Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

New 1

Download as txt, pdf, or txt
Download as txt, pdf, or txt
You are on page 1of 30

\documentclass[fleqn,s1,tex]{neptune}

\usepackage[5.5]{neptune-els-dtd}
\bookindate{2022-01-24}
%% Please do not edit bookindate/proofdate
\proofdate{2022-02-05}
%%
%% for non-CAS
\usepackage[casnptn]{neptune-els} %% for CAS

\begin{pkg}
\end{pkg}
\begin{additionalpkgs}

\end{additionalpkgs}

\begin{document}
\newproof{proof}{Proof}

\def\stmdocstextcolor#1{}
\def\stmdocscolor#1{}
\let\citep\cite
%\def\stmdocsrangecite#1{#1} %% Add missing string in BODY and uncomment the macro

%\beginTeX\let\labelrm\rm\endTeX

%\beginXML\let\labelrm\relax\endXML

\begin{macros}

\end{macros}
\begin{additionalmacros}
\jshortv
\def\xURL{URL\xspace} %% If prefix URL not required, remove "URL" from the
definition.

\end{additionalmacros}
%%%\endheaderBC

\begin{frontmatter}
%\QUERY[1]
\title{A holistic survey on the use of emerging technologies to provision secure
healthcare solutions}
\tnote{This paper is for special section VSI-cei. Reviews were processed by Guest
Editor Dr. Imran Razzak and recommended for publication.}

%
\addrai[1]{S0045790622000131-67b5d291949ba855bfe38c1c97ff37fd}
\addrai[2]{S0045790622000131-76cc8bf62f7e5276a3f90645819817d1}
\addrai[3]{S0045790622000131-d5f52df0020f5be51d7585b7e522ab43}
\addrai[4]{S0045790622000131-a1f85c5552cdcf729df3a5d282d9722a}

\begin{authorgroup}

\author[1]{Senthil Kumar Jagatheesaperumal}[bio]


\email[senthilkumarj@mepcoeng.ac.in]
%\QUERY[2]
\orcid{0000-0002-9516-0327}
\contribution{Conception and design of study, Acquisition of data, Analysis and/or
interpretation of data, Writing -- original draft, Writing -- review \& editing}
%\QUERY[3]

\author[2]{Preeti Mishra}[bio]
\email[scholar.preeti@gmail.com]
\orcid{0000-0002-3809-0887}
%\QUERY[4]
\contribution{Conception and design of study, Analysis and/or interpretation of
data, Writing -- original draft, Writing -- review \& editing}
\cormark[1]
\cortext[1]{\COR
% \TFadd[\AffNum].
}

\author[3]{Nour Moustafa}[bio]
\email[nour.moustafa@adfa.edu.au]
\orcid{0000-0001-6127-9349}
\contribution{Analysis and/or interpretation of data, Writing -- review \&
editing}

\author[4]{Rahul Chauhan}[bio]
\email[chauhan14853@gmail.com]
\contribution{Acquisition of data, Writing -- original draft}

\affiliation[1]{o={Department of Electronics and Communication Engineering, Mepco


Schlenk Engineering College}, c={Sivakasi}, s={Tamil Nadu}, cy={India}}
\affiliation[2]{o={Department of Computer Science, Doon University Dehradun},
s={Uttarakhand}, cy={India}}

\affiliation[3]{o={School of Engineering and Information Technology (SEIT),


University of New South Wales (UNSW) Canberra}, cy={Australia}}

\affiliation[4]{o={Department of Electronics and Communication Engineering, Graphic


Era Deemed to be University Dehradun}, s={Uttarakhand}, cy={India}}

\end{authorgroup}

\begin{abstractgroup}
\begin{abstract}[1]
Healthcare applications demand systematic approaches to eradicate inevitable human
errors to design a framework that systematically eliminates cyber-threats. The key
focus of this paper is to provide a comprehensive survey on the use of modern
enabling technologies, such as the Internet of Things (IoT), 5G networks,
artificial intelligence (AI), and big data analytics, for providing secure and
resilient healthcare solutions. A detailed taxonomy of existing technologies has
been demonstrated for tackling various healthcare problems, along with their
security-related issues in handling healthcare data. The application areas of each
of the emerging technologies, along with their security aspects, are explained.
Furthermore, an IoT-enabled smart pill bottle prototype is designed and illustrated
as a case study for providing better understanding of the subject. Finally, various
key research challenges are summarized with future research directions.
\end{abstract}

\begin{abstract}[2]
\begin{dfigure}
\centering
\dispfig{\jobname-ga1}
\end{dfigure}
\end{abstract}

\begin{abstract}[7]
\begin{itemize}
\item The survey provides a deep discussion on the use of:
IoT-based secure healthcare solutions for provisioning prominent healthcare
applications.
\item 5G services and their secure frameworks for healthcare applications.
\item AI for security enhancement in healthcare systems along with the specific
application areas.
\item Big Data analytics for ensuring trustworthy healthcare systems.
\item Emerging technologies for the design and development of smart and secure IoT-
pill bottle (Case Study).
\end{itemize}
%\QUERY[5]
\end{abstract}

\keywords{Healthcare \sep Security \sep Internet of Things \sep Artificial


intelligence \sep Machine learning \sep Deep learning \sep 5G networks}

\end{abstractgroup}

\end{frontmatter}
\section{Introduction}

\xlabel{intro}
Security-enhanced smart healthcare services using modern technology have the
potential to transform many aspects of our society. According to the recent report
published by the World Health Organization (WHO), they have confirmed that there
has been a drastic growth in health expenditure globally to cover the universal
health prospects. Countries are now taking serious attention and spending a lot of
money on developing smart healthcare services. In past, the usage of modern
technologies has been explored by some researchers for detecting various dangerous
diseases such as chronic psychological diseases, chronic diseases, brain tumor and
breast cancer, etc. in an initial stage. Modern healthcare services must leverage
smart technologies throughout the operating area of doctors, administrators, IT
managers who are specifically tasked with the challenge of providing a safe and
secure network setting with better connectivity. Also, challenges do exist in
providing enhanced communication between core members and providing secure
interoperability of healthcare infrastructure. Therefore, the formation of
effective health systems for handling primary and advanced healthcare problems is
strongly required for providing better universal healthcare solutions. Most of the
government-allocated funds for the healthcare sector are used for handling some
deadly diseases, tuberculosis, malaria, and reproductive health issues. To have
better and safe living conditions in a dynamic environment, there is a dire
requirement for standard healthcare facilities.

Recently, some researchers~\cite{xu2020deep} have developed AI, machine learning,


and deep learning techniques based mechanisms using the existing dataset to detect
COVID-19 patients from CT scan images. They observed the spread statistics of this
disease and did much reasonable analysis on the data for the appropriate extraction
of valuable information. The substantial progression of the IoT frameworks in the
healthcare sector drives the integrity, and security of vital healthcare data,
which
was earlier a huge challenge for healthcare service providers. Efficiency
enhancement
in operational management for traditional healthcare systems are reluctant to make
quick decisions. Also, the accuracy in treatment and diagnosis is prone to
increased
exposure to risks. In the traditional system, human interventions are prevalent in
machine-made decisions. In addition, the lack of regulation and algorithm biases in
decision-making systems are hindered to provide false decisions. Furthermore, the
privacy pressures drive the need for enhanced insights into the treatment process.
These shortcomings drive the demand towards modern healthcare systems driven
through
AI, IoT, cloud services, and Big data management. The limitations with traditional
healthcare systems motivate the researchers to incorporate advanced technologies.

Modern healthcare solutions demand the usage of cloud services for storing,
processing, and retrieval of vital health information. However, cloud services in
healthcare sectors are subjected to ethical and legal impacts. The security and
privacy of such data are one of the major security challenges. It triggers the need
for developing better secure cloud services for end-users to support better
healthcare services. Smart hospitals are expected to expand their use of connected
technologies by establishing versatile physical layer connectivity, that depends on
well-defined network infrastructure.

Moreover, the security aspects are depended on the network topology, media
selection, wireless mobility, and the consideration of multi-application support
with flexible and secure network solutions. In addition to the above healthcare
solutions, AI-based healthcare platforms can provide a secure infrastructure to
analyze the data accumulated from the wearable healthcare devices to determine the
current health status of the patients. They provide custom-made health references
along with estimated possible future issues and complexity in the monitored
disease. Furthermore, Big data analytics is getting attention and has yet to be
extensively implemented in the healthcare industry. To govern big data for
healthcare services, there is a need for better secure strategies, sustained
thrustworthy processes, robust secure policies, and supporting technologies. To
guarantee security, privacy, seamless communication, and interoperability among
those smart healthcare devices (having diversified purposes and characteristics),
the 5G architecture will play a significant role in numerous m-health
circumstances.

