Hindawi

Journal of Healthcare Engineering

Volume 2020, Article ID 8894694, 15 pages
https://doi.org/10.1155/2020/8894694

Review Article
Applications of Artificial Intelligence and Big Data
Analytics in m-Health: A Healthcare System Perspective

Z. Faizal khan and Sultan Refa Alotaibi

College of Computing and Information Technology, Shaqra University, Shaqraa, Saudi Arabia

Correspondence should be addressed to Z. Faizal khan; [email protected]

Received 23 June 2020; Revised 9 August 2020; Accepted 19 August 2020; Published 1 September 2020

Academic Editor: Victor Albuquerque

Copyright © 2020 Z. Faizal khan and Sultan Refa Alotaibi. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
Mobile health (m-health) is the term of monitoring the health using mobile phones and patient monitoring devices etc. It has been
often deemed as the substantial breakthrough in technology in this modern era. Recently, artificial intelligence (AI) and big data
analytics have been applied within the m-health for providing an effective healthcare system. Various types of data such as
electronic health records (EHRs), medical images, and complicated text which are diversified, poorly interpreted, and extensively
unorganized have been used in the modern medical research. This is an important reason for the cause of various unorganized and
unstructured datasets due to emergence of mobile applications along with the healthcare systems. In this paper, a systematic
review is carried out on application of AI and the big data analytics to improve the m-health system. Various AI-based algorithms
and frameworks of big data with respect to the source of data, techniques used, and the area of application are also discussed. This
paper explores the applications of AI and big data analytics for providing insights to the users and enabling them to plan, using the
resources especially for the specific challenges in m-health, and proposes a model based on the AI and big data analytics for
m-health. Findings of this paper will guide the development of techniques using the combination of AI and the big data as source
for handling m-health data more effectively.

1. Introduction and Bluetooth-based technology. Big data [7–9] in the

healthcare contains the medical images [10], clinical data of
Mobile health is defined as the practice of applying mobile- doctor, doctors’ prescriptions and notes, computed to-
based devices such as the mobile phones, patient monitoring mography (CT) images, MRI scans, laboratory data, doc-
devices, personal digital assistants (PDAs), and other uments from the drugstore, files from the insurance EPR
wireless devices for the medical and public health. Therefore, data, and other data related to the administrative operations.
this process requires the application of mobile phone’s one This is increasingly becoming favored within the worldwide
of the most important benefits called the voice and short communities of healthcare. However, there is a deficiency of
messaging service (SMS). At present, more than 500 projects understanding the most suitable framework based on the
are there for the m-health and nearly 40,000 medical-based computational methodologies which are required for this
mobile applications are also available worldwide [1]. There approach. Big data analytics is the process of scrutinizing
are mobile-based medical devices which are designed spe- huge volume of data from various kinds of sources of data
cifically for monitoring the heart rate [2], level of glucose [3], [11, 12]. These data are of different presentations and de-
blood pressure [4], tracking the patterns of sleep [5], and also signs. Various analytical methods such as data mining and
for monitoring the activity of brain [6]. It also uses more AI can be put in to examine the data. Approaches for big
complicated operations and services such as the General data analytics can be used to identify the abnormalities
Packet Radio Service (GPRS), 3rd and 4th generation mo- obtained as a result of combining large volume of data from
bile-based technologies, Global Positioning System (GPS), different sources of data. Big data has become closely
2 Journal of Healthcare Engineering

associative with the mobile health in recent years [13]. The with the systematic reviews and meta-analysis process.
main problems of big data analytics and the m-health are yet Section 3 depicts the definition of m-health and its schematic
to be solved. representation along with the mobile sensors and their
Various works have been done recently as proposals applications in m-health. Section 4 explains a detailed review
[14–28] or review [15, 27, 29–32] on m-health and appli- about the applications of AI in m-health along with the
cations of AI and big data analytics in healthcare sector. performance measurement indicators used to examine the
Applications of mobile phones have been successfully quality of m-healthcare. Section 5 presents the applications
proven in medical-based applications for monitoring and of big data analytics in m-health followed by the additional
have enhanced in the possibilities of assessing clinical data summary of its applications in the healthcare sector. Section
[27, 33]. Methods such as experience sampling methods 6 presents the proposed model based on the AI and big data
(ESMs) and ecological momentary assessment (EMA) were analytics for m-health. Section 7 depicts the limitations of
applied in the process of assessing the patient’s relationships the proposed review. Conclusion and the future enhance-
between events and disease course [28]. These methods, ments are shown in Section 8.
which depend on providing contents which are informative
in nature and questionnaires which were self-administered,
reduce the recall since these applications will process in real
2. Motivation and Scope
time [34]. Recently, mobile devices can also able to perform At present, there are many papers that have been published
passive gathering of data, i.e., to gather the information recently as proposals or review on m-health and applications
about the users without any effort on their part. Processes of AI and big data analytics in healthcare sector. This paper
such as actigraphy, geolocation, and communication-based outlines the characteristics and applications, scope/health-
activities are usual features of current smartphones, and they care subarea, timeframe, and number of papers reviewed.
can also be used in collecting the patient’s behavior using the This review is intended to answer the following research
m-health-based systems. These m-health-based applications questions:
were also used to remotely monitor various physical and
mental conditions [31]. Mobile-based health application can (1) What is m-health and what sensors have been de-
use various sensors for generating self-report of a patient. veloped along with their applications for m-health?
The authors in [26] proposed a mobile application for (2) What applications and benefits could AI technology
recognizing the human activity from inertial sensors to bring to m-health?
determine the user’s activity level during the recording (3) What applications and benefits could big data ana-
process. The signal from heart rate and galvanic skin re- lytics bring to m-health?
sponse are also recorded in by their method to determine the
(4) What are the challenges of adopting AI and big data
emotional state of a user.
analytics technology in m-health? anda proposed m-
Following are the provocations that are still under
health model based on the combination of the AI and
consideration from the perspective of m-health:
big data analytics.
(i) Better perception of the organized and unorganized
The following sections describe how these questions
sources of data produced from different sources of
were answered by this systematic review.
mobile and information.
(ii) Smart implementation and conversion of the big
data of health data occurred from the users of 5G 2.1. Methodology. The methodology of our review followed
mobile health. This should be performed in order to the checklist proposed by the Preferred Reporting Items for
compare the most awaited intelligent and pre- Systematic Reviews and Meta-Analyses (PRISMA) [35]. This
defined change of behavior or convincing tools for review also identified applications of AI and big data ana-
inspiring more users for comfort and improvement lytics in m-health system. The review is limited to English
of their health. articles and reports from 2007 to present date.
(iii) Resilient, precise, and secure methods for data
analytics for the explication of huge data of medical
imaging and other relevant diagnostic data which 2.1.1. Relevant Articles. Relevant articles and process of their
are created and transferred from the future gen- selection for this systematic review are described in this
eration of mobile imaging devices should be section. In order to collect the relevant articles for this
developed. systematic review, we searched eight large scientific data-
bases: the IEEE Xplore, ACM digital library, Taylor &
The paper explores the applications of AI and big data Francis online, ScienceDirect, SAGE Journals, ProQuest,
analytics for providing insights to the users and enabling Springer, and Web of Science. This is done by an advance
them to plan, uses resources especially for the specific keyword searching process. The following terms were used
challenges in m-health, and a proposes a model based on AI in the search: “Artificial Intelligence AND m-Health,” “Big
and big data analytics for m-health. data analytics AND m-Health,” and “AI AND big data
The remainder of the paper is organized as follows. analytics in M-health.” Various articles were also found from
Section 2 shows the motivation and scope of this work along the Google Scholar search. The main aim of this search is to
Journal of Healthcare Engineering 3

