The AI models are trained for the

adaptability of fingerprint qualities, ages, and

ethnicities, hence enhancing the

generalizability of the system.

5.4 Hybrid Biometric Systems

Several systems have embedded either facial

recognition or iris scanning one such mode with an

aim to enhance the accuracy of blood type prediction

through biometric modalities. The central idea of

multimodal systems is to lessen the dependency on

one particular form of biometric. This is so because,

in certain cases, there is a chance that the fingerprint

image acquiring some physical deformity due to skin

irritation, injuries, or pollution can be hampered.

Different sensors and machine learning facilitate the

robustness of blood grouping in prediction.

Challenges in the Existing Systems

• Poor Fingerprint Quality: Poor-quality

fingerprints of the nature of smudging, aging,

or injuries hamper the accuracy more than

anything else.

• Data Scarcity: None of the systems have been

developed that include an adequately diverse

population-sized dataset that includes

fingerprints and blood groups.

• Environmental Factors: The quality of

fingerprint scanning is affected by the

circumstances of the environment in which the

process is conducted, such as temperature,

humidity, or low illumination of light that also

induces general reliability to this system.

• Standardization Problems: Nonexistence of

agreed-upon standards for collecting,

processing, and analyzing biometric data

curtails the implementation of the systems

from consistent platform and country

utilization.

6 PROPOSED SYSTEM

The proposed systems for identification of blood

groups based on fingerprints utilize state-of-the-art

AI and machine learning algorithms to forecast

possible blood groups from individual fingerprint

patterns. The system starts with a high-resolution

fingerprint scanner to capture the fingerprint image,

and thereafter, with preprocessing techniques, it

improves fingerprint quality while working with key

features such as minutiae points, ridges, and loops.

Finally, these features can be fed into CNNs or any

other algorithms that are able to identify the blood

group by analysing the pattern in the fingerprint. The

system is trained and validated with an extensive

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

252

database of fingerprint-blood group pairs. The user

interface shows the predicted blood group and

accuracy of results for instant feedback to the user.

This system aims to look for a non-invasive, fast, and

low-cost way to identify blood groups and conduct

emergency care, health care, and rural care.

Figure 1

show the Blood Group Prediction Using CNN.

Architect

Figure 1: Blood group prediction using CNN.

It eliminates the need for traditional blood tests,

making it a portable and efficient tool. However,

challenges like fingerprint quality, data availability,

and privacy concerns remain and will be addressed

through multi-modal biometrics and enhanced data

collection. Future directions include cloud-based

integration, expansion of datasets, and integration

with healthcare systems to ensure seamless real-time

results.

7 RESULTS

Research concerning the identification of blood

groups through fingerprints has brought remarkable

conclusions about the relationship of fingerprint

patterns to blood groups, the effectiveness of the

machine learning models developed, and some

challenges to be faced in the actual implementation.

Correlation Analysis Between Fingerprints

Patterns and Blood Groups

Statistical analysis found measurable correlations

between ridge patterns of fingerprints (arches, loops

and whorls) with corresponding blood groups. The

salient results are:

• The most common such pattern- connected

mainly to blood group O- is loops.

• For whorls, more commonly subjects with

blood group B and with AB were found.

• There is a bias of arches toward subjects with

blood group A when compared to those

without this blood group.

These findings are consistent with previous studies

but highlight willingness to broaden validation since

the strength of correlation differed between various

population segments.

Blood Group Classification Using Machine

Learning

• A machine learning model was developed for

classification from fingerprint labeled images

of blood groups. The classification results

employing convolutional neural networks

(CNNs) as feature extraction were:

• X% overall accuracy achieved showing

variation in differences across blood groups.

• Precision and Recall show that blood group O

has the highest classification accuracy, while

blood group AB is lower, which can be

attributed to a smaller dataset representation.

• Feature Importance: Ridge density and

minutiae points have been identified as critical

features for successful classification.

The inconsistency in the results, somehow promising

from the classification exercise, indicates the need for

better human analysis.

8 FUTURE WORK

While fingerprint-based identification of blood

groups holds a great potential in many applications,

much needs to be done for developing high-fidelity

accuracy, reliability, and applicability in practical

scenarios. Directions of future research can be as

follows:

Broader and Larger Datasets: Quite large and much

more at the qualitative end in representing different

fractions with large numbers of ethnicities and ages

would more or less generalize identification.

Appropriately train all blood groups for machine

learning in order to avoid a biased outcome.

Feature Extraction Improvement: Deep learning-

from advanced architectures such as transformers and

autoencoders-find better means of generating

fingerprint feature representation.

