Content uploaded by Parag Chatterjee
Author content
All content in this area was uploaded by Parag Chatterjee on Nov 26, 2015
Content may be subject to copyright.
Internet of Things for a Smart and Ubiquitous eHealth System
Parag Chatterjee
Department of Computer Science
St. Xavier‟s College (Autonomous), Kolkata
University of Calcutta
Kolkata, India
E-mail: parag2700@gmail.com
Ricardo L. Armentano
Translational Engineering Department
School of Engineering and Exact and Natural Sciences
Favaloro University
Buenos Aires, Argentina
E-mail: armen@ieee.org
Abstract — Connected data has always been considered as a
primary source to knowledge. Internet of Things uses the
virtue of connecting this data from different entities and
creates a pool of knowledge for providing smart services to
users, based on rigorous analysis and processing over the
knowledge. The communication in this context scales to not
only between machine to machine but also between a large
number of heterogeneous entities and persons. This genius
technology of Internet of Things holds paramount importance
and application in healthcare technologies. Considering health
technologies, a large number of devices generate huge amount
of data related to a patient. Assimilating the data from
heterogeneous sources and using it to generate intelligence is
one of the primary tasks in a smart environment. In the
context of eHealth, Internet of Things is of immense
importance since connected data about patient would facilitate
treatment with more efficiency and comprehensive knowledge.
Virtually storing the patient data and making it ubiquitously
accessible to concerned healthcare personnel would be the first
step toward mutual knowledge sharing. Another important
aspect of using this connected data is the design of an
intelligent clinical decision support system which would assist
the doctors in every possible way during the treatment phase.
A model has been proposed with an inclusive approach of
Internet of Things in eHealth scenario for a smart medical
environment and providing ubiquitous services at its best.
Several issues pertaining to the system has also been discussed
accordingly. Nevertheless, the enormous spread of Internet of
Things for efficient and intelligent healthcare services holds
quite inevitable. Rather it adds to the foundation notion of
ubiquitous services by making available to everyone and
everywhere. The new age eHealth facilities are expected to
enable end-to-end monitoring systems even at remote
scenarios, helping medical services reach the unreached.
Keywords – cloud; eHealth; healthcare; internet of things;
smart; ubiquitous
I. INTRODUCTION
With the advancement in digital technologies, the
number of digital entities has seen a sharp rise in the recent
years. Since it has been a common goal to use data
efficiently and acquire knowledge, innovative methods were
employed to make the most of it. In this context, the notion
of connected data came into being for the generation of
enhanced knowledge. Using this knowledge, the machines
and devices can be programmed for working with
intelligence for better and customized output. Making use of
the connected data obtained from homogeneous sources has
been a common sector in communication system between
computers. However the notion to make use of connected
data from heterogeneous sources and obtain intelligence
from it holds the foundation of Internet of Things (Fig. 1).
Figure 1. Generation of intelligence from data.
A trivial goal of Internet of Things (IoT) is to achieve the
apex of the pyramid (Fig. 1) and to use the intelligence to
provide customized and user-specific services. Basically,
Internet of Things (IoT) actually harnesses the power of
connected data and applies it for multifarious aspects of
smart living.
Typically, IoT is expected to offer advanced connectivity of
devices, systems, and services that goes beyond machine-to-
machine communications (M2M) and covers a variety of
protocols, domains, and applications [1]. The interconnection
of embedded devices working intelligently has been
cultivated for employing automation in every sectors ranging
from home control to smart cities. With the evolving number
of devices, the number of connected devices is expected to
touch 50 billion by 2020 [2]. And this enormous power of
connected data holds IoT as a high potential technology for
the upcoming days. The predictions of Cisco IBSG also
pointed out to the enormous increment of the number of
connected devices per person, which could be 6.58 by 2020
[3].
Figure 2. Evolutionary Outline of Humans vs Devices.
(Source: Cisco IBSG, 2011)
However, the connected devices could be only used to its
best with the help of a mutual sharing platform. Also in a
typical IoT, not only the Internet-enabled devices like
smartphone, PC are under consideration. For an efficient
service, heterogeneous objects like car, refrigerator and
lights are also taken into consideration and fastened to the
same connectivity platform. Real-time communication
between the heterogeneous connected devices enables
intelligent user-specific services.