find other quality articles that might be missed during the provide instantaneous, personalized informatics for indi-
initial search in scientific databases. viduals. m-Health could be a key element in healthcare
systems [29] and can be useful in monitoring health status
and improving patient safety and quality of care.
2.1.2. Inclusion and Exclusion of Articles. After completing m-Health is becoming more popular in the smart device
the process of searching the article, the authors concealed the sector as it can provide remote assistance and data collection.
titles and abstracts of the retrieved articles using an inclusion Unlike an individual healthcare service, the collected data
and exclusion criteria. The articles that were not in English, can be expanded and used across communities to under-
the articles lacking full text, the articles that do not represent stand common trends and thus improve the standards of
the applications of AI in m-health, the articles that do not healthcare. m-Health can provide support in vulnerable and
represent the applications of big data analytics in m-health, remote communities via improvements to networks and the
and the articles with insufficient details were excluded. All emergence of IoT [44].
the duplicate articles were removed. At last, 106 articles were The application of mobile technologies and their impact
obtained and kept for the review process. The above process are likely to increase in the coming years. Surveys showed
is explained in the form of PRISMA flowchart in Figure 1. that mobile technologies and devices held about 80% of the
overall global market in 2017, whereas in 2013, it was just
2.2. Results. A total of 2543 articles were retrieved from the 39%. The number of global users of smart mobile devices is
eight scientific databases. Then, another 130 additional ar- anticipated to almost double in 2020 compared with 2014
ticles were found through the search in Google Scholar. A and will reach 2.87 billion users [45]. This may increase the
total of 2437 articles have been excluded in the initial significance of m-health globally as shown in Figure 3. Low-
screening process. Among these, 1345 articles which do not cost smartphones have the required features and capabilities
represent the applications of AI in m-health, 902 articles to cope with health-related applications and include the
which do not represent the applications of big data analytics necessary connectivity [36].
in m-health, 78 articles which were from international As the popularity of m-health increases, countries are
journals, and 12 articles with insufficient details were ex- allotting more funding to this area helping society and
cluded. Flowchart of the systematic reviews and meta-an- communities to become more health literate. This promotes
alyses (PRISMA) is shown in Figure 1. wellness rather than expensive medical intervention and
hospitalization.
3. Mobile Health
The application of mobile phones has inadequacies in in- 3.1. Mobile Sensors and Their Applications for m-Health.
frastructure in developing countries which have led to huge There are many mobile sensors which can be applied for a
changes in various healthcare sectors. Recently, mobile various applications of health [21, 47–52]. Various sensors
technology has played a significant role in various fields of such as camera sensor [53–55], microphone sensor [56–58],
technologies among various subscribers in almost all the accelerometer sensor [59–61], and gyroscope sensor [59, 62]
countries. Mobile devices and communications assist the were used in the healthcare-based applications. Table 1
evolution of the proposed systems and their employment for shows a detailed outline of how the mobile-based sensors
the healthcare called m-health [36]. This comprises the can be applied for various healthcare-based applications.
combination of mobile devices, medical-based sensors [37],
and portable devices [23, 38]. Health-based applications on 4. Applications of Artificial
smartphones are classified into the following: general health Intelligence in m-Health
and fitness-based applications, information on medicine-
based applications, and applications for managing the Artificial intelligence is the process of demonstration of
healthcare. m-Health is the innovative application of up- intelligence by machines in disparity to the natural in-
coming mobile-based technologies in concurrence with telligence depicted by the humans [24, 75, 76]. Machine
wearable devices especially in the application of healthcare learning is one of the applications of AI that lay out the
informatics in order to enhance the practices of healthcare systems to create capability to learn automatically and to
[39, 40]. m-Health has a scope of applying it to the mobile- enhance it from its training without being programmed
based technologies. As a result, it produces various tech- explicitly. It also puts emphasis on the evolution of al-
nologies such as the wearable devices, embedded systems, gorithms, can obtain data, and can adopt it for the process
trackers for location, and legacy-based sensor devices. It also of making it to train themselves. Due to the fast en-
explores the facilitation in wireless-based communication hancement of the AI, it has been employed in various
[24, 41], ubiquitous computing, and other embedded fields, such as the IoT [22, 41, 77], machine vision [78],
technologies in healthcare to improve support of healthcare- driver assistance [79, 80], and natural language pro-
based applications and also to reach into different pastoral cessing [81, 82]. AI has been put in application in various
areas [42, 43]. The schematic representation of m-health domains of healthcare [83–87] which includes cancer
scenario is shown in Figure 2. research [88], cardiology [89], diabetes [90], mental
There are many advantages of using m-health. These health [91], identification of prognosis [92], identification
devices can apprehend, save, recover, and transmit data to of Alzheimer’s disease [93], identification of difference in
4 Journal of Healthcare Engineering

Identification
Records identified through Additional records identified
database searching through other sources
(n = 2413) (n = 143)

Records after removal of duplicates

(n = 2556)
Screening

Records screened Records excluded

(n = 2556) (n = 2437)

Full-text articles excluded, with reasons

Full-text articles assessed
(n = 2437)
for eligibility
Eligibility

(n = 113) Articles not representing the applications of

AI in m-health (n = 1345)
Articles not representing the applications of
big data analytics in m-health (n = 902)
Articles in which full text was not
Studies included in
available (n = 902)
qualitative synthesis
(n = 113) Articles from international journals (n = 78)
Articles with insufficient details (n = 12)
Included

Studies included in
quantitative synthesis
(meta-analysis)
(n = 113)

Figure 1: PRISMA flowchart for the entire review process.

Telemonitoring
Online/offline Web servers
devices
1 Wi-Fi/wireless

2 3
Network
7 6

Data gathering Data transport Data

Mobile Mobile 4
processing
phone Request phone
5 and storage
8

Data from users

Results from physician

Patients
Doctors
analyze data and share results
Figure 2: Schematic representation of the m-health scenario.
Journal of Healthcare Engineering 5

for example, sadness. In their analysis, they have selected

eight patients for identifying the depressive disorder, the
depression symptoms, anxiety, etc. Even though the accu-
racy of their proposed methodology is promising, the au-
thors suggested the proposed methodology has to be
enhanced since the outcome of prediction in the case of
mood and location has to be upgraded.
Hawley et al. [98] proposed an application of automated
machine in the recognition of speech of persons who are
affected with dysarthria. It also assists in the process of voice
message generation. In their method, the authors employed
the hidden Markov models to decide the overall proximity of
a word which is spoken to a speech model and is person-
alized for a particular person. Yet, the accuracy of their
methodology for the speech recognition is only 67% for real-
life study which comprises nine persons. The persons who
participated identified that the hurdles in the process of
communication are decreased by their proposed device
when compared with the already available method of
communication while speaking. Main drawback of their
methodology is that its support is done by a usual aid for the
Figure 3: Global m-health markets [46]. voice-output communication and the accuracy of speech
recognition hardware is less.
Martin et al. in [99] proposed a predicting and an alert
the clinical groups [94], identification of cardiovascular generating methodology about multiple modalities such as
disease [39], stroke-related studies [95], etc. lung diseases or cardiovascular problems in patients. Alerts
Larburu et al. in [96] proposed an m-health application were generated and sent to professionals of healthcare who
based on artificial intelligence for avoiding heart failures in can monitor the alerts based on the predefined guidelines.
patients. At present, the doctors are applying simple Their proposed system was based on the information col-
methods for generating alerts in the identification of heart lected through the phone calls of patients. Features such as
failure. More false alerts are generated in the present linguistic and metalinguistic were extracted along with the
methodology. In this work, predictive models were proposed status of patient in order to instruct the models of prediction.
to avoid the impact of these false alarms. These predictive A 70% positive predictive value was obtained for unplanned
models are based on clinical data taken from 242 heart events by their proposed methodology. Their proposed
failure patients’ mobile accumulated in 44 months. The methodology was tested in a controlled manner with a set of
finest predictive model is acquired by the merger of various 214 patients in a time period of six months. This is the
alerts which are based on observing the data and a set of biggest testing of an algorithm in terms of patient’s par-
questions using the application of a Naive Bayes classifier. ticipation and also with respect to the time taken. This
This proposed model lowered the false alerts for a patient for methodology depicted a reduction rate of 50% in the number
a year from 28.64 to 7.8 gradually. In this method, the of participants in unplanned events of hospitals in the group
proposed system forecasts the possible risk of heart failures when compared with custom alert generating mechanism.
among the patients with more possibility of a heart failure. Morrison et al. in [100] used the push notifications to
Main drawback of their method is that the accuracy of upgrade the application of smartphone users for the process
detection is less when the patient had undergone any heart of stress management. The authors have employed a clas-
surgeries in his past. sifier called Naive Bayes for predicting the response of a user.
Burns et al. in [97] depicted the importance of mobile- Their algorithm predicts if a user would respond for a
based multicomponent that can be applied in the models of personalized intelligent mechanism for notification delivery
AI in order to analyze the different types of emotions such as when a notification is received from it. It depends on the
the mood, cognitive state, depression, motivation, various number of times a user views and reacts within a day for the
activities, environmental behavior of the patient, and social messages he received. This methodology was carried out for
behavior of the patient. Their proposed methodology gives 72 hours which includes 76 participants. The drawback in
graphs for feedback for the process of behavioral self-re- this method is that the response is less when there is a
flection, and it also provides coaching using various special distraction in the mobile networks.
trainers. The proposed methodology is based on the com- Ortiz-Catalan et al in [101] used the pattern recognition
bination of regression along with decision trees and the algorithms for controlling the virtual limb movement in
phone sensor-based devices. Overall accuracy of their patients suffering from phantom limb pain. They also used
proposed methodology was excellent for the prediction of gaming-based methodology combined with augmented re-
location about 60%–91%. Main drawback of their method is ality for the process of treatment. Their proposed meth-
that the accuracy of prediction was very less for emotions, odology was trained with a group of 14 participants. The
6 Journal of Healthcare Engineering

Table 1: Mobile-based sensors applied for various healthcare-based applications.

