Healthauth: A Multi-Modal Authentication System with ECC and
Machine Learning for Healthcare Applications
Tamilselvan R and N Thangarasu
Department of Computer Science, Karpagam Academy of Higher Education, Coimbatore-21, India
Keywords: Elliptic Curve Cryptography, Structured Convolutional Neural Network, Multi-Modal Authentication,
Healthcare Security, Biometric Authentication, Behavioral Authentication, Machine Learning in Healthcare
Abstract: In the dynamic field of digital healthcare, effective user authentication is mandatory due to safeguarding
critical health data and prevent any sort of unauthorized entry. This paper presents HealthAuth, a unique
multimodal authentication approach that is powered by Elliptic Curve Cryptography (ECC) and machine
learning to deliver robust and reliable security for healthcare-oriented solutions. It is a multi-layer
comprehensive set using biometric (such as facial recognition, fingerprints) and behavioral (such as typing
rhythm, user interaction) data for authentication, as seen in the proposed system. We propose a Structured
Convolutional Neural Network (S-CNN) to improve the processing of biometric data, a kind of CNN
architecture designed for health authentication tasks. The S-CNN is responsible for extracting hierarchical
spatial features from biometric inputs, thus providing improved accuracy and also efficiency in feature
extraction. Temoral Patterns are modelled using a Recurrent Neural Network (RNN) which makes it more
secure for user behaviour. This securing the Authentication with ECC makes it light weight and very secure
suitable for Healthcare IoT Devices, as well as its security as High because of performing challenge response
mechanisms. HealthAuth combines cryptographic measures and classification via deep learning to ensure not
only that the user is who they claim, but also how they are demonstrating themselves in real time, making a
difficult target for spoofing or replay attacks. The arbitrary experiments exhibited that HealthAuth as a system
performs better than conventional strategies as far as confirmation exactness, handling length, and security
dangers which make it an ideal answer for guaranteeing secure access to EHRs, telemedicine stages, unified
medicinal services gadgets.
1 INTRODUCTION
Healthcare technologies have been changing at a
rapid pace, catalyzed by electronic health records
(EHRs), telemedicine as well as the integration of
Internet of Things (IoT) devices, and human health
care delivery and management have evolved. While
such advances appear to be making strides towards
improving patient care and providing digital insights,
they are also open healthcare systems to significant
security threats. Since health data is extremely
sensitive and has high value, gaining unathorized
access to it might lead serious privacy breaches,
identity theft or even modification of the data which
may have a disastrous impact. Therefore, the need of
the hour is to create strong and efficient mode for
authentication which should also be secure enough to
protect patient information relying on it.
Passwords or personal identification numbers
(PINs) are traditional authentication methods that
cannot address today's security challenges in
healthcare systems. Such methods fall victim to
vulnerabilities like weak password generation,
phishing attacks, and credential theft. These
challenges have resulted in the evolution of a method
like multi-factor authentication (MFA), verifying the
identity of an individual using two or more different
ways, but they are not a solution. With multi-modal
authentication, you often use a combination of
biometric (fingerprint, facial recognition) data along
with behavioral data (typing patterns, user interaction
etc.) in an attempt to make it more challenging for an
attacker.
This paper presents HealthAuth, the next
generation multi-modal biometric authentication
system for Healthcare applications. Machine
learning-based biometric and behavioral verification
R, T. and Thangarasu, N.
Healthauth: A Multi-Modal Authentication System with ECC and Machine Learning for Healthcare Applications.
DOI: 10.5220/0013632700004664
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 3, pages 589-597
ISBN: 978-989-758-763-4
Proceedings Copyright © 2025 by SCITEPRESS – Science and Technology Publications, Lda.
589