The connected medical devices provide many vital benefits for physicians. In most
cases the manufacturer of medical devices could be able to remotely repair a
device, allowing them to get back online and assist to provide improved patient
care. Moreover, connected devices are also a target to cybercriminals and malicious
actors. As soon as a medical device is connected to a network, it becomes much
convenient for the attackers to disrupt them. Connected insulin pumps and
pacemakers are prone to be misregulated, through abnormal triggering by the
attackers. So enabling secure healthcare solutions assists the patients and
healthcare professionals to gain access to new and emerging technologies that help
improve the lives of the diseased without introducing risks to patient safety.
Modern enabling technologies and the probable research solutions have to work
together along with the manufacturers, regulators, healthcare service providers,
and patients to secure connected healthcare solutions.

The impact of secure, connected healthcare solutions will transform the healthcare
industry to a larger extent. For instance, secure cloud services provide access
control policies that limit the sharing of healthcare data and incorporate various
data security approaches. Employing secure IoT frameworks safeguards the healthcare
devices and restricts unauthorized access to them by employing appropriate security
solutions. Further high-speed streaming of medical data enabled through 5G services
guides the physicians to acquire timely medical information. Furthermore, the
inherent capability of 5G services in addressing the security threats of medical
information is far much more robust than the conventional communications used in
the healthcare systems.

In this article, a detailed comprehensive survey is provided from the security


perspective on the usage of modern technological trends for modern healthcare
applications, which involves key contributions from IoT, 5G services, AI, and Big
data. It provides a critical discussion and awareness about the security issues in
state-of-the-art modern healthcare services and associated solutions. Further, we
discuss the IoT framework for healthcare along with a few vital applications.
Various security enhancement mechanisms in modern healthcare applications have been
discussed with relevant approaches and future perspectives. In addition, the impact
of remote surgery through high stream video streaming of medical information is
highlighted. Furthermore, the impact of deep learning in biomedicine, genomics as
well as COVID-19 related studies are summarized. The design and development of
smart IoT-based pill bottles were highlighted with the results visualization
through cloud services. Finally, we highlight the key challenges in modern
healthcare systems-related research and discusses the key outcomes of the survey.
The paper will help researchers in finding solutions to address the threats in
healthcare-related systems with the use of modern technologies. The major
contributions of the paper can be summarized below.

\begin{itemize}
\item To provide a deep discussion and insight on the use of IoT for the healthcare
mechanisms and systems along with its security solutions.
\item To provide a deep discussion and insight on the use of 5G for the healthcare
mechanisms and systems and its security solutions.
\item To provide a deep discussion and insight on the use of AI for security
enhancement in healthcare systems along with the specific application areas.
\item To provide a deep discussion and insight on the use of Big Data Analytics for
providing trustworthy healthcare systems along with the specific application areas.

\item To provide a case study on the use of IoT for the design and development of
smart IoT-pill bottles for better understanding the subject of discussion.
\end{itemize}

The contribution of various researchers towards the growth of the healthcare sector
with the support of modern technologies is enormous. Numerous studies have been
conducted to better explore various healthcare solutions with the support of some
specific modern enabling technologies. However, the proposed survey provides a
detailed investigation and utilization of various modern enabling technologies for
the healthcare domain such as the use of AI, IoT, and 5G communication
technologies, and Big Data analytics for provisioning high-quality healthcare
solutions for addressing global health issues. Furthermore, the existing related
surveys are concerned with specific applications in healthcare and fail to cover
every aspect of enabling technologies (IoT, 5G \& machine learning) starting from
sensing of data, network designing issues, and data analytics, etc. In this work,
we have attempted to cover every aspect of modern healthcare solutions by making
use of IoT, 5G, and AI, and Big data analytics. It provides a comprehensive
detailed survey of the use of all these technologies in the healthcare sector. We
claim that one technology cannot be suitable to address all healthcare-related
problems. Hence, the usage of individual technology is explored in detail with
numerous examples.

The paper is organized as follows: \xref{IoThealth} summarizes the impact of IoT


and its usage in healthcare applications. \xref{5Ghealth} elaborates on the
necessity and usage of 5G on modern healthcare systems. \xref{AIhealth} discusses
the importance of utilizing AI for analyzing healthcare data. \xref{BDhealth}
provides deep insight into the use of big data analytics along with application
areas for healthcare services. \xref{CSpill} provides a case study on an IoT-based
Pill bottle for a better understanding of the subject. \xref{challenges}
elaborates on the research challenges involved in modern healthcare systems and
also summarizes the related issues. \xref{conclusion} concludes and summarizes the
work along with future research directions.

\xref{fig1} summarizes the taxonomy of modern healthcare enabling technologies \&


concepts surveyed in this article. Let us now discuss the usage of each of the
modern technologies in detail.

\begin{figure}
\caption{Classification Tree of Modern Healthcare Systems Enabling Technologies.\
xlabel{fig1}}
\includegraphics{gr1}
\end{figure}
\section{{\?{IoT}} in Healthcare}

\xlabel{IoThealth}
In IoT architectures for healthcare, it reflects the typical implementation of
remote monitoring and control of healthcare services. In this section, we introduce
the key features of IoT systems used for healthcare applications and discuss the
related literature in detail.

\subsection{Overview of \?{IoT} in Healthcare}

IoT is a network of smart devices that can monitor, control, and exchange
information with remote appliances. There are varieties of IoT-based applications
that are being deployed in different segments and they have flourished by reaping
positive paybacks to the end-users. Data Analytics has started playing a key role
in the evolution and accomplishment of various IoT applications. A lot of data
analytics tools are available for making effective use of datasets, acquired from
IoT devices connected to the network.

For the best healthcare systems in the modern era, an amalgamation of modern
technologies such as Big Data, Data Analytics, AI with IoT, is essential. IoT
healthcare systems acquire vital healthcare information from sensor data. Those
sensors are the key attributes for developing any healthcare application using IoT.
Body temperature sensor, Pulse and oxygen in blood sensor (SPO2), Blood pressure
sensor (sphygmomanometer), Patient position sensor (accelerometer), Airflow sensor
(breathing), Electrocardiogram sensor (ECG), Electromyography sensor (EMG) \&
Galvanic skin response sensor are some widely used sensors for health monitoring.

\subsection{{\?{IoT}} Computing Frameworks for Healthcare}

Various researchers have put forward different ranges of frameworks for various IoT
applications. Computing and processing of the data acquired from IoT devices are
quite challenging tasks faced by researchers. Deploying edge computing allows IoT-
enabled systems to process vital healthcare data at the edge device in the network.
However, those edge devices are located in proximity to the data source from the
medical IoT devices. A significant decrease in the bandwidth for communication is
reduced with the implementation of IoT technology. Applying data analytics to large
volumes of data becomes a challenge with the conventional mechanism. It will be
convenient enough if the volume of data acquired from the IoT devices is used for
processing, storage, and performing analytics using edge computing. \xref{fig2}
shows a sample IoT Framework employed for healthcare applications with necessary
supporting services on the data acquired from the IoT-based healthcare devices.

\begin{figure}
\caption{IoT Framework employed for healthcare applications.\xlabel{fig2}}
\includegraphics{gr2}
\end{figure}

Hardware and software integration along with different communication network


architecture for secure and safe transactions of healthcare data can be implemented
for each layer of the framework. Moreover, IoT technology, cloud-computing
solutions, edge computing, and other recent solutions for health industries with
elaborate specifications for future healthcare systems need to be secured for
promoting better healthcare services.

\subsection{Application of \?{IoT} in Healthcare}

The application of IoT in healthcare is not just limited to smart healthcare


devices
and smart healthcare systems. Moreover, it is also concerned with developing
hardware
and software solutions aiming to support quality healthcare systems. Among various
5G
services, fog computing has become one of the supporting technologies for IoT
applications. An optimization algorithm with transmission power control is proposed
by Magsi \etal~\cite{magsi_evolution_2018} for quality of service optimization in
IoT applications. The fog architecture is scaled up with IoT and along with
transmission power control, it provides maximum throughput, minimum energy, and
less
delay with better QoS. Additionally, smart hospitals with cloud connectivity, smart
ambulances are also some key features that are to be incorporated in healthcare.
There are some key applications of IoT, discussed below.

\begin{enumerate}[label={(\roman*)}]

\item \textit {\textbf{Smart Wearable Devices}--} Energy management of healthcare


devices is performed using the battery recovery algorithm and energy harvesting is
performed by optimizing the duty cycle of operations for healthcare applications.
Smart wearable healthcare devices enable to sync of patient-generated data to
electronic health records, and it assists the doctors to monitor the health
parameters without seeing the patients in person. It could reduce healthcare costs
and revolutionize how preventive medicine is carried out at present.

\item \textit {\textbf{Pill based IoT Devices}---} Data sources can be varied
depending on medical applications and requirements of the health monitoring
centers. Solutions put forward for addressing the issues in the IoT frameworks for
supporting various types of data sources for meeting the desired requirements.
Moreover, these IoT devices employed for monitoring health data will capture vital
health information, like ECG, EEG, Glucose levels, etc, and transmit them over IoT
communication protocols. Ingestible electronic pills could also monitor digestive
health and help to diagnose gastrointestinal disorders.