Mobile sensors Main area Applications in healthcare
Applied for identifying various categories of diseases, in the
Capturing photo and perspective of effects in surgery, diagnosis of diseases, observing the
Camera
video slash, analysis of skin disease [63], monitoring the health of child, etc.
[18].
Provides an access to follow the patients who are vulnerable to some
GPS Location tracking diseases such as the people with Alzheimer’s disease [64] and Ebola
[65] by the application of mobile-based applications [66].
Mobile phones which are enabled with the electrocardiographs are
Cardiovascular disease
Electrocardiograph being used in areas which are underdeveloped for the purpose of
monitoring
monitoring the patients with heart diseases [40, 67].
Data sharing and It allows a midrange data communication between mobile devises,
Bluetooth
communication various other healthcare monitoring devices, and wearable sensors.
It allows the doctors to communicate with the patients regarding the
support for identification and treatment of diseases. It also comes up
Microphone Voice recording
with the way for analyzing the audio for assessing the feeling of a
patient with various diseases such as muscular dystrophy [68].
It assists to compute the orientation of devices which are relative to
Acceleration Earth especially for calculating the motion. It can be executed in
Accelerometer
measurement various activity monitoring techniques of patients such as counting the
step of a person, gait analysis, and monitoring [19, 69].
Wi-Fi-based mobile sensor enables the mobile device to communicate
Data sharing and
Wi-Fi with the physician about the healthcare data to for the purpose of
communication
identification of a disease and its treatments.
Accelerometer, GPS, compass, Combination of hardware and the sensors present in it is being utilized
Physical activities
gyroscope, and barometer for computing the stationary vs nonstationary actions [20].
This combination makes the monitoring of psychological health by
Microphone, accelerometer, and GPS Social engagement checking the social problems, talks from the conversationalists,
consternation, strain, behaviors related to depression, etc. [70, 71].
Microphone, GPS, accelerometer, touch Combination of this hardware depicts the data of interrupted vs
Sleep pattern tracking
interface, and light sensor constant patterns of sleep in a patient [71–74].

results revealed that about 50% symptoms of phantom limb effective in assessing how far the user is satisfied by
pain in patients were decreased significantly after 6 months the mobile healthcare system.
of treatment. The authors also recommended that their novel (2) Effectiveness. Effectiveness is defined as the extent to
method of treatment could be employed after clinical which the m-healthcare system app works in the way
treatments. One of the disadvantages of their methodology is that users expect it to and the ease with which users
the time frame. Table 2 depicts the additional summary of can apply it to achieve their specific goals. This is an
various applications of machine learning in the healthcare important metric used in the case of m-healthcare
sector. quality.
(3) Veracity. It is the measure of analyzing the accuracy
and reliability of the information, data, or content
4.1. Performance Measurement Indicators Used to Examine present in the m-health application. Content in
the m-Healthcare Quality. In order to assess the quality of health apps is usually based on more than one source
m-healthcare-based apps, various performance measure- of information. The m-health application provides a
ment indicators were proposed earlier. These performance method to enable the user to identify to the complete
measure indicators were proposed by incorporating the content more easily. Most of the m-health-based
challenges of mobile health apps and strategies to ensure systems perform the functions of user or patient
appropriate design and development of the apps for management, such as computation, tracking the
healthcare providers, patients, and the general public. Fol- data, and reminders, which should be more accurate.
lowing are the various performance measurement indicators
(4) Interactivity. It is the process of providing a sense of
used to examine the quality of m-healthcare:
engagement with the user, entertainment, satisfac-
(1) Usefulness. This metric enables the m-health user to tion to the patient or user, and motivation for the
achieve his or her specific goals and motivates the users who are using the m-health systems. It also
user to use the app repeatedly whenever necessary. extends to interactivity between service providers
This metric also analyzes how the mobile platform is and patients as facilitated by the m-health app.
Journal of Healthcare Engineering 7

Table 2: Additional summary of the AI methods suitable for the healthcare sector.
Name of the
System Technique Area of application
framework
Apache Mahout Library for machine A real-time computation system which is Provides mechanisms such as clustering,
[102] learning (open source) more flexible and scalable. classification, and regression.
AI-based platform which Applies artificial intelligence for producing For processing very large organized and
Skytree [103] is applied for general complicated algorithms for more advanced unorganized datasets more accurately
purpose algorithms analytics. without performing downsampling.
Develops and issues a graphical-based
Searches and scrutinizes the web-based,
environment which assists the way finding
Karmasphere [104] Platform of big data mobile-based, and sensor-based data in
through any type of big data and identifies
Hadoop for the social media.
the recent trends and patterns present in it.
It combines the AI-based features along
Gives various tools for performing tasks
with the cloud-based infrastructure for
Platform for AI-based related to AI such as clustering, regression
BigML [105] developing applications which are cost-
programs analysis, pattern classification, detection of
effective, highly accurate, reliable, and
anomaly, and discovery of association.
flexible.
Echocardiography data are normalized
Cognitive machine using the machine learning algorithm in
Cognitive computing Associative memory classifier-based
learning algorithm order to differentiate the constrictive
tool machine learning algorithm.
[106] pericarditis from restrictive
cardiomyopathy.
Analyzes and classifies a multidimensional
Machine learning To distinguish between athlete heart and
Support vector machine echocardiographic data based on gap in
algorithms [107] hypertrophic cardiomyopathy.
present in it.
To analyze the clustering of
Classifies similar objects between the same
Phenotypic echocardiographic variables in order to
Hierarchical clustering clusters and calculates the hierarchy in the
clustering [108] compute the dysfunction in left ventricular
echocardiographic data.
and isolate high-risk phenotyping patterns.
Combination of AI and It reads the chest X-ray reports of patients It combines the AI-based features along
Convolutional
natural language and assists the antibiotic assistant system to with the natural language processing for
neural network [109]
processing alert physicians for anti-infective therapy. effective diagnosis of diseases.

(5) Customization. The main purpose of designing the are difficult to manage with normal hardware, software,
m-health-based system is to support the users in one tools, and methods for managing the data [11]. Big data in
or more healthcare domains. Examples include as- the healthcare consists of clinical details of doctors, their
sessment of diseases, its diagnostics, prevention of notes and prescriptions, CT images, MRI images, labo-
further complications, expert’s intervention, and ratory data, documents from the drugstore, files from the
recovery. Customization is crucial in aiding the insurance and other data related to the administrative
m-health-based system to achieve what the users operations, EPR data, etc. This comprises the big data.
intend to do. For example, the systems may have to More methods have been proposed by various researchers
connect to one or more EHR systems to provide the to process these types of data. Still, there is a deficiency of
medical data of a particular user/patient. understanding the most suitable framework based on the
(6) User Satisfaction. User satisfaction can be defined as computational methodologies which are required for this
the proven willingness of a user for specified tasks in approach. Hence, an enormous amount of data belonging
the overall m-health system or in using a specific to the healthcare is available for big data scientists. By
system for repeated emergencies. This user accept- understanding the advantages and disadvantages present
ability has replaced most of the traditional metrics in this, the big data analytics has to be enhanced in order
already available for assessing the usability in mobile to save the lives and to reduce the cost of processing data.
health systems. Therefore, big data can be classified into two main cat-
egories [36] as follows:
5. Applications of Big Data (i) Organized data: in general, these data refer to the
Analytics in m-Health contents having defined format and length such as
the numbers, generated date, and contents of strings.
Recently, big data analytics has various options of pro- These data are formed by various sources such
viding advanced care for the patient and clinical decision mobile phones, computers, various sensors, and logs
support in the healthcare [14–17, 110, 111]. In general, of web. Examples of these types of data include EHR,
application of big data in healthcare refers to the elec- home treatment and monitoring data, prescriptions
tronic datasets of health which are huge and complex and from the doctors, etc.
8 Journal of Healthcare Engineering