can create a secure system without sacrificing
efficiency, powered by Elliptic Curve Cryptography
(ECC). It uses ECC, which makes it very lightweight
but useful for healthcare environments where less
computational powered IoT devices are ubiquitous.
ECC provides the same level of security as standard
cryptographic systems but with lower key sizes,
which would in turn reduce the computational
overhead and could be beneficial for real-time
healthcare applications.
Aside from using ECC for secure communication,
HealthAuth involves a new Structured Convolutional
Neural Network (S-CNN) to handle real-world
biometric healthcare databases. Benefits include high
accuracy and speed of both identifying and verifying
users as S-CNN architecture has been tuned for
extracting deep spatial features from biometric inputs
such as facial images or fingerprint. Behavioral
authentication is also incorporated with a Recurrent
Neural Network (RNN) that identifies temporal
patterns in user behavior of typos, or how they
interact, making the system more secure.
When ECC and machine learning are grouped
together in HealthAuth they form a strong
authentication framework that not only authenticates
the user across several modalities, but also guarantee
that sensitive data is transmitted securely. The system
balances advanced security with lightweight
algorithms using ECC, making the system scalable
for use across a plethora of healthcare applications.
Finally, regardless of whether it is to secure access to
electronic health records or enabling secure
telemedicine consultations or protecting connected
healthcare devices; Healthcare poses unique security
challenges for HealthAuth.
2 RELATED WORKS
Servati & Safkhani (2023) (Servati and Safkhani,
2023) have proposed the ECCbAS which is an
authentication scheme for healthcare IoT systems and
based on Elliptic Curve Cryptography (ECC). The
method established ECC for key exchange and
authentication specifications as a defence mechanism
against security issues confronted with IoT healthcare
environments. Data Privacy & Computational
Efficiency. The network is intended to improve data
privacy and computational performance. Ghaffar et
al. (2024) (Ghaffar, Kuo, et al. , 2024) proposes a
machine learning attack-resistant low latency
authentication scheme for AI powered patient health
monitoring system. To allow real-time identification
of possible security threats, the methodology uses in
combination machine learning with cryptographic
mechanisms to provide low-latency secure
communication among the devices within a
healthcare IoT network.
A smart healthcare system by Mahajan &
Junnarkar (2023) (Mahajan, Junnarkar, et al. , 2023)
incorporates a lightweight ECC with private
blockchain technology. The approach involves
medical multimedia data pre-process by the ECC to
make sure capabilities of encryption due to energy
use, using also a private blockchain for secure sharing
the information control in health care. Balakrishnan
et al. (2024) (Balakrishnan, Rajkumar, et al. ,
2024)introduces quite a safe, energy-efficient data
transmission framework by EMCQLR & EKECC
algorithms. The method ensures the energy efficiency
of healthcare IoT applications using a hybrid
encryption mechanism combined with modified ECC
and quantum learning methods for secure data
encryption and transmission.
Corthis et al. (2024) (Corthis, Ramesh, et al. ,
2024) present a fog computing-enabled framework
with a hybrid cryptographic algorithm to efficiently
identify and authenticate healthcare IoT devices. Fog
computing is used in the methodology for distributed
processing, to enable controlled latency and security
while verifying authentication of devices (through a
two-level encryption) using both asymmetric and
symmetric encryption. Patnaik & Prasad (2023)
(Patnaik, Prasad, et al. , 2023) on secure
authentication and data transmission in IoMT
systems. The design methodology also covers the
lightweight cryptographic protocol generation with
elliptic curve cryptography (ECC) and secure hashing
ensuring data privacy and integrity across medical
devices or networks.
Sheik & Durai (2023), (Sheik, Durai, et al. ,
2024)proposed an adaptive deep learning-based
authentication scheme to protect user anonymity in
telecare medical systems. The approach uses a
combination of deep learning models and
cryptographic methods such as ECC to design a
robust authentication framework for patient
identification and healthcare data privacy. Chaudhary
et al. (2023) (Chaudhary, Kumar, et al. , 2023)
proposes a ring learning with errors based three-party
authenticated key exchange protocol along with ECC
cryptography. The methodology uses the power of
post-quantum cryptographic methods along with
ECC to make a safe key exchange mechanism which
may be utilized in quantum resisting health care
systems.
Sharma et al. (2024) (Sharma, Tripathi, et al. ,
2024) creates an efficient and secure authentication
INCOFT 2025 - International Conference on Futuristic Technology
590