\item \textit {\textbf{Posture Determination using IoT Devices}---} Nowadays,


digital devices have created an epidemic of poor posture. A few minutes of
slouching over a screen can cause short-term stiffness and long-term back and neck
problems. IoT devices can help you correct poor posture. Aleks is one of the
wearable posture trackers and coaches that help you avoid neck pain. This IoT-based
device vibrates to gently remind you to correct your posture. Similar posture
correcting devices are helpful to maintain perfect postures during the prolonged
sitting of members involved in work. They could alert to correct the posture and
record events when the postures are not as per specific recommendations.

\end{enumerate}

\subsection{Security in \?{IoT}-based Healthcare services}

Enhancement in the Elliptical Curve Cryptography helps to ensure the authenticity


and
integrity of communication of vital healthcare data. Communication through ECC
requires some digital signature scheme with the exchange of the private and public
keys. Further, to enhance the security in the IoT-based medical sensor data, the
improved version of ECC in combination with Substitution-Caesar cipher is put
forward
by the authors in~\cite{khan2020secure}.

In another work by the authors in~\cite{wang2020isa}, an Identified Security


Attributes (ISA) framework was proposed to evaluate the security features of IoT
devices used in the healthcare environment. In this work, initially with the aid of
an analytical hierarchical process, the weights of the attributes are derived and
the
security assessment was performed with the preference towards similarity features,
which provides an ideal solution.

The electrical activities happening in the human heart are due to the contraction
and
relaxation of the heart muscles, and the signal generated in this action is known
as
Electrocardiogram (ECG). As IoT-assisted ECG monitoring is getting popular, it is
also
highly prone to security threats as well. Authors in~\cite{xu2020iot}, experimented
to ensure secure data transmission using lightweight secure IoT services and
lightweight access control mechanisms.

While vital medical data communication is taking place between two end-users, to
hide
them from the third parties, Steganography helps to embed the secret message with
other obvious data during the communication process. One such instance of Medical
Data Authentication using Steganography was reported in~\cite{hashim2020based},
where a multilevel security scheme is developed with random iterations and Henon
function. Here, the medical images are secured.

\begin{table}
\caption{Security threats in healthcare applications}
\xlabel{Tab:securitythreats}
\begin{tabular}{LP{80pt}P{80pt}P{180pt}}
\beginthead
{Ref.} & {Healthcare applications} & {Security threats} & {Key inferences} \\
\endthead
\rowhead Wang et al.\citep{wang2020secure} & Secured medical data retrieval &
Illegal medical data acquisition & Security in medical data delivery tasks in this
framework was ensured by making use of ciphertext and signature. \\ \hline
Khan et al.\citep{khan2020secure} & Wearable sensor devices for e-health & Illegal
access to patient data & Improved version of Elliptical Curve Cryptography in
combination with Substitution-Caesar cipher is used for enhancing the security. \\
\hline
Wang et al.\citep{wang2020isa} & IoMT & Eavesdropping and intrusion during
transmission & An Identified Security Attributes (ISA) framework were used to
provide reliable and secure solutions. \\ \hline
Hashim et al.\citep{hashim2020based} & Medical Authentication & Decryption of
secret information & Secured steganography using multilevel security scheme
developed with random iterations and Henon function. \\ \hline
Xu et al.\citep{xu2020iot} & IoT-based ECG monitoring & Illegal access to ECG data
& Light weight secure IoT services and lightweight access control mechanisms are
used to classify the ECG signals \\ \hline
Li et al.\citep{li2020secured} & IoT-Enabled Healthcare & Exploitation of edge
systems & A secure framework for healthcare services, enabled with SDN-based Edge
computing using lightweight authentication scheme. \\ \hline
Akhbarifar et al.\citep{akhbarifar2020secure}& Early diagnosis of COVID-19 &
Exploitation of sensitive healthcare data & A Lightweight secure block encryption
is used to enhance the security in cloud-assisted IoT services. \\ \hline
Sengupta et al.\citep{sengupta2020secure} & IoT enabled e-Health services &
Illegal access to health data & NoSQL based secure model for cloud-based
services. \\ \hline
Abou et al.\citep{abou2020ditrust} & IoT-based healthcare systems & impediments to
reliable information & Decentralized Interoperable Trust framework based on the
blockchain ensures trustworthy healthcare data communication. \\ \hline
Alam et al.\citep{alam2020mhealth} & mHealth Communication & Security exploitation
in electronic health data & Established secured wireless communication protocols
using blockchain and IoT.\\ \hline
Saba et al.\citep{saba2020secure} & IoMT & Insecure Internet services & Secure and
energy-efficient framework for e-healthcare.\\ \hline
Gope et al.\citep{gope2020secure} & IoT based healthcare devices & Physical
attacks on health data & Physical Unclonable Function based authentication scheme
ensure data-driven fault tolerant decision making.
\botline
\end{tabular}
\end{table}
Acceptance of patients to participate in the remote health monitoring activities
will
assist them to positively avoid periodic clinic visits and provides remote support
from healthcare officials to the patients at home. With the Internet of Medical
Things (IoMT), various health parameters could be accessed by healthcare officials
from a remote place. Due to the vital health parameters being transmitted across
the
diversified range of communication protocols, they could be hacked. A secure remote
health monitoring developed in~\cite{akhbarifar2020secure}, enforces by collecting
data from bedside IoT devices, diagnosing in real-time and the entire system of
patient care could be improved with the cloud-assisted IoT services.

Hackers are prowling with various loopholes to exploit vital healthcare


information,
particularly on the data aggregated into centralized systems. Also, universal
patient records store the patient identity that could be embedded into modern
transactions and it demands information sharing through secured means. As
blockchain-based trust models are gaining popularity, their implications towards
the
healthcare sector data management help to optimize the process flow and reduce the
operational costs. Recent works in this field and the future of blockchain
technology
integration with the IoT and healthcare sectors are explored in the survey
presented
in~\cite{srivastava2020future}. The authors in~\cite{abou2020ditrust} proposed a
Decentralized Interoperable Trust framework based on blockchain technology for
IoT-based healthcare systems. Here, trustworthy healthcare data communication is
ensured through the ripple chain established through private blockchain. It thereby
helps to aggregate the requests of healthcare data from the trusted zones.

The proliferation of smartphones, telecommunications, and consumer electronics


enables medical and health services to be delivered outside conventional medical
services. This leads to mHealth Communication that aids the telemedicine sector
with
the best wireless connectivity services. Alam \etal~\cite{alam2020mhealth} used
blockchain and IoT frameworks for securing such mHealth communication established
using the popular wireless communication protocols. It helps to provide quick
service
to patients through remote monitoring and eliminates unnecessary hospitalization by
finding the real diagnosis.

Technical and economic aspects make it inevitable to prevent the failures of


potential healthcare systems from faults. As healthcare devices need to be
fault-tolerant, great care needs to be taken to address their potential without
getting exploited to faults. Particularly, when integrated with AI, the
decision-making process is largely under threat. To address such issues in IoT-
based
healthcare systems, Gope \etal~\cite{gope2020secure} introduced a data-driven
fault-tolerant decision-making scheme integrated with a Physical Unclonable
Function
authentication scheme. This work paves the way for the evolution of modern IoT-
based
trustworthy healthcare systems.

As the outbreak of COVID-19 needs quick attention from the research community,
Medical IoT devices with the support of deep neural networks are under development.
Such neural networks are also prone to severe vulnerabilities in terms of
adversarial
perturbations. Rahman \etal~\cite{rahman2020adversarial} addressed such security
threats in the COVID-19 deep learning systems. Here, the attack model is
implemented
with the perturbations in the COVID-19 deep learning systems, and the awareness of
adversarial attacks is addressed to encourage trust in the systems.

A few of the most popular security solutions for healthcare services are summarized
in
\tabref{Tab:securitythreats}.
\section{{\?{5G}} Networks in Healthcare}

\xlabel{5Ghealth}

This section provides a summary of the aspects of using 5G services and its
associated network infrastructures along with its deployment strategies for
healthcare services.
\subsection{Overview of {\?{5G}} in Healthcare}

The impact of modern technologies is hugely positive which makes people envision
and deliver things in a simplified manner. However, with 5G services, it becomes
much more flexible since 5G is highly superior to its earlier counterpart 4G
technologies. It provides a massively faster means of communication with a larger
bandwidth. It is designed in a way to be much more flexible and follows software-
based network architecture, where the earlier ones have physical connectivity. For
performing the diagnostics and treatment of human health, 5G provides super-fast
capabilities along with the IoT. The technical specifications and requirements are
varied for indoor and outdoor hotspots as well as for urban and rural environments.
Investigations are done focusing on the applicability of core components and design
principles of Industry 4.0 for the healthcare sector and in particular, inspects
the scalability of Industry 4.0 design principles into relevant sub-domains of the
pharmaceutical industry.