(ii) Unorganized data: in general, these data refer to the A patient-centric personalized framework for healthcare
contents which do not have a predefined format of based on the collaborative filtering approach was proposed by
big data. The majority of the data are generated from Chawla and Davis in [113]. It apprehends the similarities in
various sources, such as the data from social media, different patients and generates the personalized profiles for
mobile data, and content from the video and web. risky diseases for individuals. Collaborative filtering is one
Examples of unorganized health data include health form of data analysis technique which is designed to guess the
data from the social platform such as from Twitter, opinion of user regarding an entity item or its service; it is
Facebook, user blogs, notes of clinicians, and diaries based on the preferences from a known group of a large
of medication and its instructions. number of users. In their framework, healthcare history of
individual patients was collated with all the medical histories
The process of analyzing a huge amount of data from of other available patients. This is based on the following
various sources of data and different formats in order to similarity constraints. Some of these are occupation, symp-
convey the perception of enabling a decision-making pro- tom, result from the laboratory, history of family, data of
cess in real time is called big data analytics. Various concepts demography, etc. Based on the computation of similarity, a
of analytics such as data mining and AI can be used to collection of patients who are similar is chosen and the
analyze the obtained data. These analytical approaches in big prediction of diseases is done. Since the application of elec-
data can be used to identify the anomalies by analyzing a tronic healthcare records was increased, their framework
huge amount of data from various datasets and their sources. depicts a proactive healthcare solution with respect to the
Figure 4 shows an example of the smartphone-based context of big data. Even though their proposed methodology
m-health model with the combination of AI and big data has various advantages, their proposed methodology handles
analytics. Nowadays, the conversion of digital version of all only the identification of codes for various diseases.
exams done in clinical and medical fields yields huge data An analytical framework of big data that employs
and records, which has formed a standard and has been ubiquitous healthcare system was proposed by Kim et al. in
widely accepted and implemented in practice. [114]. Their proposed framework analyzes the vital signs
EHRs are defined as the computerized form of medical obtained from accelerometers in order to provide healthcare
records for all the patients. It has various information regarding services. Vital signs are continuous time series data which are
the previous, current, and upcoming physical and the mental unstructured in nature having inadequacy to be stored in the
health situation of an individual. These electronic systems are traditional databases. Data obtained from ECG and from the
used to apprehend, transfer, obtain, stock, connect, and change respiratory system are considered as vital signs. Their pro-
the data of multimedia. The primary purpose of this electronic posed framework used a platform of open standard in order to
system is to provide services related to the health [45]. Main support the inability of data exchange between various de-
advantages of these EHRs are that they enable faster retrieval of vices. This platform has been enlarged by including various
data and the professionals in healthcare have an enhanced algorithms for the process of extracting feature values from
access to the whole history of the patient about his medical the fresh vital signs data and then storing them for the process
details. Its benefits include providing better healthcare by of real-time analysis. Even though their proposed method-
making better classifications of the patient’s health. ology has various advantages, their work has a major dis-
Similar to EHR, another record called electronic medical advantage in delivering considerable analytical models.
record (EMR) is used to store the medical and clinical data A detailed survey on the inference of computational
which are gathered from the patients. These are standard in methods in the big data-based health informatics has
nature. EHRs, EMRs, PHR, software for the medical practice been done by Fang et al. in [30]. They focused on a novel
management, and various other components of the framework called “Health informatics processing pipe-
healthcare data increase the quality and efficiency of service line” which incorporates various steps to obtain signifi-
and reduce the overall cost of healthcare and medical errors. cant patterns from healthcare-based big data. Their
The healthcare big data consist of the data from healthcare proposed framework consists of pipeline process such as
provider and various experiments done in the laboratories capturing the data, storing the data, analysis, extraction,
and various other data obtained from the IoT-based devices. and decision support systems. Apart from the proposed
Raghupathi and Raghupathi [112] proposed a novel ar- framework, some directions for research in the hetero-
chitecture for the healthcare-based system applying the analytics geneity of data such as organized and unorganized data of
of big data. Their methodology comprises various layers for data the healthcare, existing complexities which are available
source, transformation, big data platform, and analytics. The in the available data, issues of privacy, and analysis of the
layer for data source mainly focuses on the data sources of identified patterns are also traversed in their entire work.
internal and external healthcare which can be found in different Their proposed healthcare-based framework offers a
locations and in different formats. The layer for transformation systematic pipeline of data processing for various stages
is accountable for various tasks such as removal, conversion, of informatics of big data such as data acquisition, saving,
and uploading of data in the platform of big data for the process finding, and analyzing data from diversified sources.
of doing specific operations on the Distributed File System using Hence, the authors focused on enhancing the aspects of
a programming model called Map-Reduce. The main task of technological development by using the tools and tech-
analytical layer is to do various operations such as inquiring, niques of big data. Due to the enhancement of mobile
announcing, online analytical processing, and mining the data. devices and wireless sensor networks, healthcare services
Journal of Healthcare Engineering 9

1. Collect
2. Analyze and
and record the
process the data
data

4. Action taken 3. Trigger

by clinician alert to the patient
Wireless medical
sensor/wearable Trigger alerts
Automated
connectivity
IoT data Mobile care
collection monitor

Smartphone Artificial intelligence and big data analytics Notification

EHR

Figure 4: Smartphone-based m-health model with AI and big data analytics.

were improved. As a result, the services are offered at any the applications of big data in the healthcare sector is
time and at anywhere in the health informatics domain. provided in Table 3.
Pramanik et al. in [115] performed a detailed analysis on
the latest improvements in healthcare-based systems. Their
work mainly focuses on the applications of technologies 6. Proposed Model Based on AI and Big Data
based on smart system. They focused on a novel framework Analytics for m-Health
for the smart healthcare system enabled by big data for
maintaining ubiquitous solutions for healthcare. It also The proposed framework comprises three essential parts
offered a reduced cost with improved advancements. It such as the medical data obtained from the patients through
consisted of the following layers: the mobile phone and the telemonitoring devices, AI and big
data analytics platform, and the output towards the mobile
(i) Data source layer: designed especially for main- care monitor. The architecture of the proposed system is
taining the organized, unorganized, and semi- shown in Figure 5. The entire process of analyzing a huge
organized data sources. amount of data obtained from various sources of data in
(ii) Data analytics layer: designed mainly for processing different formats is processed by the combination of AI and
calculations on big data, its visualization, and big data platform. These are combined to convey the per-
management. ception of enabling a decision-making process in real time.
(iii) Smart service layer: designed mainly for making Various concepts of analytics such as data mining and AI are
ease of various favors such as the monitoring of used to analyze the obtained data from a patient. These
data, agreement on privacy, and security between analytical approaches in big data can be used to identify the
the providers, consumers, and their services. Also, anomalies by analyzing a huge amount of data from various
this layer proposes various smart services and their datasets and their sources such as biomedical signals,
infrastructure with the help of various devices and physiological sensing data, genomic data, and biomedical
software. imaging. The AI-based engine comprises two modules such
as the stream analysis module and the AI-based report
(iv) Knowledge discovery layer: improved functional-
management tool. These analyze the queries obtained from
ities such as the guessing necessity of entities,
the big data analysis engine.
proposal, and its cost evaluation of mechanism for
The main aim of the AI-based report management tool is
providing the healthcare service were also
to generate a better decision using the AI technology in order
included.
to report the status of the patient’s health. It is also used as a
The authors proposed a framework for the organizations platform for the disease control, treatment, and diagnosis
in healthcare in providing intelligence-based smart services. tool. In this model, the AI-based report management tool
Their detailed research depicts a novel framework for the collects, analyzes, performs, and triggers the action by
smart healthcare system based on big data and also makes classifying the code of a disease or condition using the free
the research directions interdisciplinary. In fact, the pro- text approach. It also extracts the features from the EHR. It
posed framework is the combination of three technical also detects the irregular records which are present in the
streams such as the AI, agent-based systems, and data EHR. All the processed streams are stored and updated in
mining along with the smart health. Additional summary of the big data engine.
10 Journal of Healthcare Engineering