protocol for healthcare IoT systems employing deep
learning-based key generation. The technique
employs deep learning algorithms to create
cryptographic keys in real-time, making those keys
more secure than typical IoT devices and providing
efficient way for the data encryption during
communication. Gupta et al. (2023) (Gupta,
Mazumdar, et al. , 2023) proposes a safe data
authentication and access control protocol for
industrial healthcare systems. The technical approach
for the project is address these using techniques such
as role-based access control with ECC for securing
healthcare data along with only allowing authorized
users to read/write sensitive data. The HIPAA
protocol is crafted to ensure that data privacy and
security are maintained seamlessly in healthcare
environments.
3 METHODOLOGY
There are several key steps involved in the
methodology that helps user authentication to
maintain security. It all starts by gathering various
biometric information like facial images, fingerprints
in addition to different behavioral data like typing
patterns and mouse activities.
Figure 1: Architecture of Proposed Model
This data is preprocessed (normalized, image
enhancement also data augmentation) to improve the
quality. The proposed system uses a S-CNN to
classify biometric data and the RNN-LSTM to
analyze behavioral patterns. Key exchange ECC is
used for generating keys, encrypting user credentials,
and digital signing during communication. User
Enrollment — This is the first step in the
authentication process that involves capturing data
attributes and hashing them into credentials. When a
user logs in, their current user data is classified with
S-CNN and also RNN and the results are merged for
the final authentication by tools such as majority
voting, thus providing more secure and better
experience of authentication with healthcare apps. An
overall architecture is shown in fig 1.
3.1 Data Collection
3.1.1 Biometric Data Collection
Biometric data is collected to provide a strong user
authentication method using unique physical
attributes. In this procedure, facial images and finger
prints are two leading types of biometric information
used. To ensure a complete and generalizable dataset,
participants will be recruited from across the
demographics (age, region/ state characteristics, race,
ethnicity and guerrillas). Before data collection,
participants will receive consent explaining both the
nature of the study and its intended purposes.
3.1.2 Behavioral Data Collection
The idea is that this behavioral data will improve the
accuracy of the authentication system, as users'
interaction patterns will act as an additional check.
Behavioral data of the following manner are being
recorded: keyboard patterns, mouse movements.
Behavior during user sessions. The platform will
introduce behavioural monitoring tools to silently
monitor and log user interactions, without any
interference in order to comply with Data privacy
regulations & ethical guidelines as well as guide the
users about the nature of data they are collecting.
3.2 Data Preprocessing
3.2.1 Biometric Data Processing
It is important to pre-process biometric data, so that
the input given to S-CNN remains normalized and of
good quality. As the first step, this is standardizing
the input size of biometric images, all biometric
images will be normalized to a default common size
ensuring with consistent processing by S-CNN. This
will be quite important since it will prevent any
Healthauth: A Multi-Modal Authentication System with ECC and Machine Learning for Healthcare Applications
591

misleading through image size difference and
improve the system performance. The first step is
normalization, then will follow the image
Enhancement techniques like histogram equalization
which make it more visible. This technique helps to
enhance the contrast of the images, thus making key
features more distinguishable for better learning and
generalization by model.
Augmentations Augmentation strategies will be
used to enhance the model to be more reliable against
variations and better generalize on unseen data. These
can be referred to as transformations or geometric
changes which include the rotation of images, scaling
and flipping that has been described as taking place at
different angles scales and conditions where
biometric data might be captured. With such a way of
artificially expanding our dataset we allow the model
to learn how to match more accurately also even if
some users have a different appearance or under
slightly differing conditions.
3.2.2 Behavioral Data Processing
In behavioral data processing, we try to find features
as direct indicators of certain user interaction
patterns. From the collected behavioral data,
extracted features will include typing speed and
frequency of key presses (with a natural keyboard),
mouse movement patterns, etc. These are important
details in establishing a behavioural profile for a
particular user to distinguish real requesters from
possible fraudsters.
Normalization will be performed on the features
after extraction to bring all of the features under one
consistent range. This standardization is necessary to
conduct any meaningful analysis and to avoid the
biases that may come from using different scales or
units of measures. Secondly, the normalization of the
features improves their comparison and combination
with biometric data, which can then be used to build
a more complete and true picture. In short, the
preprocessing phase is crucial in making both
biometric and behavioral subspace that is ready for
tight fusion into HealthAuth system, besides
increasing the security and reliable of authentication
in procedures.
3.3 Model Development
3.3.1 Structured Convolutional Neural
Network (S-CNN)
The Structured Convolutional Neural Network (S-
CNN) is designed to effectively analyze and classify
biometric data, such as facial images and fingerprints.
The architecture consists of multiple convolutional
layers that are essential for capturing spatial
hierarchies in the input data. Each convolutional layer
applies a set of filters (kernels) to the input image to
create feature maps. The mathematical operation for
a convolutional layer can be expressed as:
𝑌
[
𝑖,
𝑗
]
=
(
𝑋∗𝐾
)
[
𝑖,
𝑗
]
=𝑋
[
𝑚,𝑛
]
𝐾 [𝑖