\subsection{Applications of {\?{5G}} in Healthcare}

The transformation of global healthcare services diversified with 5G has started


playing a significant role. Those technological advancements can also be enabled
with IoT, machine learning, AI, and big data. The 5G enabled solutions for
healthcare systems are hindered by the potential shortcomings and the challenges in
existing healthcare systems.

\begin{enumerate}[label={(\roman*)}]
\item \textit{\textbf{Remote Surgery}--} Remote monitoring of health and prevention
of illness using the support of 5G networks has started providing extensive
support. \xref{fig3} shows one such scenario of using 5G services for accelerating
the utilization of wearable healthcare devices and conventional health information
devices for the nursing of diverse biometric and observation of health parameters
of the patients from remote places.

\item \textit{\textbf{High-Speed Video Streaming of Medical Information}--} Few


medical health services also adopt video communications. Quality and reliable
streaming of video services through wireless networks using 5G are used. Scalable
adaptive video communication for real-time medical health applications helps to
optimize the adaptation level on the group of pictures.

\end{enumerate}
\begin{figure}
\caption{Remote diagnosis of patients using 5G communication networks.\
xlabel{fig3}}
\includegraphics{gr3}
\end{figure}

\subsection{Security in {\?{5G}} enabled \?{IoT}-based Healthcare services}

As the landscape of the healthcare industry is changing at a rapid phase on a


global scale, the increasing demand to provide healthcare services for the aging
and growing population could be made sustainable with the support of 5G services.
5G enabled IoT services ensures dynamic trustworthy services among the medical data
shared between the cloud and the edge terminals. Such zero-trust architecture
enforces end-to-end security for the users of 5G-based smart medical systems and
helps them with the provision of industry-grade active defense against
cyberattacks.

IoT-based healthcare services that are prone to connectivity-related issues impacts


delay in providing medication to the indented patient with the speculated time
intervals. 5G enhanced connectivity with the integration of IoT healthcare services
helps to provide energy-efficient and secure healthcare solutions. Network routers
across the globe are looking for updates to go beyond 5G(B5G)communication to
address
the new markets across various industrial sectors. The healthcare industry is one
among them, which is in demand for such updates in communication services. Its
impact
greatly enhances the healthcare industry by providing high-speed and more secure
communication services. Particularly, they help to improve the performance of deep
learning networks employed in healthcare decision-making systems.
In~\cite{qiu2021hospital}, key 5G-based research models, medical classifications,
and their research challenges are explored. For further enhancing the trust on 5G
enabled healthcare devices and addressing the security-related issues blockchain
frameworks for 5G-enabled IoT are highlighted in the automation perspective of the
healthcare industry.

\section{{\?{AI}} in Healthcare}

\xlabel{AIhealth}

In this section, we discuss the usage of AI segments and their interactions with
other supporting services that are normally employed to support healthcare
services.

\subsection{Overview of {\?{AI}} in Healthcare}

AI and high-performance computing have started redefining healthcare in the new era
of precision medicine. The healthcare industry has quickly taken advantage of AI,
machine learning, and deep learning. Medical application nowadays has started using
them to accelerate and improve patient outcomes. Automation, cloud technologies,
and streamlined information technology towards healthcare have become more
important than ever before. Machine learning and AI have a strong set of scientific
components, through which people express their thought processes, create newer
ones, and test the existing concepts. Machine learning relies on algorithms, which
are deployed to analyze the dataset and extract meaningful conclusions out of it.
Machine learning algorithms process the medical data and even suggest treatments
and steps in drug development. It also helps us to identify tumors, fractures, and
other complex medical conditions.

Advances in deep learning are driven by various powerful GPU accelerators; the
healthcare industry is heading towards more sophisticated services such as
personalized medicine, wearable medical devices, and automated robotic surgery.
Researchers and start-ups have already begun to improve accuracy from the medical
data and carry out analysis at high speeds. Deep learning networks are also used
for predicting heart conditions and cancerous genetic mutations before they make
significant implications for treatment. In addition, advancements in AI technology
also have enhanced the way AI-assisted robotic surgery is performed, supporting
quick and accurate clinical diagnosis, and providing virtual nursing assistants.

\subsection{Applications of {\?{AI}} in healthcare}

According to a survey carried out in the year 2015, medical error and misdiagnosing
illness caused 10\% of deaths in the US. Therefore, improving the diagnostic
process is one of the major AI healthcare goals. In the majority of cases, it has
been observed that AI-based diagnosis can perform better than medical
professionals. Machine learning and deep learning are some AI algorithms and
techniques that are being extensively used in healthcare over the years. A
comprehensive survey is carried out on the use of deep learning algorithms in
healthcare.

Deep learning is very important to each technical and scientific field, especially
in medical image analysis. Deep learning is concerned with the use of deep hidden
layer neural networks for pattern recognition \& classification. Deep learning is a
type of representational learning as the machine develops its representations by
using the raw data fed into the system, which is required for pattern recognition.
Unlike conventional ANN, the deep neural network has more number of hidden layers
for better data abstraction.

\begin{figure}
\caption{A CNN based deep neural network with heterogeneous type of input data and
multiple hidden layer neurons and multiple output
class~\cite{nivaashini_deep_2018}.\xlabel{fig4}}
\includegraphics{gr4}
\end{figure}

There are some deep learning algorithms like Convolution Neural Network (CNN),
Autoencoders, and Recurrent Neural Network (RNN) that can be used extensively in
medical image processing, analysis of biomedical signals, and classifications of
diseases. Objection detection and image classification provide aid to the
conventional disease diagnosis pattern to the fully automatic diagnosis with the
efficiency and accuracy better than the human-centric analysis. There are some
complex medical diagnostic problems, which require an expert level of knowledge and
inference, and often an efficient diagnosis would not be possible. In such cases,
CNN based object detection and classification can be best utilized (refer \
xref{fig4}) for spanning dermatology, ophthalmology and radiology.

\begin{enumerate}[label={(\roman*)}]

\item \textit {\textbf{Deep learning applied to Physiological (EEG) signals}--} For


disease detection, treatment, and rehabilitation, the physiological signal proved
to be an aid. Various sensors are mounted on the body or region of interest and
thus determine the characteristics of the region. Hence, to have a proper
diagnosis, the relevant information must be extracted from the signals for
providing a better healthcare support system and application. A Major problem with
physiological signal interpretation is their parameters require extensive training
to become an expert in signal interpretation.

\item \textit {\textbf{Deep learning in biomedicine}--} The amount of data


generated in the biomedicine field is huge and it is increasing at a fast rate. But
the heterogeneity of data renders the integrative analysis difficult.
Alternatively, one pathology may appear to differ from itself when imposed on a
variety of different experimental ``backgrounds'' and in several different states
of progression.

\item \textit {\textbf{Deep learning in genomics}--} The accuracy of clinician


recommendations, treatment, and diagnosis is dependent on the understanding of the
genetics of disease. Determine the novelty variant in a patient's genome is a key
challenge. In part, this decision relies on predicting the pathogenicity of
mutations; a task that already uses features like protein structure and
evolutionary conservation to train learning algorithms. Deep-learning algorithms
provide more accurate pathogenicity predictions with greater power and the ability
to effectively integrate disparate data types. In the prediction of phenotype from
genetic data, machine learning \& deep learning play an important role including
complex heredity traits such as height, weight as well as disease risk. Deep
learning can further enhance such models by integrating additional modalities such
as medical images, clinical history, and wearable device data. A particularly
promising approach to phenotype prediction is to predict intermediate molecular
phenotypes---e.g.,~gene expression or gene splicing which is then fed into
downstream disease predictors.

\item \textit {\textbf{Diabetic retinopathy with deep learning}--} One of the major
side effects of diabetes is retinopathy. The manual detection of diabetic
retinopathy
is very difficult as the disease shows very low symptoms at an early stage.
Computer-aided diabetic retinopathy has shown some significant results in the past.
The deep learning model has better optimization \& accuracy in the prediction of
diabetic retinopathy as compared to the traditional machine learning approach.
Leung
\etal~\cite{leung_machine_2015} trained on the kaggle fundus dataset, which is
publically available with the deep convolutional neural network (DCNN) algorithm.
The
author used dropout layer techniques. The same algorithm is also tested on DRIVE
and
STARE dataset fundus images for the classification of the fundus. The author
achieves
the classification accuracy of 94\%--96\%, which is very high. Pratt
\etal~\cite{pratt_convolutional_2016} used the 80,000 digital fundus images from
the
kaggle dataset to implement the NVIDIA CUDA DCNN library. The network has been
validated on 5000 images. Five classes with features such as exudates, hemorrhages,
and micro-aneurysms are chosen for classification which achieved up to 95\%
specificity, 30\% sensitivity, and 75\% accuracy.