Table 3: Additional summary of the applications of big data in the healthcare sector.
Name of the framework Source of data Technique Area of application
Substructure for preserving Reliable healthcare-based services.
Privacy preservation
privacy in healthcare systems Data produced from the tags of RFID Enhanced isolation in healthcare
methods
based on RFID [116] system based on RFID.
Providing security
Novel framework for Secure healthcare system.
limitation and control
distributed and secured HIS Electronic-based health records Distributed and secured multitier
mechanisms for accessing
[46] framework.
the data
Technologies based on smart system
especially for the healthcare system.
Smart framework for EHR, report on diagnosis, data from Providing services of smart
Combining the healthcare
healthcare system enabled the social media, biometric data, and healthcare by infrastructure
knowledge data mining strategies
with big data [115] monitoring data which is service oriented
with the infrastructure of smart
services.
Patients’ various biological
Providing access control
Framework for policy parameters, data related to Smart health applications for
based on policy mechanism
enforcement towards IoT- environmental factors, and data avoiding threats in security for large
for offering resources of
based smart health [117] generated from the instruments such scale and heterogeneous scenarios.
healthcare
as RFID
Framework for prediction of
Ensemble learning Design of drugs. Depicts a
protein structure using big
Protein structure dataset technique based on distributed framework with
data and ensemble learning
distributed tree enhanced accuracy.
[118]
Big data-based analytics for the
Datasets of the patient from various applications of smart healthcare.
Framework for smart health sources such as the health Pattern recognition and its Improving the services of healthcare
[44] information system and the matching techniques by combining the sensor-based
radiology department technologies along with the cloud
computing and big data analytics.
A semantic web-based Classification of patient based on
Building the ontology
technology for maintaining various clinical criteria. Combining
EHR through ontology web
and reusing the archetypes the semantic-based resources along
language
present in clinical data [119] with the EHR.

Large medical data AI and big data analytics platform Mobile care
monitor

Data from doctors and IoT-based

clinicians devices
AI-based report
Stream analysis
management
module
tool
Data from the patients Mobile phone

Doctors and
Medical images
medical
service
providers

RFID data Big data Statistical data

storage analysis tool
Web server

Figure 5: Architecture of the proposed AI and big data analytics-based m-health system.
Journal of Healthcare Engineering 11

The big data analysis engine consists of two modules based systems for providing privacy and security in m-health
such as storage for big data and a statistical data analysis tool. big data.
The statistical data analysis tool retrieves the input data,
processes it into queries, and then sends it to the AI-based Data Availability
engine. All the processed queries and streams were given as
output towards the mobile care monitor. The data that support the findings of this study are available
The proposed model enhances the overall performance from the corresponding author on request.
of m-health since AI and big data analytics are combined.
The proposed methodology improves the process of Conflicts of Interest
m-health by processing each and every query, and it also
enables a decision-making process in real time. The authors declare that there are no conflicts of interest
regarding the publication of this paper.

7. Limitations References
Despite the various advantages of the proposed m-health [1] P. S. Sousa, D. Sabugueiro, V. Felizardo, R. Couto, I. Pires,
model based on AI and big data analytics, some limitations and N. M. Garcia, “mHealth sensors and applications for
were also there that need to be considered: a large section of personal aid,” in Mobile Health, S. Adibi, Ed., Springer,
population, system can never be too accurate, have to de- Berlin, Germany, Springer Series in Bio-/Neuroinformatics
pend completely on the technology, and several privacy and vol 5, 2015.
security issues. [2] S. Majumder and M. J. Deen, “Smartphone sensors for health
With regard to a large section of population, the access to monitoring and diagnosis,” Sensors, vol. 19, no. 9, p. 2164,
m-health-based system is denied because of their numbers, 2019.
their incapacity to afford it, and the lack of knowledge and [3] M. Darrell, “Improving health care through mobile medical
skill to use it. The system can never be too accurate to replace devices and sensors,” Brookings Institution Policy Report,
vol. 10, pp. 1–13, 2013.
the humans and their predictions. These systems have been
[4] H. Jamaladin, “Mobile apps for blood pressure monitoring:
made to ease out the health structure but they cannot be a systematic search in app stores and content analysis,” JMIR
substitute to human. Even the most well designed and mHealth and uHealth, vol. 6, no. 11, 2018.
technologically best developed apps can also never be [5] J.-Y. Jenny, A. Bureggah, and Y. Diesinger, “Measurement of
hundred percent accurate. the knee flexion angle with smartphone applications: which
These m-health systems also make a user/patient to be technology is better?” Knee Surgery, Sports Traumatology,
dependent completely on them. If the user loses his or her Arthroscopy, vol. 24, no. 9, pp. 2874–2877, 2016.
mobile phone and user id/password, there is a possibility for [6] J. F. Huckins, “Fusing mobile phone sensing and brain
all the information to be lost temporarily or even perma- imaging to assess depression in college students,” Frontiers in
nently. There might be a chance for various issues in the Neuroscience, vol. 13, no. 248, p. 21, 2019.
[7] A. McAfee, E. Brynjolfsson, T. H. Davenport, D. J. Patil, and
privacy and security of the health data present in it. In such
D. Barton, “Big data: the management revolution,” Harvard
cases, there is a chance for the personal information to be Business Review, vol. 90, no. 10, pp. 60–68, 2012.
leaked and shared to unauthorized users. [8] C. Lynch, “How do your data grow?” Nature, vol. 455,
no. 7209, pp. 28-29, 2008.
8. Conclusion and Future Works [9] A. Jacobs, “The pathologies of big data,” Communications of
the ACM, vol. 52, no. 8, pp. 36–44, 2009.
m-Health is a technique which uses mobile devices and tech- [10] F. Ritter, T. Boskamp, A. Homeyer et al., “Medical image
analysis,” IEEE Pulse, vol. 2, no. 6, pp. 60–70, 2011.
nology for health interventions and is the biggest technological
[11] X. Wu, X. Zhu, G.-Q. Wu, and W. Ding, “Data mining with
advancement of recent research. Similarly, the application of AI big data,” IEEE Transactions on Knowledge and Data Engi-
and the analytics of big data in healthcare are considered as one neering, vol. 26, no. 1, 2014.
of the important achievements for the intelligent healthcare [12] A. Belle, R. Thiagarajan, S. M. Reza Soroushmehr, F. Navidi,
system. In this paper, a detailed review of the m-healthcare D. A. Beard, and K. Najarian, “Big data analytics in
system is proposed based on the application of AI and big data healthcare,” BioMed Research International, vol. 2015, Ar-
analytics. Various advantages from this combination for the ticle ID 370194, 16 pages, 2015.
m-health perspective are presented. Particularly, all applications [13] A. Y Sun and B. R. Scanlon, “How can Big Data and machine
of relevant technological areas and the building blocks such as learning benefit environment and water management: a
communications, sensors, and computing which are associated survey of methods, applications, and future directions,”
with mobile health are explained in detail. The role of various Environmental Research Letters, vol. 14, no. 7, Article ID
073001, 2019.
tools of machine learning within the current m-health model is
[14] T. Huang, L. Lan, X. Fang, P. An, J. Min, and F. Wang,
also illustrated. Future works can be a comprehensive review on “Promises and challenges of big data computing in health
the retrospective validation of models of the AI and combining sciences,” Big Data Research, vol. 2, no. 1, pp. 2–11, 2015.
them with various digital health tools and evaluating their [15] B. Cyganek, M. Graña, B. Krawczyk et al., “A survey of big
clinical validation and efficacy issues on these systems. Future data issues in electronic health record analysis,” Applied
works can be the proposal of application of intelligent agent- Artificial Intelligence, vol. 30, no. 6, pp. 497–520, 2016.
12 Journal of Healthcare Engineering