−𝑚,
𝑗
−𝑚]
(1)
Where Y is the output feature map, X is the input
image, K is the kernel and m and n are the indices of
the input.
After each convolutional layer, an activation
function, such as the Rectified Linear Unit (ReLU), is
applied to introduce non-linearity into the model:
𝑓
(𝑥) = 𝑚𝑎𝑥(0,𝑥)
(2)
This helps the model learn complex patterns in the
data. Following the activation functions, pooling
layers (e.g., Max Pooling) are utilized to reduce the
spatial dimensions of the feature maps, which
decreases the computational load and mitigates
overfitting. The pooling operation can be defined as:
Y
[
i,j
]
=max
(
,
)
∈
𝑋[𝑚,𝑛]
(3)
Where P denotes the pooling window.
The final layer of the S-CNN is the output layer,
which is designed to classify users based on their
biometric features. This layer typically employs a
softmax activation function to output class
probabilities for multiple classes, defined
mathematically as:
P
(
y=
k
|
x
)
=
𝑒

∑
𝑒



(4)
Where z

the score for class k, and K is the total
number of classes.
3.3.2 Recurrent Neural Network (RNN)
The Recurrent Neural Network (RNN) component
of the model processes the extracted behavioral
features. Given that user interactions exhibit temporal
dependencies, RNNs are particularly well-suited for
INCOFT 2025 - International Conference on Futuristic Technology
592

this task. The architecture may include Long Short-
Term Memory (LSTM) units, which are a type of
RNN that can effectively capture these temporal
dependencies.
The LSTM cell consists of several key
components, including input, output, and forget gates,
which control the flow of information. The equations
governing the LSTM cell are as follows:
f

= 𝜎 𝑊

∙
[
ℎ

,𝑥

]
+𝑏


(5)
𝑖

= 𝜎 (𝑊

∙
[
ℎ

,𝑥

]
+𝑏

)
(6)
𝐶


= 𝑡𝑎𝑛ℎ (𝑊

∙
[
ℎ

,𝑥

]
+𝑏

)
(7)
C

=
𝑓

∙𝐶

+𝑖

∙ 𝐶


(8)
𝑜

= 𝜎 (𝑊

∙
[
ℎ

,𝑥

]
+𝑏

)
(9)
h

=𝑜

∙tanh
(
𝐶

)
(10
)
Where, 𝑓

,𝑖

,𝑜

are the forget, input and output
gate activations, respectively, 𝐶

is the cell state, and
ℎ

is the hidden state, W represents the weight
matrices and b represents the bias vectors.
The output layer of the RNN will classify user
behavior patterns, providing an additional layer of
identity verification. Similar to the S-CNN, the RNN
output layer will typically use a softmax activation
function to generate probabilities for different
behavior classes:
f

