A health care monitoring system that uses ontology

agents

Stella C. Christopoulou, Theodore Kotsilieris Ioannis Anagnostopoulos

Department of Business Administration Department of Computer Science and Biomedical
TEI Peloponnese Informatics
Kalamata, Greece University of Thessaly
[email protected], [email protected] Lamia, Greece
[email protected]

Abstract— The healthcare domain is mission critical and vast tems. The aim of the proposed system is to suggest an agent-
research efforts keep being funded in order to improve life based solution to overcome the automation deficiencies of
quality of people. Health management consists of specific medical data monitoring. Towards this objective we study the
cooperation intensive activities, while medical knowledge and applicability and usefulness of the mobile agent technology in
systems are spatially and functionally distributed. Also, medical the healthcare domain. Data management, decision-making and
data flows are inefficient for dynamic data delivery to multi-source information retrieval are some indicative
stakeholders while the manifold of medical care derives from applications in the field.
their diversity and autonomy. In such an environment we
propose the implementation of vhMentor based on a solid Agent technology proves to be ideal for developing
ontological schema. The aim of the proposed system is to suggest healthcare applications due to its distributed operation over
an agent-based solution to overcome the automation deficiencies systems and data sources of high heterogeneity. Agents are
of medical data monitoring. Towards this objective we study the able to perform their tasks by acting pro-actively in order to
applicability and usefulness of the mobile agent technology in the assist individuals to overcome burdens posed during accessing
healthcare domain. Furthermore, we implemented the medical data. Thus, the proposed system fills the gap between
encapsulation of remote sensing and medical devices signals in an healthcare systems and community care provision and at the
ontology-supported healthcare information system. The resulting same time it allows for the introduction of next generation
system encourages the future implementation of next generation
distributed services in healthcare domain.
services (e.g. decision making and reasoning) through
autonomous - intelligent agents and reasoning engines. Agent technology has been extensively used in healthcare
applications development due to it is distributed operation
Keywords— Ubiquitous & Pervasive computing, software efficacy over systems and data sources of high heterogeneity
agents, health care monitoring, medical knowledge, medical degree. It has been extensively used to overcome
information systems interoperability weaknesses and inefficiencies that are typically
found in legacy medical information systems [1], [2], [3]. The
mobile collaboration features of agents have been proposed in
I. INTRODUCTION – RELATED WORK [4] towards developing a u-healthcare system. VIMS [5] is a
The healthcare domain is mission critical and vast research mobile agent approach that achieves the secure integration of
efforts keep being funded in order to improve life quality of distributed health information over heterogeneous medical
people. Health management consists of specific activities that records systems. Moreover, multi-mobile agent systems have
been used in the implementation of a distributed system for
requires the collaboration of physicians, nurses, managers,
indoor healthcare monitoring of elderly people [6]. All the
patients where each of them holds different expertise and above mentioned research efforts exploit the intrinsic features
responsibilities. Also, medical knowledge is spatially and of mobile agents for mobility, persistence, flexibility and
functionally distributed. Furthermore, medical data flows are adaptability to dynamic environments.
inefficient for dynamic data delivery to stakeholders while the
manifold of medical care derives from their diversity and The rest of the paper is organized as follows. Section 2
autonomy. introduces the ontology enabled JADE (Java Agent
In such an environment we propose the implementation of DEvelopment Framework – jade.tilab.com) agent framework.
vhMentor that is a solid ontological schema, over a Foundation Section 3 outlines the system’s architecture and functional
for Intelligent Physical Agents (FIPA – www.fipa.org) specifications, while it also describes a prototype experimental
compliant agent framework. Agent technology is endorsed for environment. In section 4 discusses the main challenges as well
the development of healthcare applications because of its as the future research directions. Finally, section 6 concludes
distributed and seamless operation over diverse operation sys- our work.