[16] H. Asri, H. Mousannif, H. Al Moatassime, and T. Noel, “Big [30] R. Fang, S. Pouyanfar, Y. Yang, S.-C. Chen, and S. S. Iyengar,
data in healthcare: challenges and opportunities,” in Pro- “Computational health informatics in the big data age,”
ceedings of the international Conference on cloud Technologies ACM Computing Surveys, vol. 49, no. 1, pp. 1–36, 2016.
and applications (CloudTech), pp. 1–7, Marrakech, Morocco, [31] E. Reinertsen and G. D. Clifford, “A review of physiological
June 2015. and behavioral monitoring with digital sensors for neuro-
[17] W. Raghupathi and V. Raghupathi, “Big data analytics in psychiatric illnesses,” Physiological Measurement, vol. 39,
healthcare: promise and potential,” Health Information no. 5, 2018.
Science and Systems, vol. 2, no. 1, p. 3, 2014. [32] S. Berrouiguet, M. L. Barrigón, J. L. Castroman et al.,
[18] L. De Greef, M. Goel, M. J. Seo et al., “Bilicam: using mobile “Combining mobile-health (mHealth) and artificial intelli-
phones to monitor newborn jaundice,” in Proceedings of the gence (AI) methods to avoid suicide attempts: the
2014 ACM International Joint Conference on Pervasive and Smartcrises study protocol,” BMC Psychiatry, vol. 19, p. 277,
Ubiquitous Computing, pp. 331–342, ACM, Seattle, WC, 2019.
USA, 2014. [33] S. Sakr and A. Elgammal, “Towards a comprehensive data
[19] Z. Chen, C. Jiang, and L. Xie, “A novel ensemble ELM for analytics framework for smart healthcare services,” Big Data
Research, vol. 4, pp. 44-45, 2016.
human activity recognition using smartphone sensors,” IEEE
[34] J. Asselbergs, J. Ruwaard, M. Ejdys, N. Schrader,
Transactions on Industrial Informatics, vol. 15, no. 5,
M. Sijbrandij, and H. Riper, “Mobile phone-based unob-
pp. 2691–2699, 2019.
trusive ecological momentary assessment of day-to-day
[20] C. A. Ronao and S.-B. Cho, “Human activity recognition
mood: an explorative study,” Journal of Medical Internet
with smartphone sensors using deep learning neural net-
Research, vol. 18, no. 3, 2016.
works,” Expert Systems with Applications, vol. 59, pp. 235– [35] D. Moher, A. Liberati, J. Tetzlaff, D. G. Altman, and
244, 2016. P. Group, “Preferred reporting items for systematic reviews
[21] Z. Chen, Q. Zhu, Y. C. Soh, and L. Zhang, “Robust human and meta-analyses: the PRISMA statement,” PLoS Medicine,
activity recognition using smartphone sensors via CT-PCA vol. 6, no. 7, Article ID e1000097, 2009.
and online SVM,” IEEE Transactions on Industrial Infor- [36] R. S. H. Istepanian and B. Woodard, m-health: Fundamentals
matics, vol. 13, no. 6, pp. 3070–3080, 2017. and Applications, John Wiley, Hoboken, NJ, USA, 2017.
[22] T. Zhang, “A joint deep learning and internet of medical [37] M. Muzammal, R. Talat, A. H. Sodhro, and S. Pirbhulal, “A
things driven framework for elderly patients,” IEEE Access, multi-sensor data fusion enabled ensemble approach for
vol. 8, pp. 75822–75832, 2020. medical data from body sensor networks,” Information
[23] A. H. Sodhro, S. Pirbhulal, G. H. Sodhro, A. Gurtov, Fusion, vol. 53, pp. 155–164, 2020.
M. Muzammal, and Z. Luo, “A joint transmission power [38] S. Balandin, E. Balandina, Y. Koucheryavy, V. Kramar, and
control and duty-cycle approach for smart healthcare sys- O. Medvedev, “Main trends in mHealth use scenarios,”
tem,” IEEE Sensors Journal, vol. 19, no. 19, pp. 8479–8486, Journal on Selected Topics in Nano Electronics and Com-
2019. puting, vol. 1, no. 1, pp. 64–70, 2013.
[24] Z. F. Khan and A. Kannan, “Intelligent approach for seg- [39] A. Triantafyllidis, D. Filos, R. Buys et al., “Computerized
menting CT lung images using fuzzy logic with bitplane,” decision support for beneficial home-based exercise reha-
Journal of Electrical Engineering and Technology, vol. 9, no. 4, bilitation in patients with cardiovascular disease,” Computer
pp. 742–752, 2014. Methods and Programs in Biomedicine, vol. 162, pp. 1–10,
[25] A. H. Sodhro, G. Fortino, S. Pirbhulal, M. Muhammad 2018.
Lodro, and M. A. Shah, Energy-efficiency in Wireless Body [40] Qualcomm Wireless Reach Initiative, “Wireless heart health:
Sensor Networks”, Book Title: Networks of the Future Ar- using 3G to assist underserved patients with cardiovascular
chitectures, Technologies, and Implementations, Chapman & disease,” 2011, https://www.qualcomm.com/media/documents/
Hall/CRC Computer and Information Science Series, p. 492, files/china-heart-health.pdf.
CRC Press (Taylor & Francis Group), Boca Raton, FL, [41] S. Pirbhulal, N. Pombo, V. Felizardo, N. Garcia,
USACRC Press (Taylor & Francis Group), 2017. A. H. Sodhro, and S. C. Mukhopadhyay, “Towards machine
[26] A. Nazabal, P. Garcia-Moreno, A. Artes-Rodriguez, and learning enabled security framework for iot-based health-
care,” in Proceedings of the 2019 13th International Confer-
Z. Ghahramani, “Human activity recognition by combining
ence on Sensing Technology (ICST), pp. 1–6, Sydney,
a small number of classifiers,” IEEE Journal of Biomedical
Australia, December 2019.
and Health Informatics, vol. 20, no. 5, pp. 1342–1351, 2016.
[42] L. H. Iwaya, M. A. L. Gomes, M. A. Simplı́cio et al., “Mobile
[27] E. Dogan, C. Sander, X. Wagner, U. Hegerl, and E. Kohls,
health in emerging countries: a survey of research initiatives
“Smartphone-based monitoring of objective and subjective
in Brazil,” International Journal of Medical Informatics,
data in affective disorders: where are we and where are we vol. 82, no. 5, pp. 283–298, 2013.
going? Systematic review,” Journal of Medical Internet Re- [43] S. P. Bhavnani, J. Narula, and P. P. Sengupta, “Mobile
search, vol. 19, no. 7, 2017. technology and the digitization of healthcare,” European
[28] S. Berrouiguet, M. L. Barrigón, S. A. Brandt et al., “Ecological Heart Journal, vol. 37, no. 18, pp. 1428–1438, 2016.
assessment of Clinicians’ antipsychotic prescription habits in [44] S. Madanian, D. Airehrour, N. Kumar et al., “Smart cap for
psychiatric inpatients: a novel web- and Mobile phone-based visually impaired in disaster situations,” in Proceedings of the
prototype for a dynamic clinical decision support system,” ISCRAM Asia Pacific, Wellington, New Zealand, 2018.
Journal of Medical Internet Research, vol. 19, no.1 page, 2017. [45] M. Reisman, “EHRs: The challenge of making electronic data
[29] A. Bassi, O. John, D. Praveen et al., “Current status and usable and interoperable,” Pharmacology & Therapeutics,
future directions of m-Health interventions for health system vol. 42, no. 9, pp. 572–575, 2017.
strengthening in India: systematic review,” JMIR Mhealth [46] S. Madanian, D. T. Parry, D. Airehrour, and M. Cherrington,
Uhealth, vol. 6, 2018. “mHealth and big-data integration: promises for healthcare
Journal of Healthcare Engineering 13