As a part of reaching every sector with the connectivity
of IoT, eHealth is an important area of concern. Primarily
eHealth is a healthcare practice supported by electronic
process and communication. But with the extensiveness of
communication, mutual sharing between healthcare entities
has become inevitable. Inducing IoT technologies in
healthcare, the standard procedure of monitoring,
investigation and treatment would be ushered in with a
digital revolution of connectivity. One of the focal goals in
this context is to embed the computing power inside the
devices and communication channel so that the entire system
remains subtle and ubiquitous. In this connection, an eHealth
model has been illustrated, considering different aspects of
the system.
II. FOCAL AREAS OF E-HEALTH
To induce IoT as an embedded technology in the existing
healthcare applications, the focal services need to be
addressed first.
TABLE I. FOCAL HEALTHCARE APPLICATIONS
Service
Description
Health Records
A lot of records linked to a patient are generated
during a treatment phase. Also for a chronic
treatment, a lot of patient‟s medical history needs to
Service
Description
be taken into account for efficient treatment.
Efficient management of all the patient data is an
important task to perform.
Diagnosis
Detailed treatment involves a large number of
diagnoses very often. These diagnosis process is
often distributed which needs a comprehensive
assimilation for finalized reports. This entire process
is often tedious.
Monitoring
The entire monitoring process involves huge amount
of real-time data. This data is processed and
analyzed (mostly manually).
Post-medication
phase
Even after the medication phase is over, the
rehabilitation phase also involves recording patient‟s
data which is also very vital in the context of
ensuring complete cure.
The chief reason why the induction of IoT in healthcare
systems has been taking over is quite clear. Most healthcare
applications and processes work distributed and remotely.
Especially the areas where active manual intervention is
inevitable (like medication, prescription) work as remote
entities and data sharing has been quite tedious. IoT comes in
as the connecting platform for all the respective entities
involved in a typical healthcare system. Moreover it supplies
the subtle and embedded power of computing which extracts
the data from the environment and exchanges it mutually for
a ubiquitous information system, heading toward a
ubiquitous intelligent system. Transforming to a healthcare
system based on IoT, the resultant system performs much
more to a typical clinical information system. The main
objective of IoT is not to cater a mere information system but
to provide an automated decision support system. However,
the level of autonomy is extremely variable from case to
case. This gives emphasis on the method of passive
monitoring and medication. In many cases, even after the
end of an active treatment phase, this passive treatment is an
important phase which is one of the focal areas of interest in
IoT based healthcare systems.
III. PROPOSED E-HEALTH MODEL BASED ON IOT
To promote a ubiquitous and smart environment of
healthcare services, an IoT based model of eHealth has been
proposed. First, the objective of such system needs to be
addressed. Accordingly the connecting entities along with
their method of interaction would be discussed.
A. Patient Records
Maintaining the patient data electronically has been quite
common in all major healthcare facilities. However the
increasing number of distributed healthcare entities results in
generating redundant data. Also the historic data of a patient
couldn‟t be used efficiently if the data belongs to another
entity. To make use of the power of connected data, the
entities dealing with patient data needs to be linked up via a
common and ubiquitous platform. IoT would serve as this
connecting link between the entities dealing with patient
entities. Primarily, the mutual exchange of patient records
within remote entities would simplify the record maintaining
purpose. Likewise, keeping track of the medical history of a
patient would also become easier. Mere sharing of data
requires a distributed database but using the IoT approach, it
works little different. Getting deeper into the system, the
patient is the center of all data in the environment. Hence
whatever interactions take place between the sharing nodes
would use patient‟s ID as the unique key. Accordingly a
virtual storage (cloud) space for every new patient is
proposed. The registration to this storage space could be
done on the first encounter with the health services.
Alternatively, every persons registered with a National
Unique Identity (UID) could be allocated this storage space
specially meant for health records. However, the storage
mechanism needs to be efficient for systematic access when
required. Accordingly the data system needs to be structured
properly in an efficient schema so that even if the data is
entered by different remote entities, storing the data uses a
common form for easier and systematic access.
Figure 3. Relationship schema for storing patient related records.
Holding patient_id as the unique key, separate tables has
been used for storing records related to diagnosis,
investigation, medication and rehabilitation of a patient (Fig.
3). For every set of records at a single instance a separate ID
would be created at that table itself. Another important table
for family details has also been put up. Here a respective
patient_id is linked with other unique patient_id of their
family members with the relationship details. It could be
excellent is treatment where knowing family traits is of
paramount importance. Also specific family traits would
generate precautionary alerts which could help many
diseases get detected well beforehand.
The mechanism to operate this system would also
be different compared to ordinary e-record management
system. The primary difference is that unlike ordinary
electronic record management system, in this system any
healthcare unit or even a private doctor could upload the
medical records of a patient in the proper format. Since the
key to this personal storage space of the patient is the UID, a
login system needs to be made functional ubiquitously.