= 𝜎 𝑊

∙
[
ℎ

,𝑥

]
+𝑏


(11)
P
(
y=k
|
ℎ

)
=
𝑒





∑
𝑒







(11)
Where h

is the hidden state at time t and W

are
the weights corresponding to class k.
This detailed explanation of the Model
Development process includes the design and
equations for both the S-CNN and the RNN,
providing a comprehensive view of the architectures
and their functionalities within the HealthAuth
system.
3.3.3 Integration of ECC
ECC within the HealthAuth system provides a robust
framework for secure communication between clients
and the authentication server. ECC is favored for its
efficiency and strong security features, particularly in
resource-constrained environments like healthcare
applications. The integration involves three core
components: key generation, encryption, and digital
signatures.
Key Generation. The first step in integrating ECC is
the generation of key pairs for both users and the
authentication server. Each entity in the system will
have its own unique key pair, consisting of a public
key and a private key. The public key is shared with
other parties, while the private key is kept secret.
1. Secure Random Number Generation: Key
generation begins with selecting a secure
random number, which serves as the private
key. This random number must be
sufficiently large and unpredictable to
ensure security. For example, in a 256-bit
ECC system, the private key can be
generated using a secure random number
generator (RNG), denoted as:
𝑘 ← 𝑆𝑒𝑐𝑢𝑟𝑒𝑅𝑎𝑛𝑑𝑜𝑚(256)
(12
)
where k is the generated private key.
Public Key Computation: The public key is derived
from the private key using a predefined elliptic curve
E and a base point G. The public key P is calculated
as:
𝑃 = 𝑘 ⋅ 𝐺
(13
)
where P is the public key, k is the private key, and G
is the elliptic curve generator point. This public key
can now be shared securely with the authentication
server or other users without compromising security.
Encryption. Once the key pairs are established, ECC
is used to encrypt sensitive data, such as user
credentials and authentication tokens, during
transmission between clients and the server. This
encryption process ensures that even if data is
intercepted, it remains unreadable to unauthorized
parties.
1. Encryption Process: To encrypt data, the sender
generates a unique ephemeral key ke for each
session. This ephemeral key is also a random
number, which ensures that each encryption is
Healthauth: A Multi-Modal Authentication System with ECC and Machine Learning for Healthcare Applications
593

unique, even for identical plaintexts. The sender
then computes the ephemeral public key P

:
P

=𝑘

∙𝐺
(14
)
The actual encryption is performed using the
recipient’s public key P

as follows:
𝐶 = 𝑀 ⊕ 𝐸𝑛𝑐𝑟𝑦𝑝𝑡 (𝑃

,𝑘

)
(15
)
Where, C is the ciphertext, M is the plaintext message
(e.g., user credentials or tokens),
(
𝑃

,𝑘

)
represents
the ECC encryption process using the recipients
public key.
2. Transmission: The ciphertext C and the
ephemeral public key are sent to the recipient.
Upon receipt, the recipient uses their private key
𝑃

to decrypt the message:
𝑀=𝐶 ⊕𝐷𝑒𝑐𝑟𝑦𝑝𝑡 (𝑃

,𝑘

)
(16
)
Where 𝑘

𝑖s the recipients private key.
Digital Signatures. To ensure the integrity and
authenticity of messages exchanged between clients
and the authentication server, ECC-based digital
signatures are employed. Digital signatures provide a
mechanism to verify that a message has not been
altered and confirm the identity of the sender.
1. Signing Process: When a user sends a message,
they generate a digital signature using their
private key. The signing process involves
hashing the message M with a cryptographic
hash function (e.g., SHA-256) to create a
message digest H(M). The signature S is then
created using the private key k:
𝑆 = (𝐻(𝑀) + 𝑘 ⋅ 𝑟)𝑚𝑜𝑑 𝑛
(17
)
where r is a random nonce and n is the order of the
elliptic curve.
Pseudocode for HealthAuth System
function collectUserData():
return captureBiometric(), captureBehavioral() // Collect biometric and behavioral data
function preprocess(data):
return normalizeAndEnhance(data) // Preprocess the data
function enrollUser(user):
b
iometric, behavioral = collectUserData()
storeData(user, preprocess(biometric), preprocess(behavioral)) // Store processed data
function authenticateUser():
currentBiometric, currentBehavioral = collectUserData()
b
iometricResult = S
_
CNN.predict(preprocess(currentBiometric)) // Predict with S-CNN
b
ehavioralResult = RNN.predict(preprocess(currentBehavioral)) // Predict with RNN
return combineResults(biometricResult, behavioralResult) // Combine results
S
_
CNN, RNN = buildModels() // Build models
enrollUser(newUser) // Enroll use
r
isAuthenticated = authenticateUser() // Authenticate use
r
if isAuthenticated:
g
rantAccess() // Access
g
rante
d
else:
den
y
Access() // Access denied
INCOFT 2025 - International Conference on Futuristic Technology
594