system in India,” BMJ Health & Care Informatics, vol. 26, Cybernetics—SMC, pp. 182–187, San Diego, CA, USA, Oc-
Article ID e100071, 2019. tober 2014.
[47] D. De, P. Bharti, S. K. Das, and S. Chellappan, “Multimodal [63] A. Bourouis, A. Zerdazi, M. Feham, and A. Bouchachia, “M-
wearable sensing for fine-grained activity recognition in health: skin disease analysis system using smartphone’s
healthcare,” IEEE Internet Computing, vol. 19, no. 5, camera,” Procedia Computer Science, vol. 19, pp. 1116–1120,
pp. 26–35, 2015. 2013.
[48] A. Wang, G. Chen, J. Yang, S. Zhao, and C.-Y. Chang, “A [64] B. Jeffrey, M. Bagala, A. Creighton et al., “Mobile phone
comparative study on human activity recognition using applications and their use in the self-management of type 2
inertial sensors in a smartphone,” IEEE Sensors Journal, diabetes mellitus: a qualitative study among app users and
vol. 16, no. 11, pp. 4566–4578, 2016. non-app users,” Diabetology & Metabolic Syndrome, vol. 11,
[49] H. Li and M. Trocan, “Deep learning of smartphone sensor no. 84, 2019.
data for personal health assistance,” Microelectronics Journal, [65] World Health Organization, Contact Tracing during an
vol. 88, pp. 164–172, 2019. Outbreak of Ebola Virus Disease, WHO, Geneva, Switzer-
[50] M. M. Hassan, M. Z. Uddin, A. Mohamed, and A. Almogren, land, 2014.
“A robust human activity recognition system using smart- [66] A. Grünerbl, A. Muaremi, V. Osmani et al., “Smartphone-
phone sensors and deep learning,” Future Generation based recognition of states and state changes in bipolar
Computer Systems, vol. 81, pp. 307–313, 2018. disorder patients,” IEEE Journal of Biomedical and Health
[51] A. Vaughn, P. Biocco, Y. Liu, and M. Anwar, “Activity Informatics, vol. 19, no. 1, pp. 140–148, 2015.
detection and analysis using smartphone sensors,” in Pro- [67] A. Bánhalmi, J. Borbás, M. Fidrich, V. Bilicki, Z. Gingl, and
ceedings of the 2018 IEEE International Conference on In- L. Rudas, “Analysis of a pulse rate variability measurement
formation Reuse and Integration (IRI), pp. 6–9, Salt Lake using a smartphone camera,” Journal of Healthcare Engi-
City, UT, USA, July 2018. neering, vol. 2018, Article ID 4038034, 15 pages, 2018.
[52] F. J. Ordóñez and D. Roggen, “Deep convolutional and [68] J. M. Statland, Y. Wang, R. Richesson et al., “An interactive
LSTM recurrent neural networks for multimodal wearable voice response diary for patients with non-dystrophic
activity recognition,” Sensors, vol. 16, no. 115, 2016. myotonia,” Muscle & Nerve, vol. 44, no. 1, pp. 30–35, 2011.
[53] WHO, Health: New Horizons for Health through Mobile [69] J. Barth, J. Klucken, P. Kugler et al., “Biometric and mobile
Technologies, World Health Organization, Geneva, Swit- gait analysis for early diagnosis and therapy monitoring in
zerland, 2011.
Parkinson’s disease,” in Proceedings of the 2011 Annual
[54] A. Bastawrous, “Increasing access to eye care there’s an app
International Conference of the IEEE Engineering in Medicine
for that. Peek: smartphone technology for eye health,” In-
and Biology Society, pp. 868–871, IEEE, Boston, MA, USA,
ternational Journal of Epidemiology, vol. 45, pp. 1040–1043,
September 2011.
2016.
[70] M. Matthews, S. Abdullah, G. Gay, and T. Choudhury,
[55] G. Eysenbach, S. Sawesi, J. Jones et al., “Acceptability, us-
“Tracking mental well-being: balancing rich sensing and
ability, and views on deployment of peek, a mobile phone
patient needs,” Computer, vol. 47, no. 4, pp. 36–43, 2014.
mhealth intervention for eye care in kenya: qualitative
[71] G. Eysenbach, J. Proudfoot, B. Dear et al., “Harnessing
study,” JMIR mHealth and uHealth, vol. 4, no. 2, 2016.
context sensing to develop a mobile intervention for de-
[56] E. C. Larson, M. Goel, G. Boriello, S. Heltshe, M. Rosenfeld,
and S. N. Patel, “SpiroSmart: using a microphone to measure pression,” Journal of Medical Internet Research, vol. 13, no. 3,
lung function on a mobile phone,” in Proceedings of the 2012 2011.
ACM Conference on Ubiquitous Computing—UbiComp’12, [72] S. Kumar, W. J. Nilsen, A. Abernethy et al., “Mobile health
Pittsburgh, PA, USA, September 2012. technology evaluation,” American Journal of Preventive
[57] Y.-C. Lin, Y.-H. Lai, H.-W. Chang et al., “SmartHear: a Medicine, vol. 45, no. 2, pp. 228–236, 2013.
smartphone-based remote microphone hearing assistive [73] R. Nandakumar, S. Gollakota, and N. Watson, “Contactless
system using wireless technologies,” IEEE Systems Journal, Sleep Apnea Detection on Smartphones,” in Proceedings of
vol. 12, no. 1, pp. 20–29, 2018. the 13th Annual International Conference on Mobile Systems,
[58] T. Thap, H. Chung, C. Jeong et al., “High-resolution time- Applications, and Services-MobiSys’15, pp. 45–57, Florence,
frequency spectrum-based lung function test from a Italy, May 2015.
smartphone microphone,” Sensors, vol. 16, no. 8, 2016. [74] F. Wahle, T. Kowatsch, E. Fleisch, M. Rufer, and S. Weidt,
[59] C. Catal, S. Tufekci, E. Pirmit, and G. Kocabag, “On the use of “Mobile sensing and support for people with depression: a
ensemble of classifiers for accelerometer-based activity pilot trial in the wild,” JMIR mHealth uHealth, vol. 4, no. 3,
recognition,” Applied Soft Computing, vol. 37, pp. 1018–1022, 2016.
2015. [75] Z. F. Khan, “Automated segmentation of lung parenchyma
[60] V. Natale, M. Drejak, A. Erbacci, L. Tonetti, M. Fabbri, and using colour based fuzzy C-means clustering,” Journal of
M. Martoni, “Monitoring sleep with a smartphone accel- Electrical Engineering & Technology, vol. 14, no. 5,
erometer,” Sleep and Biological Rhythms, vol. 10, no. 4, pp. 2163–2169, 2019.
pp. 287–292, 2012. [76] Z. F. Khan and A. Kannan, “Intelligent segmentation of
[61] E. Thammasat and J. Chaicharn, “A simply fall-detection medical images using fuzzy bitplane thresholding,” Mea-
algorithm using accelerometers on a smartphone,” in Pro- surement Science Review, vol. 14, no. 2, pp. 94–101, 2014.
ceedings of the 2012 5th Biomedical Engineering International [77] M. Chiang and T. Zhang, “Fog and IoT: an overview of
Conference (BMEiCON), pp. 1–4, Ubon Ratchathani, Thai- research opportunities,” IEEE Internet of Things Journal,
land, December 2012. vol. 3, no. 6, pp. 854–864, 2016.
[62] M. Seera, C. K. Loo, and C. P. Lim, “A hybrid FMM-CART [78] Y. Guo, Y. Liu, A. Oerlemans, S. Lao, S. Wu, and M. S. Lew,
model for human activity recognition,” in Proceedings of the “Deep learning for visual understanding: a review,” Neu-
2014 IEEE International Conference on Systems, Man and rocomputing, vol. 187, pp. 27–48, 2016.
14 Journal of Healthcare Engineering