Using this portal any authorized person could access the
patient data. However, to restrict access, the authentication
could be based on some biometric information and password.
Also the diagnostic information could be classified into
hierarchical and historic levels. Like for a cardiac treatment,
the medical team would have primary access to the cardiac
reports only. However, secondary and tertiary access could
be granted based on the nature of request.
B. Clinical Decision Support System
Providing a decision support system for the physicians
based on connected knowledge would be of immense help.
The system consists primarily of three key parts – the
knowledge base, an inference engine and a communication
mechanism. Basically in an IoT system for clinical decision
support system, the connected data of patient‟s history is
primarily taken into concern. This includes specific allergies
and sensitivities. Also the decision system is trained with
conditional statements which would take the historic data
and current data, match with the inbuilt logic and would pose
specific alerts and decisions as an advice to the medical
personnel. For example, if a patient has past history of
disease „A‟, drug „B‟ could be potentially risky. In that case
drug „C‟ needs to be applied instead of drug „B‟. The
decision system would work like –
if (apply_drug(B) then
check_history(A);
if (check_history(A) == „true‟) then
apply_drug(B) = „false‟;
apply_drug(C);
end if
end if
The effectiveness of this IoT based decision support system
is that, the patient need not to exclusively keep track of all
prior medical history. Also a new physician handling the
case needs not to search the entire medical history before
applying a drug. Rather the trained decision support system
performs the search process and generates alert if a drug
creates potential risk in connection to prior medical issues.
However the training phase is extremely vital for a good
performance decision support system. Primarily the system
is divided into knowledge based and non-knowledge based
approaches [5]. IoT acts as the missing link in this context.
The knowledge is obtained from the input and logic provided
to the system beforehand. Also several conditional logics are
also provided to the system as pre-inputs. The intelligence
works on the patterns obtained from the recorded data linked
to the patient, dealing with the machine-learning
mechanisms [6]. Considering the knowledge and
intelligence, the system would provide the best support to the
doctor for prescribing the best thing for the patient.
C. Real-time and Remote Monitoring
The IoT based healthcare system could also work for
passive medication or lifestyle. But to record lifestyle data,
real-time recording is needed. However, real-time recording
involves generation of huge data. Lifestyle data could be
obtained through embedded sensors. Sensors could be
embedded in devices having bodily contact like smart
watches and wrist bands. However, the data obtained in this
way contains only non-invasive data. To address the problem
of voluminous nature of real-time data, a method has been
proposed. The sensors need to be provided with minimal
processing capacity. On collection of data, at specific time
intervals, running average is calculated. Periodic peak values
which stand far away from the average are recorded. While
storing the data to the virtual storage, only two values would
be stored – the average value and the peak value
(that do not comply well with the average). Metabolism
rates, calorific values, pulse, blood glucose levels could be
some of the recordable data from this non-invasive real-time
monitoring process.
Figure 4. Scheme for real-time monitoring and reporting .
Considering a typical monitoring system of heart pulse (Fig.
4), the running average taken at every 12 seconds is 70.416.
The normal pulse being pre-programmed at 72 along with
the normal range being given, the average value remains
within normal range. However at a particular instance, the
value drops to 65 which is beyond the normal range and
needs to be reported separately. So during storing the
summarized value of this report, only two attributes
(average, peak) are taken into concern. Hence at every
twelve seconds the values are being refreshed, the first
dataset being reported as (70.416, 65.0). Definite alerts could
also be generated if the dataset goes beyond normal range,
approaching to potential emergency. It is good to take the
home-recorded data of a patient than the recorded data at
doctor‟s place because home-recorded data is more likely to
be realistic since it corresponds more to a normal condition
[4]. Also abnormal values addressed beforehand helps tackle
many critical health issues in an early and better way.
D. Remote treatment
Many health issues get addressed at a much later stage
due to lack of immediate connection with the healthcare
personnel. This issue is severe in case of rural and remote
areas where healthcare facilities are not always at the
fingertips. Due to this issue, often it is seen that patients are
more inclined to follow previous medication (medicines
prescribed earlier for similar health problems) rather than
consulting a physician afresh. In this context, remote
treatment counts significant because it helps connecting to
healthcare facilities in a better way. Telemedicine and
Cybermedicine have been quite popular terms these days but
using IoT based smart platform would make the things
simplified and efficient together. The foundation of this IoT
based platform of remote treatment is based not only on
symptoms as reported by the patient but also on the health
data remotely collected from the embedded sensors at the
patient‟s end and transmitted through the IoT platform to the
concerned physician.