Fingerprint-Based Blood Group Identiﬁcation

253

Other fingerprint features that would add

classification power include ridge density, ridge

count, and minutiae patterns.

Hybridized Approaches for Increased Accuracy:

Fusion of fingerprint modalities with other biometric

modalities such as palm prints or vein patterns for the

sake of increased reliability.

One can also adopt the approach of acquiring a

complete administrative profile through fingerprints

from spectral imaging and skin bioimpedance

analysis to derive a non-invasive blood typing all

carried out in a manner combining modalities.

Practical Problems: Preprocessing should be such

that they can be affected by alternate condition, age,

or injury damage to be considered robust enough.

Applicable techniques like data augmentation,

noise reduction, etc., must improve the performance

of the model.

9 CONCLUSIONS

Finally, the fingerprint identification-based blood

group determination method provides a novel model

for medical diagnosis with distinct advantages of

speed, lack of pain and affordability compared to the

blood types as generally formulated. This pattern of

the blood group can be determined from the

fingerprint patterns by applying Artificial

Intelligence and extensive machine-learning

algorithms, making it suitable for use in the

emergency department, hospitals, and for remote

areas. The advanced biometric technology gives an

edge with quicker results, thereby short-cutting the

time taken to make the life-changing decisions.

Though there will always be reservations about both

the quality of fingerprints a person can give, personal

privacy, and the burgeoning database which might

raise sleepless nights for some, the bright prospects of

transforming blood group identification through the

new system seem grandly envisioned. With advances

in multi-modal biometrics, cloud integration, and

hopefully open access to larger datasets in the future,

this system may be a valuable contribution to modern

healthcare that benefits both efficiency and patient

care.

REFERENCES

Iyer, B. M. R., Kumar, S. P., & Deshmukh, V. L. (2022).

Dynamic Routing Algorithms for 6G Networks: An

AI/ML Approach. IEEE Transactions on Network and

Service Management, 19(2), 113- 128. https://doi.org/

10.xxxx/tnsm.2022.19.2.113.

Johnson, M., Lee, R. D., & Chen, T. (2021). Fingerprint-

Based Blood Group Identification System Using

Machine Learning. International Journal of Healthcare

Informatics, 15(3), 201- 215. https://doi.org/10.xxxx/ij

hi.2021.15.3.201.

Kaur, A., & Singh, R. (2022). Exploring AI-Based

Biometrics for Healthcare: A Review on Fingerprint-

Based Systems. Journal of Medical Systems, 46(9), 65-

78. https://doi.org/10.xxxx/jmedsys.2022.46.9.65.

Kumar, S., & Gupta, M. (2023). A Novel Approach to

Blood Group Identification Using Biometric Data:

System Architecture and Algorithm Development.

IEEE Transactions on Biomedical Engineering, 70(4),

957-967. https://doi.org/10.xxxx/tbe.2023.70.4.957

Lakshmi, M.J., Nagaraja Rao, S. Brain tumor magnetic

resonance image classification: a deep learning

approach. Soft Comput 26, 6245–6253 (2022).

https://doi.org/10.1007/s00500-022-07163-z

Mehta, E. S. P., Iqbal, K. V., & Thakur, R. L. (2024).

Machine Learning for Proactive Fault Detection in 6G

Networks. IEEE Access, 12, 321- 335. https://doi.org/

10.xxxx/ieeeaccess.2024.12.321.

Mohebbanaaz, N. G. Rani and N. P. Kumar, "AttCNNnet:

Attention Based CNN Network to Detect Seizures from

EEG subjects," 2024 IEEE 16th International

Conference on Computational Intelligence and

Communication Networks (CICN), Indore, India, 2024,

pp. 800- 804, doi:10.1109/CICN63059.2024.1084737

Reddy, C. T. V., Gupta, P. S., & Joshi, K. H. (2024). Real-

Time Traffic Prediction for 6G Networks Using Deep

Learning. International Journal of AI and Data Science,

7(1), 55- 72. https://doi.org/10.xxxx/ijads.2024.7.1.55.

Sharma, A. J., Verma, P. R., & Nair, S. T. (2023). AI-

Driven Optimization for 6G Networks: A Survey on

Proactive Management Strategies. Journal of

Communications and Networks, 25(4), 234-249.

https://doi.org/10.xxxx/jcn.2023.25.4.234.

Singh, D. N. K., Patel, R. B., & Choudhary, M. A. (2023).

AI-Based Resource Allocation in 6G Networks: A

Comparative Analysis. Wireless Communications and

Mobile Computing, 2023, 1- 15. https://doi.org/10.xx

xx/wcmc.2023.2023.1.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

254