2. Verification Process: Upon receiving the
signed message, the recipient can verify the
signature using the sender’s public key PPP.
The verification checks that the signature
SSS corresponds to the message digest. The
verification process is expressed as:
Verify(H(M),S,P)→True/False
If the verification returns true, the recipient can be
confident that the message was sent by the legitimate
user and has not been tampered with.
4 RESULTS AND DISCUSSIONS
The results section presents the findings from the
implementation of the HealthAuth system,
highlighting the effectiveness of using S-CNN and
RNN in conjunction with ECC for secure
authentication in healthcare applications.
4.1 Results
4.1.1 Model Performance Metrics
To evaluate the performance of the S-CNN and RNN
models, several metrics were used, including
Table 1: Performance Metrics of S-CNN and RNN Models
Metric S-CNN
(Biometric
Data
)
RNN
(Behavioral
Data
)
Combined
Model
Accuracy 95.2% 92.5% 96.0%
Precision 94.0% 91.0% 95.5%
Recall 96.5% 93.5% 97.2%
F1-score 95.2% 92.2% 96.6%
AUC 0.98 0.95 0.99
Figure. 2: Performance of Combined Models
accuracy, precision, recall, F1-score, and area under
the ROC curve (AUC). These metrics provide
insights into the models' classification capabilities
regarding biometric and behavioral data as given in
Table 1 and Fig 2.
4.1.2 Authentication Time
Authentication time is a crucial factor in evaluating
user experience. The time taken for the system to
process the biometric and behavioral data and
produce an authentication result was measured as
given in Table 2 and Fig 3.
Table 2: Authentication Time Comparison
Method Average Time (seconds)
Biometric Only (S-CNN) 1.2
Behavioral Only (RNN) 1.5
Combined Approach 1.8
Figure 3: Time Comparison
4.1.3 Security Analysis
To assess the security of the HealthAuth system, the
effectiveness of ECC in securing user data during
transmission was evaluated. The success rate of
unauthorized access attempts was also analyzed.
The discussion section analyzes the results obtained
and their implications for the effectiveness of the
HealthAuth system in healthcare applications as
given in Table 3 and Fig 4.
Table 3: Unauthorized Access Attempt Analysis
Attempt Type Success Rate (%)
Without ECC 75%
With ECC 5%
85.00%
90.00%
95.00%
100.00%
S-CNN (Biometric Data)
RNN (Behavioral Data)
Combined Model
0
0.5
1
1.5
2
Biometric Only
(S-CNN)
Behavioral
Only (RNN)
Combined
Approach
AVERAGE TIME
(SECONDS)
Healthauth: A Multi-Modal Authentication System with ECC and Machine Learning for Healthcare Applications
595