\item \textit {\textbf{COVID-19 Screening with deep learning}--} One of the major
issues in the novel coronavirus screening is the quick identification of the COVID-
19
infected patients before the treatment. Due to the minimum number of test kits
available at certain hospitals, an automated COVID-19 detection system using deep
neural networks can be a better alternative. The classification between COVID-19
and
non-COVID-19 CT images is performed by Gozes \etal~\cite{gozes2020rapid} on the
testing performed on 157 patients from China and the United States. From the chest
X-ray (CXR) images of 13,725 patients using the modified version of the CNN, COVID-
19
symptoms are predicted by Wang \etal~\cite{wang2020covid}. Also, from the observed
results, deep insight into the explanations of the impact is given by the developed
system on the COVID-19 cases. COVID-19 infected regions are segmented from the CT
images using 3-dimensional deep learning models for early diagnosis of the COVID-19
patients. Using human-in-the-loop (HITL) methodology quantification of the COVID-19
from the CT images of 300 COVID-19 patients is done using deep neural networks by
Shan \etal~\cite{shan+2020lung}. \xref{fig5} shows the pipeline stages in the
quantifying pipeline stages involved in the prediction of the infected regions from
the CT images of COVID-19 affected persons. Three different types of CNN are used
by
Narin \etal~\cite{narin2020automatic} for automatic detection of COVID-19 from the
chest X-ray radiographs of the infected patients.

\end{enumerate}

\begin{figure}
\floatsource{Redrawn from~\cite{shan+2020lung}.}
\caption{A deep neural network based pipeline for quantifying COVID-19 infection
from
chest CT images.\xlabel{fig5}}
\includegraphics{gr5}
\end{figure}

\subsection{Security in {\?{AI}}-enabled \?{IoT}-based Healthcare services}

The advent of AI-based solutions helps to drive the healthcare sector in providing
enhanced security solutions. Security management systems with AI could learn the
parameters of the networks, behaviors of the patients, and other vital entities
that
can address security breaches. One of the unique aspects of AI is its capabilities
of
provisioning centralized training and processing of tasks. This leads to the usage
of
huge footprints with more processing power and resources. It needs a fundamental
change in the design and nature of AI applications. This could be done by instead
of
having a centralized AI in a data center, we can have AI in the edge devices. Greco
\etal~\cite{greco2020trends} highlight the evolution of IoT in the healthcare
sector
by moving AI to the edge of IoMT devices. The authors also exploited the
flexibility
of edge and fog computing for running the AI algorithms for healthcare
applications.

Safety and security of healthcare devices, patient and staff data being a top
priority in healthcare centers they need to operate robustly in a tightly charged
environment. AI-based security solution platforms help to understand the cultural
challenges and healthcare devices operating actively in the environment. Authors
in~\cite{gopalan2021iot}, compiled the usage of AI tools for protecting the IoT
networks used in healthcare applications against cyber-attacks. Similarly, Bhuiyam
\etal~\cite{bhuiyan2021internet} proposed a new security architectural model for
IoT-based healthcare applications. Furthermore, the authors reviewed imparting
sustainable IoT healthcare development and the discovery of possible features for
its
implementation.

The revolution of the healthcare industry from doctor centric approach was followed
by the introduction of electronic health records (EHRs). Now the AI has transformed
the next phase of the patient-centric approach towards telehealthcare and IoT
technology plays a lead role to support them. Such a dominant role of AI in the
healthcare industry is termed Healthcare 4.0, were framing a secure infrastructure
for healthcare applications has been a challenging task. To address these
challenges
in the AI-driven Healthcare 4.0 Hathaliya \etal~\cite{hathaliya2020exhaustive} made
an exhaustive review on the privacy and security issues in Healthcare 4.0
considering
the data breach occurring in the patient reports, patient's contact details, and
messages.

Medical image analysis systems developed using deep neural networks are often prone
to adversarial attacks that may affect the biological textures. The authors
in~\cite{ma2021understanding} developed a framework to understand and detect such
adversarial attacks thereby suppressing the over parameterization issues and
providing high detectable accuracy.

AI techniques can identify the information present in healthcare data. They can
find and predict possible connections on health issues from the large size
datasets, and recognize knowledge using Big Data Analytic techniques such as
machine learning, deep learning, or natural language processing. AI-based systems
incorporate these techniques to automate and accelerate the pre-processing stages
of healthcare data, preparation of data for the models, and are also helpful in
assisting the data exploration process. The details of various Big Data Analytics
approaches are given in the section below.

\section{Big Data Analytics in Healthcare}

\xlabel{BDhealth}

For healthcare services, most of the predictions are usually based on big data and
analytics. This section offers means of integrating big data into healthcare
services. Moreover, we discuss the means of using the existing healthcare datasets
in the big data analytics platforms.

\subsection{Overview of Big Data Analytics in Healthcare}

Big data analytics has evolved over the years and is very useful in analyzing data
collected through IoT applications with good accuracy. The process of finding
useful information from big datasets is called data mining. Dynamic processing of
large volumes of data and storing them using distributed file systems are key
challenges in big data analytics. Many open-source tools from Apache have made
significant contributions to the development of solutions for Big Data analytics.
Data analytics for mobile health and its relevant issues are needed to be carefully
managed with appropriate solutions.

The deployment of big data analytics for healthcare profiles enables the end-users
to enjoy various benefits. They utilize the IT infrastructure with better
operational efficiency. It enables a better and smart organization with strategic
management of tasks in the healthcare centers which provides quick updates of
health information and makes accurate predictions on the huge volume of healthcare
data. In healthcare systems deployed at healthcare centers, sharing vital
healthcare information, and the prediction of accurate health information are
mandatory requirements. It can be achieved by the implementation of IoT-enabled
systems for real-time health monitoring and performing big data analytics.
Awareness of existing choices of big data analytics techniques for IoT-based
healthcare systems is essential when assessing and selecting a suitable methodology
for prediction and decision-making.

To classify huge volumes of healthcare data, supervised machine learning techniques


can be utilized. This approach uses previous knowledge acquired from the historic
data to categorize the healthcare data into subcategories and distinct groups. The
predefined categories are assigned to certain objects based on some rules about
that particular category for forecasting a group of health information or class of
health data for that particular defined object.

The Clustering of healthcare data provides a higher level of abstraction that helps
in extracting the hidden details of health information from the large volume of
data. The clustering of healthcare data can be done based on either by using
partitioning clustering, hierarchical clustering, or co-occurrence clustering
mechanisms. Clustering mechanisms fall under the category of unsupervised
learning, in which the system discovers the features of the input data without
using any prior set of categories. The clustering mechanisms such as K-mean,
Hierarchical clustering, One-class SVM, DBSCAN implements unsupervised learning
approach on healthcare data. They are used to categorize the groups with certain
healthcare information, based on their meaningful characteristic features.

\subsection{Applications of Big Data Analytics in Healthcare}

Data mining plays a significant role in big data analytics, and most of the
analytics techniques are developed using data mining algorithms based on the
situation and applications. Big data has large amounts of data that may not be well
structured in practice. From big health datasets, data mining can contribute
several values to a healthcare organization. For example, a medical equipment
company can analyze its recent medical equipment sales data to identify the most
important clients. Then the company can provide some discounts or promotions to
clients like hospitals and clinics to increase equipment sales in the future.
Moreover, in the coming future, every individual would carry an average of 4--5
health gadgets.

\subsection{Security of Healthcare Big data}

Data protection laws have been established by various agencies across the globe to
secure vital data. Healthcare data are one among them, that gives individuals and
data subjects greater control over how the healthcare data are stored, processed,
and communicated. It leads to developing a better healthcare information system
with enhanced quality of patient care through secure use and exchange of healthcare
information. With the aid of Big data, healthcare information systems help to
analyze electronic health records as well as how they correlate to privacy,
confidentiality, and security. When managing healthcare information from the
perspective of Big data, need to address the quality of the data, security of
different types of clinical information in telemedicine, telehealth, and databases.
Various researchers have put forward solutions to address the security, privacy,
and challenges in healthcare data.

Retrieval of medical data from electronic health records in a streamlined fashion


helps to provide an improved patient experience and also saves a lot of time for
clinicians. Secured medical data retrieval from edge devices in a cloud
infrastructure that offers abundant resources is most used in conjugation with the
medical IoT devices. Authors in~\cite{wang2020secure} implemented IoT-based edge
cloud infrastructure with the aid of named data networking for improving the
efficiency of the data retrieval process. Security in medical data delivery tasks
in
this framework was ensured by making use of ciphertext and signature. Also, with
the
quantitative analysis, the latency in medical data retrieval was considerably
reduced.
Security enhancement in cloud services is also becoming increasingly concerned with
the ability to store, handle and process healthcare data. The
VMShield~\cite{mishra2021vmshield}, which could be used in cloud-based service
platforms, which secures the virtual domains by performing memory introspection
from
trusted domains and restricts malware in cloud services.

\section{Case Study: \?{IoT} \?{P}ill Bottle}

\xlabel{CSpill}
In the era of modern enabling technologies, society is shifting towards a
personalized healthcare system rather than the standard healthcare system. This
means that diagnosis and prescription are now patient-oriented rather than disease-
oriented. The reason for this shift is in the advancement of data analytics and
technologies like IoT, cloud computing, and machine learning. Nowadays, we are not
supposed to take a generalized prescribed medicine for a particular disease. Rather
than the researchers are developing the technologies wherein one can even identify
the effectiveness of a particular drug and accordingly suitable drug can be taken.
The history of patient ailment can be analyzed via data analytics and with
connected technology, the information can be accessed at any time, anywhere with
cloud computing and IoT. This motivation results in the development of various
handheld smart devices and applications in healthcare.