[79] H. Nguyen, L.-M. Kieu, T. Wen, and C. Cai, “Deep learning [94] A. Tsanas, M. A. Little, P. E. McSharry, J. Spielman, and
methods in transportation domain: a review,” IET Intelligent L. O. Ramig, “Novel speech signal processing algorithms for
Transport Systems, vol. 12, no. 9, pp. 998–1004, 2018. high-accuracy classification of Parkinson’s disease,” IEEE
[80] D. Yang, K. Jiang, D. Zhao et al., “Intelligent and connected Transactions on Biomedical Engineering, vol. 59, no. 5,
vehicles: current status and future perspectives,” Science pp. 1264–1271, 2012.
China Technological Sciences, vol. 61, no. 10, pp. 1446–1471, [95] S. Datta, R. Barua, and J. Das, Application of Artificial In-
2018. telligence in Modern Healthcare System, In techOpen, Lon-
[81] S. Alshahrani and E. Kapetanios, “Are deep learning ap- don, UK, 2019.
proaches suitable for natural language processing?,” in Nat- [96] N. Larburu, A. Artetxe, V. Escolar, A. Lozano, and
ural Language Processing and Information Systems, NLDB J. Kerexeta, “Artificial intelligence to prevent mobile heart
2016, E. Métais, F. Meziane, M. Saraee,V. Sugumaran, and S. failure patients decompensation in real time: monitoring-
Vadera, Eds., Springer, Cham, Switzerland, Lecture Notes in based predictive model,” Mobile Information Systems,
Computer Science, vol 9612, 2016. vol. 2018, Article ID 1546210, 11 pages, 2018.
[82] T. H. Kim, G. S. Tomar, W. I. Grosky, T. H. Kim, [97] M. N. Burns, M. Begale, J. Duffecy et al., “Harnessing context
S. Mohammed, and S. K. Saha, “Emerging approach of sensing to develop a mobile intervention for depression,”
natural language processing in opinion mining: a review,” in Journal of Medical Internet Research, vol. 13, no. 3, 2011.
Ubiquitous Computing and Multimedia Applications, [98] M. S. Hawley, S. P. Cunningham, P. D. Green et al., “A voice-
pp. 121–128, Springer, Berlin, Germany, 2010. input voice-output communication aid for people with se-
[83] M. M. Churpek, T. C. Yuen, C. Winslow, D. O. Meltzer, vere speech impairment,” IEEE Transactions on Neural
D. P. Kattan, and D. P. Edelson, “Multicenter comparison of Systems and Rehabilitation Engineering, vol. 21, no. 1,
machine learning methods and conventional regression for pp. 23–31, 2013.
predicting clinical deterioration on the wards,” Critical Care [99] C. M. Martin, C. Vogel, D. Grady et al., “Implementation of
Medicine, vol. 44, no. 2, pp. 368–374, 2016. complex adaptive chronic care: the Patient Journey Record
[84] K. Seetharam, S. Shrestha, and P. P. Sengupta, “Artificial system (PaJR),” Journal of Evaluation in Clinical Practice,
intelligence in cardiovascular medicine,” Current Treatment vol. 18, no. 6, pp. 1226–1234, 2012.
Options in Cardiovascular Medicine, vol. 21, no. 6, 2019. [100] L. G. Morrison, C. Hargood, V. Pejovic et al., “The effect of
[85] W. Bai, M. Sinclair, G. Tarroni et al., “Automated cardio- timing and frequency of push notifications on usage of a
smartphone-based stress management intervention: an ex-
vascular magnetic resonance image analysis with fully
ploratory trial,” PLoS One, vol. 12, no. 1, Article ID e0169162,
convolutional networks,” Journal of Cardiovascular Magnetic
2017.
Resonance, vol. 20, no. 65, 2018.
[101] M. Ortiz-Catalan, R. Brånemark, and B. Håkansson, “Bio-
[86] S. J. Al’Aref, K. Anchouche, G. Singh et al., “Clinical ap-
PatRec: a modular research platform for the control of ar-
plications of machine learning in cardiovascular disease and
tificial limbs based on pattern recognition algorithms,”
its relevance to cardiac imaging,” European Heart Journal,
Source Code for Biology and Medicine, vol. 8, no. 1, 2013.
vol. 40, no. 24, pp. 1975–1986, 2019.
[102] Apache Mahout, http://mahout.apache.org/.
[87] K. Seetharam, N. Kagiyama, and P. P. Sengupta, “Application
[103] Skytree, http://www.skytree.net/products/.
of mobile health, telemedicine and artificial intelligence to
[104] Karmasphere (F. I. C. O), http://www.fico.com/en/
echocardiography,” Echo Research and Practice, vol. 6, no. 2, communications/patientadherence#overview.
pp. 41–52, 2019. [105] BigML, http://bigml.com/.
[88] A. W. Forsyth, R. Barzilay, K. S. Hughes et al., “Machine [106] P. P. Sengupta, “Cognitive machine-learning algorithm for
learning methods to extract documentation of breast cancer cardiac imaging: a pilot study for differentiating constrictive
symptoms from electronic health records,” Journal of Pain pericarditis from restrictive cardiomyopathy,” Circulation:
and Symptom Management, vol. 55, no. 6, pp. 1492–1499, Cardiovascular Imaging, vol. 9, no. 6, 2016.
2018. [107] S. Narula, K. Shameer, A. M. Salem Omar, P. P. Dudley, and
[89] S. F. Weng, J. Reps, J. Kai, J. M. Garibaldi, and N. Qureshi, P. P. Sengupta, “Machine-learning algorithms to automate
“Can machine-learning improve cardiovascular risk pre- morphological and functional assessments in 2D echocar-
diction using routine clinical data?” PLoS One, vol. 12, no. 4, diography,” Journal of the American College of Cardiology,
Article ID e0174944, 2017. vol. 68, no. 21, pp. 2287–2295, 2016.
[90] I. Kavakiotis, O. Tsave, A. Salifoglou, N. Maglaveras, [108] M. C. Lancaster, A. M. Salem Omar, S. Narula, H. Kulkarni,
I. Vlahavas, and I. Chouvarda, “Machine learning and data J. Narula, and P. P. Sengupta, “Phenotypic clustering of left
mining methods in diabetes research,” Computational and ventricular diastolic function parameters: patterns and
Structural Biotechnology Journal, vol. 15, pp. 104–116, 2017. prognostic relevance,” JACC: Cardiovascular Imaging,
[91] J. Kelly, P. Gooding, D. Pratt, J. Ainsworth, M. Welford, and vol. 12, no. 7, p. 1149, 2019.
N. Tarrier, “Intelligent real-time therapy: harnessing the [109] M. Fiszman, W. W. Chapman, D. Aronsky, R. S. Evans, and
power of machine learning to optimise the delivery of P. J. Haug, “Automatic detection of acute bacterial pneu-
momentary cognitive-behavioural interventions,” Journal of monia from chest X-ray reports,” Journal of the American
Mental Health, vol. 21, no. 4, pp. 404–414, 2012. Medical Informatics Association, vol. 7, no. 6, pp. 593–604,
[92] J. A. Cruz and D. S. Wishart, “Applications of machine 2000.
learning in cancer prediction and prognosis,” Cancer In- [110] M. O. Gökalp, K. Kayabay, M. A. Akyol, A. Koçyiğit, and
formatics, vol. 2, pp. 59–77, 2006. P. E. Eren, “Big data in mHealth,” Current and Emerging
[93] C. Salvatore, A. Cerasa, P. Battista et al., “Magnetic resonance mHealth Technologies, E. Sezgin,S. Yildirim, S. Özkan-ildirim,
imaging biomarkers for the early diagnosis of Alzheimer’s and E. Sumuer, Eds., Springer, Cham, Switzerland, 2018.
disease: a machine learning approach,” Frontiers in Neuro- [111] F. A. Lateef, “Big data: applications in healthcare and medical
science, vol. 9, no. 307, 2015. education,” Education in Medicine Journal, vol. 8, no. 1, 2016.
Journal of Healthcare Engineering 15

[112] W. Raghupathi and V. Raghupathi, “Big data analytics in

healthcare: promise and potential,” Health Information
Science and Systems, vol. 2, no. 3, 2014.
[113] N. V. Chawla and D. A. Davis, “Bringing big data to per-
sonalized healthcare: a patient-centered framework,” Journal
of General Internal Medicine, vol. 28, no. S3, pp. 660–665,
2013.
[114] T. W. Kim, K. H. Park, S. H. Yi, and H. C. Kim, “A big data
framework for u-healthcare systems utilizing vital signs,” in
Proceedings of the International Symposium on Computer,
Consumer and Control, pp. 494–497, Taichung, Taiwan, June
2014.
[115] M. I. Pramanik, R. Y. K. Lau, H. Demirkan, and
M. A. K. Azad, “Smart health: big data enabled health
paradigm within smart cities,” Expert Systems with Appli-
cations, vol. 87, pp. 370–383, 2017.
[116] F. Rahman, M. Z. A. Bhuiyan, and S. I. Ahamed, “A privacy
preserving framework for RFID based healthcare systems,”
Future Generation Computer Systems, vol. 72, pp. 339–352,
2017.
[117] S. Sicari, A. Rizzardi, L. A. Grieco, G. Piro, and A. Coen-
Porisini, “A policy enforcement framework for Internet of
Things applications in the smart health,” Smart Health,
vol. 3-4, no. 4, pp. 39–74, 2017.
[118] L. Xavier, R. Thirunavukarasu, and R. Thirunavukarasu, “A
distributed tree-based ensemble learning approach for effi-
cient structure prediction of protein,” International Journal
of Intelligent Engineering and Systems, vol. 10, no. 3,
pp. 226–234, 2017.
[119] M. D. C. Legaz-Garca, C. Martinez-Costa, M. Menarguez-
Tortosa, and J. T. Fernandez-Breis, “A semantic web based
framework for the interoperability and exploitation of
clinical models and EHR data,” Knowledge-Based Systems,
vol. 105, pp. 175–189, 2016.