Figure 5. Schematic Diagram of IoT based Remote Treatment Model.
Basically the entire process of remote treatment works as an
event management procedure (Fig. 5) which is described
briefly as follows. As the patient generates the request for
remote care, the system asks for the symptoms. Once the
symptoms have been recorded, the request is transmitted to
the healthcare personnel, the request containing the
patient_id and primary symptoms. Analyzing the request, the
doctor could ask for more symptoms if required. Then she
sends query to the patient‟s system for the recent (or
previous, if needed too) health data, sensed through
embedded sensors (smart band etc.). The decision support
system at the doctor‟s end assists in choosing the right
medicine for the patient, considering previous medical issues
and special sensitivity towards a drug. Analyzing the data
obtained, the doctor either generates an e-prescription, or
expresses the inability to treat remotely (after giving primary
advice). This could be an efficient method for reaching the
unreached with the best possible medical assistance,
remotely.
IV. DISCUSSIONS ON THE MODEL
The focal concept of the IoT based eHealth system is
using the ubiquitous network to communicate with the
concerned entities and using the knowledge from the data to
take intelligent actions. However, there have been several
issues involved. First, availability of a live data connection
would have been essential for maintaining a constant
connection. But it is never recommended to maintain a live
connection between all the entities, nor is it possible also.
Alternatively, periodic updates of the system could be
broadcast to all the concerned entities to refresh the status
pages. Embedded chips providing data connection would be
immensely helpful in this context. Second is the security
structure of the system. Since the system deals with patient‟s
data, utmost care is essential. However the data becomes
more vulnerable when it moves across several networks,
often unspecified beforehand. In this case, maintaining the
data locally is recommended. The devices and entities
seeking remote data need to use a query service with a
request which would be granted with the exact (or
summarized) value, as required. This minimizes the chance
of bulk data leakage and putting the security at stake. Also
specific gateways need to be employed at strategic points to
ensure security. Another important issue is the fate of the
patient‟s data stored in the personal virtual space. This area
is vital because it would contain all the medical history
related to a person. To ensure non-fabrication of the data, it
is recommended to put up any modification facility. Only
authorized entities would be able to write new data (medical
records) and access previous data, based on query services
only.
V. CONCLUSION
Inclusion of technologies in health sector has been quite
enthusiastic in the recent times. Just with the advent of IoT,
inclusion of this technology for eHealth sounds significant as
a part of the ubiquitous revolution. Moreover the proposed
system of eHealth based on IoT would not only provide a
smarter approach toward health services but also makes the
decision making process intelligent. On a whole this system
could address several health issues as a mass. Since the
foundation of the proposed eHealth model is based on
Internet, it would be easier to transform the outputs to second
screen and mobile devices. Accessibility to remote
monitoring would receive a thrust in this way. Scalability is
also an important advantage of this system. Performing big
data analytics on the connected data obtained from the
patients‟ virtual storage of medical records, health issues of
the mass could be identified at its best and could be
addressed efficiently. So the benefits of this connected data
would not only be personal but also significant in the larger
aspect. It would also help in exchanging the medical issues
for worldwide research which would help addressing the
problems and providing better solutions. Despite several
issues of connectivity and security, IoT based health services
truly hold a promising future for extending the services on a
larger vision.
REFERENCES
[1] J. Höller, V. Tsiatsis, C. Mulligan, S. Karnouskos, S. Avesand, D.
Boyle, “From Machine-to-Machine to the Internet of Things:
Introduction to a New Age of Intelligence,” Elsevier, 2014, ISBN
978-0-12-407684-6.
[2] Philip N. Howard (8 June 2015). “How big is the Internet of Things
and how big will it get?,” The Brookings Institution. Retrieved 26
June 2015.
[3] Dave Evans, “The Internet of Things – How the Next Evolution of
the Internet is Changing Everything,” White Paper-Cisco Internet
Business Solutions Group, April 2011.
[4] David Lake, Rodolfo Milito, Monique Morrow and Rajesh
Vargheese, “Internet of Things: Architectural Framework for eHealth
Security”, Journal of ICT, Vol. 3 and 4, pp. 301-308, doi:
10.13052/jicts2245-800X.133
[5] Berner, Eta S., ed, “Clinical Decision Support Systems,” New York,
NY: Springer, 2007.
[6] "Tanveer Syeda-Mahmood plenary talk: The Role of Machine
Learning in Clinical Decision Support", SPIE Newsroom, March
2015.