There are billion prescriptions being written every year around the world and thus
numbers of people taking pills are growing. Medical adherence is a serious concern
for society. The recommended dose of medicine necessary to be taken at a specified
time and frequency is needed to get the full benefit of medicine. Especially for
the patients who suffer from chronic diseases like diabetes \&, high blood pressure
is supposed to be more careful of the time and frequency of the recommended dose.
This reluctance leads to costly hospital bills and fatal deaths. Therefore, the
researchers are looking forward to the solution to this medical adherence.
Therefore, numerous solutions were given in the past to tackle this problem, such
as smart pill bottles working on the principle that it generates an alarm whenever
a patient forgot to take medicine at a fixed time and date.

This connected pill bottle has the potential to tackle medical adherence by
generating medical alert messages and emails with a supported network. It also
facilitates the use of data analytics to generate the prediction of disease
seriousness and analyze the response of a particular drug on the patient. Thus, it
helps in developing personalized drugs for a particular ailment. \xref{fig6} shows
the approach and design of an IoT-enabled smart pill bottle. The design
consideration and technical specification of the Pill-bottle are mentioned with the
necessary hardware \& software requirements.

\subsection{Design consideration \& technical specifications of an \?{IoT} enabled


smart \?{P}ill-bottle}

A pill bottle is having seven different sections corresponding to the 7 days in a


week. Each sub-section is stocked with the prescribed medicine. A detailed
description of each sub-component is mentioned below.
\begin{enumerate}[label={(\arabic*.)}]
\item \textit{ESP8266 module (Node MCU)}: Developed by the Espressif system, it
is a cost-effective integrated Wi-fi chip module. Operated on 3.3 V and provided
with Graphical programming input--output (GPIO) pin. It supports serial
communication protocols like I2C, SPI, and UART. Powered by a microcontroller and
ESP8266 Wi-fi chip, this module is programmed for updating the information of the
pill taken on thing speak server.

\item \textit{LDR Sensor:} Light-dependent resistor (LDR) is a sensor that


works on the principle of variable resistance with the light intensity falling upon
it.

\item \textit{Battery \& Voltage regulator:} The operating voltage for the Wi-
fi module is 3.3 V. Therefore, a chargeable lithium-ion battery is required for
adequate power supply.

\item \textit{Thingspeak server:} Records of Pill consumed is being updated on


the Thingspeak server. A dashboard is created and from collected data, alert
information of pill taken or not is being sent to three different entity namely
family member, pharmacist, and doctor.

\item \textit{Pharmacist and Family App:} There are some dedicated apps
designed for pharmacists and family members. The family member is being informed in
case a patient missed the specific dosage. Similarly, the pharmacist is being
informed whenever the pill is getting shortened.

\item \textit{Expert System:} A recommender system can be designed based on the


data and information available to the server.

\end{enumerate}

\subsection{Working Mechanism}

This is just an attempt to make a test bench for the IoT pill bottle. This cost-
effective solution is designed with an optimal cost. A basic methodology adopted
here is explained as follows:

\begin{enumerate}[label={(\arabic*)}]

\item A dashboard (personalized cloud space) on the Thingspeak server,


concerning the patient, is created wherein the medical history of a patient,
prescription by the doctor for an ailment is being uploaded.

\item LDR sensor is mounted inside the bottle cap and acts as a trigger whenever
the bottle cap is being opened. There are seven separate sections concerning 7
days a week. LDR sensor ensures whether the recommended dose has been consumed or
not according to its inherent property of sensing light intensity.

\item An ESP8266 Wi-fi module is programmed with an LDR sensor and whenever a
bottle cap is opened, it triggers an event to the dashboard created on the
Thingspeak server, and accordingly, the information is being saved. As shown in \
xref{fig7}, the pill bottle is equipped with an ESP8266 Wi-fi module and an LDR
sensor. As soon as the bottle cap is open, an event is triggered on the dashboard
and it can be visualized in the graph.

\item If a person forgets to take medicine on prescribed time duration then an


alert message from the server is being sent to the family member app (specifical
design for this project) concerning the missed dosage.

\item Besides that, the server sends the alert message to the pharmacist app as
soon the medicines in the bottle get shortened ensuring the sufficient stock of
medicine to the patient.

\item As mentioned earlier, the medical records of the patient (health


parameters like disease history, weight, hemoglobin, sugar level, allergy to
specific drugs, etc.) are available on the server. Moreover, with the help of
present data, prescribed medicine, time, and frequency of medicine dosage; an
expert recommender system can be designed in the future. With the data analytics
techniques and deep learning algorithms, a prediction of disease progression, and
the effectiveness of a particular drug on a patient can be estimated. Therefore,
effective diagnosis measures can be taken and besides that, it can help the drug
companies to analyze the effectiveness of a particular drug on a patient.

\end{enumerate}

\begin{figure}
\caption{Design principle of an IoT enabled smart Pill-bottle with expert
information system.\xlabel{fig6}}
\includegraphics{gr6}
\end{figure}

In case of critical illness, where the time and frequency of medicine dosage is
critical, this pill bottle might prove to be aid through a timely alert generation
facility. It ensures that the patient is taking the medicines at a prescribed time
and frequency according to the alert messages. Besides, by keeping the record of
dosage taken by the patient, one can avoid a lack of medication adherence. As the
pharmacist is being informed on a timely basis so the need to remember the medicine
stock is not required. In the future with the help of an expert system, a
personalized drug delivery system can be developed as well as various other
features related to the disease can be estimated.

\begin{figure}
\caption{The triggered event on Thingspeak server dashboard is visualized through a
graph.\xlabel{fig7}}
\includegraphics{gr7}
\end{figure}
\section{Open Research Challenges in Modern Healthcare Systems}

\xlabel{challenges}

In this section, we provide the state of the art research challenges in healthcare
and, various challenges in modern healthcare systems are summarized based on the
survey carried out with potential research gaps to give future research directions
to researchers. The \xref{fig8} shows the most prominent open-end research
challenges that could motivate the researchers in this domain for enhancing the
security aspects in modern healthcare systems.

\begin{itemize}
\item \emph{Security in IoT healthcare data:}
\para
The quality of the data acquired from the IoT devices is playing a crucial role in
the analysis of health information for the patients. Faithful accumulation of
quality information from the devices is one of the premier challenges in healthcare
systems. In addition to that, the physical components of the healthcare system, the
framework was chosen for the management are also one of the challenging factors in
health services. Services provided by the IoT-based existing healthcare systems may
also be prone to attacks by unauthorized access. Therefore, securing the resources
and providing quality services also plays a crucial role in the deployment of the
best healthcare services.

\item\emph{Connectivity of healthcare devices:}


As IoT healthcare wearables are getting popular, most of them are not shared with
common communication protocols. This may lead to wireless connectivity issues
unless the physical layer protocol of the IoT-based smart health monitoring devices
is well understood by the devices to which the communication is established.
Research challenges persist in this domain to a larger extent to provide a common
standard and architecture for establishing communication with wireless healthcare
smart devices.

\item\emph{High-Speed Communication of healthcare data:}


Medical experts might not be always near the patients. In such cases, the health
information of the patients needs to send to them at a very high speed,
particularly in emergencies. Conventional communication architecture is prone to a
lot of hindrances. Even though 5G communication provides better solutions, still
many rural areas persist without the reach of proper wireless connectivity.
Researchers are working on this challenge to provide unmanned air vehicles (UAVs)
based 5G architecture targeting to establish reliable high-speed communication.
\begin{figure}[,belowfloat=-10pt]
\caption{Prominent open research challenges in security management for modern
healthcare systems.\xlabel{fig8}}
\includegraphics{gr8}
\end{figure}
\item\emph{Managing Big data healthcare information:}
Data involved in the healthcare sector is quite large with a variety of data
available as text, image, and video. Storing, processing, and transmitting those
vital data are highly challenging. The healthcare data acquired from the patients
are of high volume with heterogeneous complexity and they have the characteristics
of high dynamism. Researches are exploring big data services of cloud computing to
provide simplified access to provide medical aids to the needed persons with the
processed healthcare data and information.

\item\emph{Quality of the healthcare datasets:}


Performance indication of the healthcare data is mandatory to estimate the
credibility of the healthcare data. The data quality index and quality assurance
play a major role in data mining tasks for performing data analytics. A large set
of healthcare providers issue large volumes and a variety of data. Challenges are
involved in assessing their performance, quality, and reliability to gain
meaningful and accurate predictions from the healthcare datasets for providing
quality healthcare services for the people.

