IEEE Network • November/December 2019
30
0890-8044/19/$25.00 © 2019 IEEE
AbstrAct
A remarkable increase of personalized health-
care services has been powered by the broad
implementation of the networks and their ram-
ifications. The core value of using personalized
healthcare systems is to create customized med-
ical service offerings and establish interconnec-
tions between patients and physicians. However,
the treatment delivery for individual patients is
encountering a challenge in effectively creating
plans in which medical services can be retrieved
from multiple sources at various costs. This article
focuses on the problem of minimizing total cost
of service retrieval and proposes an approach that
uses intelligent agents to dynamically make ser-
vice retrieval strategies. The approach is called
the Smart Treatment for Personalized Healthcare
(STPH) model, which is designed to produce an
optimal solution to minimizing total cost on the
intelligent agent. The performance of the pro-
posed approach has been validated by experi-
mental evaluations.
IntroductIon
The rapid expansion of deploying network-based
technologies has created a variety of new services
throughout people’s lives, from tele-health to
mobile apps. This expansion has brought a bunch
of new technical terms to the eyeshot of our
society, for example, the Internet of Things (IoT)
and cloud computing. A key factor of this trend
addresses the optimization of computing resource
usage, considering the scenario within a connect-
ed environment [1]. As one of the popular novel
applications in the networking domain, person-
alized healthcare (PHC) is powered up from dis-
tinct perspectives, such as distributed treatment
support, real-time patient monitoring, and drug
delivery [2]. The extent of the medical service
generally takes advantage of the advanced com-
putation capability provided by centralized com-
puting as well as distributed sensor deployment.
A characteristic of the network is that it also
enables enhanced connection between patients
and medical service institutes. An issue is gradually
emerging as a quickly growing number of patients
results in the expanded computation workload,
which is a major aspect for restricting the perfor-
mance of customized healthcare. Considering the
service model of customized healthcare, normally,
a patient shall have an individual medical treat-
ment plan that is associated with his/her health
status/ condition. Real-time service availability is a
high-quality requirement, since the system needs
to deal with consistent dynamic input (medical
situations). Centralized computing, for example,
a cloud data center, encounters an observable
restriction when the user group is oversized.
Multiple factors are negatively impacting its per-
formance, such as complexity of the treatment
planning and network traffic jams. It seems to be
a contradiction to concurrently achieve high per-
formance and service customization on PHC.
Our study has explored the solution to the
issue above. We emphasize the importance of the
service availability and intend to minimize the “dis-
tance” between patients and medical treatment.
The distance is a metaphor for the gap restrict-
ing patients from reaching medical services, for
example, latency, equipment availability, opera-
tion room scheduling, or medical service quality.
In this article, we present a problem that con-
cerns multiple treatment/therapy providers who
offer various service qualities at different costs.
Each patient is considered to be a service target
unit who may have a number of medical options
(different treatments/therapies or different med-
ical service institutes). We name the proposed
approach as a Smart Treatment for Personalized
Healthcare (STPH) model. Figure 1 illustrates the
architecture of the proposed model.
According to the illustration of the figure,
there are two major participant groups from the
user side, who are “patients” and “physicians.”
Among these two roles, on one hand, patients are
treatment receivers, and multiple sensors maybe
deployed for the purpose of data collection. Our
approach introduces an intelligent agent (IA) part,
which assists in the construction of customized
treatment/medical plans and forwards data pro-
cessing tasks if applicable. Specifically, the IA sec-
tor is a multi-role who records medical data, filters
data, allocates/forwards tasks, and synchronizes
data with edge/cloud servers. Two options for
task forwarding are set, edge servers and remote
cloud servers, the use of which depends on the
complexity of the task. A cloud server generally
deals with heavy processing jobs (e.g., high com-
putation complexity), while an edge option mainly
is aimed at simpler tasks.
On the other hand, physicians play a major
role in improving the service quality via sending
feedback to medical treatments/therapy pro-
viders. In the figure, the service provider side
deploys two subsystems, including the medical
Toward Smart Treatment Management for Personalized Healthcare
Keke Gai, Zhihui Lu, Meikang Qiu, and Liehuang Zhu
ENABLING NETWORKED SERVICES AND TECHNOLOGIES FOR
CONNECTED HEALTHCARE
Digital Object Identifier:
10.1109/MNET.001.1900075
Keke Gai and Liehuang Zhu are with Beijing Institute of Technology; Zhihui Lu (corresponding author) is with Fudan University and Engineering Research Center
of Cyber Security Auditing and Monitoring, Ministry of Education, China; Meikang Qiu (co-corresponding author) is with Columbia University.