Figure 4: Success Rate Comparison
4.2 Discussion
4.2.1 Model Performance
The results indicate that the Combined Model, which
integrates both the S-CNN for biometric data and the
RNN for behavioral data, outperforms individual
models in terms of accuracy, precision, recall, F1-
score, and AUC. The accuracy of 96.0%
demonstrates the potential of combining multiple
modalities for improved authentication, addressing
the limitations of using a single data type.
The S-CNN's high recall rate of 96.5% indicates
that it is effective in correctly identifying genuine
users, which is essential in a healthcare context where
unauthorized access can lead to severe consequences.
On the other hand, the RNN also shows strong
performance, with a recall of 93.5%, indicating its
reliability in capturing user behavior patterns.
4.2.2 Authentication Time
While the combined approach shows slightly longer
authentication times (1.8 seconds) compared to
individual models (1.2 and 1.5 seconds), it remains
within acceptable limits for user experience. The
marginal increase in time is justified by the enhanced
security and accuracy achieved through multi-modal
authentication. In real-world applications, this trade-
off is critical to ensure robust security without
significantly impacting user convenience.
4.2.3 Security Analysis
The analysis of unauthorized access attempts reveals
a significant improvement in security when ECC is
employed. The success rate of unauthorized access
attempts drops to 5% with ECC compared to 75%
without it. This stark contrast highlights the
effectiveness of ECC in securing user credentials and
authentication tokens during transmission, making
the HealthAuth system resilient against potential
attacks. The implementation of digital signatures
further enhances the integrity and authenticity of
messages exchanged between users and the
authentication server, ensuring that malicious actors
cannot tamper with the data.
5 CONCLUSIONS
The HealthAuth system is an essential breakthrough
on secure authentication to the healthcare
applications combining S-CNN and RNN
architectures with ECC. The results indicate that the
joint method not only improves user authentication
accuracy and reliability by utilizing multi-modal data
analysis, but also establishes more secure protection
for transmission of sensitive information. The system,
which performed at 96.0% accuracy takes in all
biometric and behavioral traits we identified that are
unique to healthcare settings. Also, purpose of giving
proper security through keyless signatures is justified
with the sizeable reduction in unauthorized access
attempts (from 75% without ECC to only about 5%
with ECC) showing how any approach-based security
enhancement will be highly resistant against various
known attacks. So, users are essentially trading a tiny
bit of time for much higher security—and they seem
to think that it's well within an acceptable amount.
Finally, the HealthAuth architecture provides
potential substantially more secure and efficient
authentication mechanisms to meet immediate
challenges of the health care market which points to
future research avenues needed in a direction that
may help future enhanced systems using other
features data types and specification or machine
learning algorithms.
REFERENCES
Servati, M. R., & Safkhani, M. (2023). ECCbAS: An ECC
based authentication scheme for healthcare IoT
systems. Pervasive and Mobile Computing, 90,
101753.
Ghaffar, Z., Kuo, W. C., Mahmood, K., Tariq, T., Bashir,
A. K., & Omar, M. (2024). A Machine Learning Attack
Resilient and Low-Latency Authentication Scheme for
AI-Driven Patient Health Monitoring System. IEEE
Communications Standards Magazine, 8(3), 36-42.
Mahajan, H. B., & Junnarkar, A. A. (2023). Smart
healthcare system using integrated and lightweight
ECC with private blockchain for multimedia medical
data processing. Multimedia Tools and
Applications, 82(28), 44335-44358.
0%
20%
40%
60%
80%
Without ECC With ECC
Success Rate (%)
INCOFT 2025 - International Conference on Futuristic Technology
596

Balakrishnan, D., Rajkumar, T. D., Dhanasekaran, S., &
Murugan, B. S. (2024). Secure and energy-efficient
data transmission framework for IoT-based healthcare
applications using EMCQLR and EKECC. Cluster
Computing, 27(3), 2999-3016.
Corthis, P. B., Ramesh, G. P., García-Torres, M., & Ruíz,
R. (2024). Effective Identification and Authentication
of Healthcare IoT Using Fog Computing with Hybrid
Cryptographic Algorithm. Symmetry, 16(6), 726.
Patnaik, A., & Prasad, K. K. (2023). Secure Authentication
and Data Transmission for Patients Healthcare Data in
Internet of Medical Things. International Journal of
Mathematical, Engineering and Management
Sciences, 8(5), 1006.
Sheik, S. A., & Durai, S. (2024). Cryptography with
optimal deep learning-based authentication scheme for
preserving anonymity in telecare medical information
system. Multimedia Tools and Applications, 1-20.
Chaudhary, D., Kumar, U., & Saleem, K. (2023). A
construction of three party post quantum secure
authenticated key exchange using ring learning with
errors and ecc cryptography. IEEE Access.
Sharma, S., Tripathi, S., Bhatt, K. K., & Chhimwal, N.
(2024). An Efficient and Secure Authentication
Protocol with Deep Learning Based Key Generation
toward Securing Healthcare Data in IoT. Cybernetics
and Systems, 55(4), 848-871.
Gupta, D. S., Mazumdar, N., Nag, A., & Singh, J. P. (2023).
Secure data authentication and access control protocol
for industrial healthcare system. Journal of Ambient
Intelligence and Humanized Computing, 14(5), 4853-
4864.
Healthauth: A Multi-Modal Authentication System with ECC and Machine Learning for Healthcare Applications
597