\item\emph{High-performance hardware for training health data:}


For training the healthcare data particularly while implementing deep learning
algorithms, usage of high-performance hardware is required. Conventional systems
might not have such a capability. Using Graphical Processing Units(GPUs) with
multiple cores reduces the time taken for training the healthcare data and improves
the efficiency of the process. Potential challenges exist in the cost and power
consumption of those GPUs while training the data. Particularly for medical
imaging, robot-assisted surgery, natural language processing in electronic health
records deep learning algorithm requires high-performance hardware with minimum
power consumption and requires maximum efficiency and accuracy.

\item\emph{Accuracy in the prediction of healthcare information:}


Machine learning and deep learning algorithms are not ready-made for all kinds of
healthcare data. They need to be tailored to suit the application with appropriate
algorithms with the best efficiency and accuracy. Challenges exist in the choice of
the algorithm and the development of novel algorithms based on the complete study
and analysis of the healthcare data.

\item\emph{Lack of security-by-design models with existing healthcare Systems:}


Besides the cloud architectures for healthcare being a preferable computation model
that encourages scalable, flexible, and economic solutions, the lack of security-
by-design models can be attended to efficiently by exploring the challenges in
securing the modern healthcare services. Also, domain-specific vulnerabilities need
to be addressed through preliminary analysis. Moreover, security-by-design can be
achieved by deployment of big data frameworks over cloud services.

\item\emph{Security Models to deal with adversarial attacks with AI-based


healthcare systems:} A mechanism to deal with the adversarial attacks is highly
essential for the AI/deep learning frameworks particularly targeted towards medical
imaging applications. So, the implementation of robust models considering the
vulnerabilities caused by the unexplored factors of adversarial effects requires
focusing on the black box setting of the adversarial attacks. The approaches should
be able to decrease the attack possibilities by exploiting the disjoint training
data between the attacker model and the target.

\end{itemize}
\section{Conclusion}

\xlabel{conclusion}
To tackle the ever-growing and increasing trends in the healthcare environment with
numerous challenges, patients and healthcare providers need to be equipped with
modern technological trends. The proposed survey provides a detailed comprehensive
analysis on the usage of modern enabling technologies such as IoT, 5G, AI, and Big
data analytics for providing secured healthcare solutions. The use of modern
technologies enables better prediction capability along with better decision
support infrastructure with AI-based healthcare models, capable of providing
enhanced security features, evaluation of the robustness of the system, and
addressing adversarial attacks. Various existing surveys are discussed and compared
with the proposed survey paper. The use of big data analytics in extracting the
unstructured data in a secured manner by using classification and clustering
techniques is summarized. Besides, the usage of 5G services for enhancing the
healthy lifestyle of the patients is discussed in detail with appropriate examples.
Better analysis of privacy in healthcare data through IoT devices offers a
significant road map towards trustworthy healthcare systems. Extensive support of
AI is discussed with the summary of healthcare-related works in the application of
machine learning and deep learning approaches. Also, a case study on IoT Pill Box
is discussed to demonstrate the use of IoT for addressing some healthcare problems.
Furthermore, various key research challenges have been discussed in detail which
provides greater opportunities to emerging research scientists.

\conflictofinterest
\cfist

\back{}
%\QUERY[6]
{}

\nocite{*}

\bibliographystyle{stm-xml-sort-num}

\bibliography{\jobname}

\begin{bibwrite}{\jobname.bib}
%\QUERY[7]
@custombib{xu2020deep,
mypattern={[authors,atitle,midc][booktag,bktitle,yr,endbooktag,bibpages,bibdoi]
[post,myehost]},

serialno={1},
author={Xu, Xiaowei and Jiang, Xiangao and Ma, Chunlian and Du, Peng and Li,
Xukun and Lv, Shuangzhi and others},
year={2020},
title={Deep learning system to screen coronavirus disease 2019 pneumonia},
post={arXiv preprint \myehost{arXiv:2002.09334}},
}

@inbook{magsi_evolution_2018,
serialno={2},
author={Magsi, Hina and Sodhro, Ali Hassan and Chachar, Faheem Akhtar and
Abro, Saeed A. Khan and Sodhro, Gul Hassan and Pirbhulal, Sandeep},
year={2018},
title={Evolution of {{\?{5G}}} in {Internet} of medical things},
booktitle={2018 international conference on computing, mathematics and
engineering technologies},
pages={1--7},
orgname = {IEEE},
isbn={1-5386-1370-0},
}

@article{khan2020secure,
serialno={3},
author={Khan, Mohammad Ayoub and Quasim, Mohammad Tabrez and Alghamdi, Norah
Saleh and Khan, Mohammad Yahiya},
year={2020},
title={A secure framework for authentication and encryption using improved \?{ECC}
for \?{IoT}-based medical sensor data},
journal={\rvtIEEEA},
volume={8},
pages={52018--52027},
orgname = {IEEE},
}

@article{wang2020isa,
serialno={4},
author={Wang, Lanjing and Ali, Yasir and Nazir, Shah and Niazi, Mahmood},
year={2020},
title={{ISA} evaluation framework for security of internet of health things
system using \?{AHP}-\?{TOPSIS} methods},
journal={\rvtIEEEA},
volume={8},
pages={152316--152332},
orgname = {IEEE},
}

@article{xu2020iot,
serialno={5},
author={Xu, Guangyu},
year={2020},
title={\?{IoT-assisted \?{ECG} monitoring framework with secure data transmission
for
health care applications}},
journal={\rvtIEEEA},
volume={8},
pages={74586--74594},
orgname = {IEEE},
}

@inbook{hashim2020based,
serialno={6},
author={Hashim, Mohammed Mahdi and Rhaif, Suhad Hasan and Abdulrazzaq, Ali A
and Ali, Adnan Hussein and Taha, Mustafa Sabah},
year={2020},
title={Based on {\?{IoT}} Healthcare Application for Medical Data Authentication:
\?{T}owards A New Secure Framework Using Steganography},
booktitle={IOP conference series: Materials science and engineering, vol. 881},
pages={012120},
organization={IOP Publishing},
}

@article{wang2020secure,
serialno={7},
author={Wang, Xiaonan and Cai, Shaohao},
year={2020},
title={Secure healthcare monitoring framework integrating \?{NDN}-based {\?{IoT}}
with
edge cloud},
journal={\rvtFGCS},
volume={112},
pages={320--329},
orgname = {Elsevier},
}

@article{li2020secured,
serialno={8},
author={Li, Junxia and Cai, Jinjin and Khan, Fazlullah and Rehman, Ateeq Ur
and Balasubramaniam, Venki and Sun, Jiangfeng and others},
year={2020},
title={A Secured Framework for \?{SDN}-Based Edge Computing in \?{IoT}-Enabled
Healthcare System},
journal={\rvtIEEEA},
volume={8},
pages={135479--135490},
orgname = {IEEE},
}

@article{akhbarifar2020secure,
serialno={9},
author={Akhbarifar, Samira and Javadi, Hamid Haj Seyyed and Rahmani, Amir
Masoud and Hosseinzadeh, Mehdi},
year={2020},
title={A secure remote health monitoring model for early disease diagnosis in
cloud-based {\?{IoT}} environment},
journal={Pers Ubiquitous Comput},
pages={1--17},
orgname = {Springer},
}

@article{sengupta2020secure,
serialno={10},
author={Sengupta, Souvik and Bhunia, Suman Sankar},
year={2020},
title={Secure data management in cloudlet assisted iot enabled e-health
framework in smart city},
journal={\rvtIEEESenJ},
volume={20},
number={16},
pages={9581--9588},
orgname = {IEEE},
}

@article{abou2020ditrust,
serialno={11},
author={Abou-Nassar, Eman M and Iliyasu, Abdullah M and El-Kafrawy, Passent M
and Song, Oh-Young and Bashir, Ali Kashif and Abd El-Latif, Ahmed A},
year={2020},
title={\?{DITrust chain: towards blockchain-based trust models for sustainable
healthcare {\?{IoT}} systems}},
journal={\rvtIEEEA},
volume={8},
pages={111223--111238},
orgname = {IEEE},
}

@article{alam2020mhealth,
serialno={12},
author={Alam, Tanweer},
year={2020},
title={mHealth Communication Framework using blockchain and {\?{IoT}}
Technologies},
journal={Int J Sci Technol Res},
volume={9},
number={6},
}

@article{saba2020secure,
serialno={13},
author={Saba, Tanzila and Haseeb, Khalid and Ahmed, Imran and Rehman, Amjad},
year={2020},
title={Secure and energy-efficient framework using \?{Internet of Medical Things}
for e-healthcare},
journal={J Infect Publ Health},
volume={13},
number={10},
pages={1567--1575},
orgname = {Elsevier},
}

@article{gope2020secure,
serialno={14},
author={Gope, Prosanta and Gheraibia, Youcef and Kabir, Sohag and Sikdar,
Biplab},
year={2020},
title={A secure \?{IoT}-based modern healthcare system with fault-tolerant
decision making process},
journal={IEEE J Biomed Health Inform},
orgname = {IEEE},
}

@article{srivastava2020future,
serialno={15},
author={Srivastava, Gautam and Parizi, Reza M and Dehghantanha, Ali},
year={2020},
title={The future of blockchain technology in healthcare internet of things
security},
journal={Blockchain Cybersecur Trust Priv},
pages={161--184},
orgname = {Springer},
}

@article{rahman2020adversarial,
serialno={16},
author={Rahman, Abdur and Hossain, M Shamim and Alrajeh, Nabil A and Alsolami,
Fawaz},
year={2020},
title={Adversarial examples--security threats to \?{COVID}-19 deep learning
systems in medical {\?{IoT}} devices},
journal={IEEE Internet Things J},
orgname = {IEEE},
}

@article{qiu2021hospital,
serialno={17},
author={Qiu, Famei},
year={2021},
title={Hospital Archives Intelligent Management System Based on {\?{5G}} Network
and
Internet of Things System},
journal={Microprocess Microsyst},
volume={80},
artnum={103564},
orgname = {Elsevier},
}

@article{nivaashini_deep_2018,
serialno={18},
author={Nivaashini, M. and Soundariya, R.S.},
year={2018},
title={Deep Boltzmann machine based breast cancer risk detection for
healthcare systems},
journal={Int J Pure Appl Math},
volume={119},
pages={581--590},
}

@article{leung_machine_2015,
serialno={19},
author={Leung, Michael KK and Delong, Andrew and Alipanahi, Babak and Frey,
Brendan J.},
year={2015},
title={Machine learning in genomic medicine: a review of computational
problems and data sets},
journal={Proc IEEE},
volume={104},
number={1},
pages={176--197},
orgname={ISBN: 0018-9219 Publisher: IEEE},
}

@article{pratt_convolutional_2016,
serialno={20},
author={Pratt, Harry and Coenen, Frans and Broadbent, Deborah M. and Harding,
Simon P. and Zheng, Yalin},
year={2016},
title={Convolutional neural networks for diabetic retinopathy},
journal={\rvtPComS},
volume={90},
pages={200--205},
orgname={ISBN: 1877-0509 Publisher: Elsevier},
}

@custombib{gozes2020rapid,
mypattern={[authors,atitle,midc][booktag,bktitle,yr,endbooktag,bibpages,bibdoi]
[post,myehost]},
serialno={21},
author={Gozes, Ophir and Frid-Adar, Maayan and Greenspan, Hayit and Browning,
Patrick D and Zhang, Huangqi and Ji, Wenbin and others},
year={2020},
title={Rapid \?{AI} development cycle for the coronavirus (\?{COVID}-19) pandemic:
\?{I}nitial results for automated detection \& patient monitoring using deep
learning \?{CT} image analysis},
post={arXiv preprint \myehost{arXiv:2003.05037}},
}

@custombib{wang2020covid,
mypattern={[authors,atitle,midc][booktag,bktitle,yr,endbooktag,bibpages,bibdoi]
[post,myehost]},
serialno={22},
author={Wang, Linda and Wong, Alexander},
year={2020},
title={\?{COVID-Net: \?{A} tailored deep convolutional neural network design for
detection of \?{COVID}-19 cases from chest radiography images}},
post={arXiv preprint \myehost{arXiv:2003.09871}},
}

@custombib{shan+2020lung,
mypattern={[authors,atitle,midc][booktag,bktitle,yr,endbooktag,bibpages,bibdoi]
[post,myehost]},
serialno={23},
author={Shan+, Fei and Gao+, Yaozong and Wang, Jun and Shi, Weiya and Shi, Nannan
and Han, Miaofei and others},
year={2020},
title={Lung infection quantification of covid-19 in ct images with deep
learning},
post={arXiv preprint \myehost{arXiv:2003.04655}},
}

@custombib{narin2020automatic,
mypattern={[authors,atitle,midc][booktag,bktitle,yr,endbooktag,bibpages,bibdoi]
[post,myehost]},
serialno={24},
author={Narin, Ali and Kaya, Ceren and Pamuk, Ziynet},
year={2020},
title={Automatic detection of coronavirus disease (\?{COVID}-19) using \?{X-ray}
images and deep convolutional neural networks},
post={arXiv preprint \myehost{arXiv:2003.10849}},
}

@article{greco2020trends,
serialno={25},
author={Greco, Luca and Percannella, Gennaro and Ritrovato, Pierluigi and
Tortorella, Francesco and Vento, Mario},
year={2020},
title={Trends in {\?{IoT}} based solutions for health care: \?{M}oving \?{AI} to
the edge},
journal={\rvtPatRecLet},
volume={135},
pages={346--353},
orgname = {Elsevier},
}

@inbook{gopalan2021iot,
serialno={26},
author={Gopalan, Subiksha Srinivasa and Raza, Ali and Almobaideen, Wesam},
year={2021},
title={{IoT} Security in Healthcare using \?{AI}: \?{A} Survey},
booktitle={2020 international conference on communications, signal processing,
and their applications},
pages={1--6},
organization={IEEE},
}

@article{bhuiyan2021internet,
serialno={27},
author={Bhuiyan, Mohammad Nuruzzaman and Rahman, Md Mahbubur and Billah, Md
Masum and Saha, Dipanita},
year={2021},
title={Internet of Things ({\?{IoT}}): \?{A} review of its enabling technologies in
healthcare applications, standards protocols, security and market
opportunities},
journal={IEEE Internet Things J},
orgname = {IEEE},
}

@article{hathaliya2020exhaustive,
serialno={28},
author={Hathaliya, Jigna J and Tanwar, Sudeep},
year={2020},
title={An exhaustive survey on security and privacy issues in \?{Healthcare 4.0}},
journal={\rvtComCom},
volume={153},
pages={311--335},
orgname = {Elsevier},
}

@article{ma2021understanding,
serialno={29},
author={Ma, Xingjun and Niu, Yuhao and Gu, Lin and Wang, Yisen and Zhao,
Yitian and Bailey, James and others},
year={2021},
title={Understanding adversarial attacks on deep learning based medical image
analysis systems},
journal={\rvtPatRec},
volume={110},
artnum={107332},
orgname = {Elsevier},
}

@article{mishra2021vmshield,
serialno={30},
author={Mishra, Preeti and Aggarwal, Palak and Vidyarthi, Ankit and Singh,
Pawan and Khan, Baseem and Alhelou, Hassan Haes and others},
year={2021},
title={VMShield: \?{M}emory Introspection-based Malware Detection to Secure
Cloud-based Services against Stealthy Attacks},
journal={IEEE Trans Ind Inform},
orgname = {IEEE},
}

\end{bibwrite}

\begin{biography}
%\includepic
\textbf{Senthil Kumar Jagatheesaperumal} is currently working as an Associate
Professor in the Department of Electronics and Communication Engineering, Mepco
Schlenk Engineering College, Sivakasi, Tamilnadu, India. He
received his Ph.D. in Robotics from Anna University, Chennai in 2017.
He received his Post Graduation degree from Anna University, Chennai,
in 2005. He received his B.E. degree from Madurai Kamaraj University,
Tamilnadu, India in 2003. His area of research includes Robotics, Internet
of Things, Embedded Systems, and Wireless Communication.
\end{biography}

\begin{biography}
%\includepic
\textbf{Preeti Mishra} is currently working as an Assistant Professor in the
Department of Computer Science in Doon University, Dehradun, UK, India (State
Government University). She has 10+ years teaching and research experience. She
received her Ph. D. in the field of Cloud Security from Malaviya
National Institute of Technology Jaipur. Her area of interest includes Cloud
Security, Internet of Things, Blockchain, Cyber Security, Mobile Security
etc.
\end{biography}

\begin{biography}
%\includepic
\textbf{Nour Moustafa} is Senior Lecturer and leader of Intelligent Security at
SEIT, University of New South Wales (UNSW)’s UNSW Canberra Australia. He was a
Post-doctoral Fellow at UNSW Canberra from June 2017
till December 2018. He received his Ph.D. degree in the field of Cyber Security
from UNSW Canberra in 2017. He obtained his Bachelor and Master
degree of Computer Science in 2009 and 2014, respectively, from the Faculty of
Computer and Information, Helwan University, Egypt. His areas
of interest include Cyber Security, in particular, Network Security, IoT security,
intrusion detection systems, statistics, Deep learning and machine
learning techniques.
\end{biography}

\begin{biography}
%\includepic
\textbf{Rahul Chauhan} received his M.Tech degree in the area of Instrumentation
\& Controls Systems in the year 2013 at Graphic Era University, Dehradun,
India. He received the B.Tech (Electronics \& Communication Engineering)
degree from Lovely Professional University, Jalandhar, Punjab, India 2005.
He has 7 years of teaching \& Research experience and currently working as
Assistant Professor in the Graphic Era Hill University, Dehradun, Uttarakhand,
India. He has published more than 15 Research papers in reputed
journals and conferences. His area of research includes Robotics, Internet
of Things, Machine learning \& deep Learning.
\end{biography}

\end{document}

You might also like