978-1-5090-5246-2/16/$31.00 ©2016 IEEE

 II. THE ONTOLOGY SUPPORTED AGENT PLATFORM Every Main-Container hosts two special purpose agents,
JADE is the most popular open-source agent platform used namely Agent Management Service (AMS) and Directory
not only for research but commercial purposes as well. It is Facilitator (DF). AMS is a special agent that provides naming
fully implemented in JAVA and provides several APIs for services to the platform, while it enforces intra (within the
distributed applications development and interconnection with same platform) and inter (with remote platforms) authority
legacy, centralized information systems through agent services. DF is also named Yellow Pages agent and is aware
communication with FIPA Agent Communication Language about services provided by other agents within the platform.
(ACL) message passing.
Several studies on the field of agent systems have
employed JADE as the software platform of choice. The
research efforts have described systems that span from big data
analysis [7], smart home automation [8] and mobile-cloud
computing [9] to crisis management [10], personalized
recommendation systems [11] and ontology based application
in financial sector [12].
However, it has to be mentioned that remarkable research
has been conducted on applying JADE framework in the field
of health care. A semantic-driven system has been described
for decision support in epidemic management has been
described in [13]. The K4Care is fully developed in JADE to
support knowledge and data dissemination [14]. JADE mobile
agents have also been proposed for drug safety surveillance
[15] and decision support for optimal drug prescription [16].
In our work we adopted JADE as the mobile agent
development platform due to its dominating role in mobile
agent computing and support for emerging operating systems
including Android [17].
A. JADE characteristics
JADE is a platform that supports the development of agent
applications that conforms to the FIPA specification. A typical
Fig. 1. A typical JADE platform deployment
JADE based system consists of the runtime environment, the
agent development APIs and the graphical agent management B. BeanGenerator
tools. Every instance of the JADE runtime environment is
called a Container that act as agent hosts, which is the place The JADE ontology support mechanism defines the process of
where one or more agents may be created and exist. In order to converting information from FIPA compatible ACLMessage
create a new platform, a Main-Container that plays the role of format to Java ontology objects. The ontology support
the coordinator, must be always active so that every other mechanism of the JADE platform and the Protégé ontology
typical Container may register with it. The first Container development environment are bound through an appropriate
activated in the platform is assigned with the role of Main- middleware named BeanGenerator allows the transformation of
Container, while all others become its members. Typical a JADE compliant ontology that is developed in Protégé into
containers “are told” how to register with Main-Container by JAVA class instances.
being provided the relevant host ID and port number. In the III. SYSTEM ARCHITECTURE AND FUNCTIONALITY
event that a new Main-Container is initiated, it is considered to
be a different platform that may accept new containers through A. The proposed system architecture
registration. The following figure depicts the employed JADE The purpose of our system is to properly blend ontological
Platform architecture. representation of data and activities with the agent
The system depicted in Figure 1, includes two JADE programming framework and technological services (wireless
platforms, namely A and B. Platform A comprises of a Main- sensors & networks, knowledge management etc.) towards an
Container (1), with two simple Containers (A and B) registered active health data monitoring system through a well-defined
on it. Platform B is composed of two containers; a main (Main knowledge ontology base in order to gather and disseminate
Container 2) and a normal one (Container C). Within the vital clinical data (i.e. Data Management Tier - DMT).
abovementioned environment, JADE agents are uniquely Figure 2 depicts the conceptual model of our system.
defined by their name, and are able to communicate regardless Patients, family members and healthcare professionals are the
of their actual location or registered main container. main actors identified. Accordingly, each functional entity
Specifically, agents 1A and 2A interact as members of the include the JADE agent management platform along with static
same container, 5A and 6A as members of the same platform and/or mobile agents that are dedicated to perform health data
and agent 4A with agent 7A as members of different platforms. management through ontology supported functionality.

42
 The raw medical data that are sensed from patients are actions or alarm generations. Thus, care providers are able to
gathered from JADE agents and DMT acts as an intermediate review the status of the patients through the software platform.
between the network and our proposed services platform (i.e. The objective of IPT is to introduce into daily medical practice
Information Processing Tier - IPT). characteristics such as dynamic interaction, continuous data
update and cooperative management. Also, it allows for
As data are continuously fed into the system through the maximum customization, so as users of legacy systems easily
DMT, IPT is designed to cope with real-time data requests, and adapt while medical, nursing and technical staff can adopt a
constant queries from the end-users against ontological and digital mode of operation in patient monitoring.
knowledge bases. This layer performs the interpretation of
health information related content to personalized medical
JADE Platform

Ambient Sensor
JADE Platform

Vital signal sensors

JADE Platform

Ambient Sensor

Information Processing Data Management

JADE Agent Interaction

Communication

Fig. 2. The conceptual model of the proposed system

B. System functionality It should be mentioned that the experimental environment

The proposed architecture focuses on patients health described in the following sub-section are developed within the
monitoring. As soon as the vital signs are collected, the scope of a 3-year research framework. Thus the objective of
relevant data are measured against their normal range values this work is twofold: on the one hand it attempts to describe the
and any changes (improvement or deterioration) or stability of system requirements, specifications and design aspects and on
health status are recorded. The values exceeding the normal the other hand it studies the applicability and usefulness of the
range are classified as mild or severe changes. In case of mobile agent paradigm in the healthcare domain.
deterioration, the system broadcasts alerts to healthcare IV. USE CASE SCENARIO AND SYSTEM VALIDATION
professionals and family members. Regarding medical data, the
system is able to collect, store and forward vital signs recorded This section describes the validation not only of the
by a variety of medical sensors (e.g. blood pressure, heart rate, ontological schema but the agent implementation as well. In
temperature etc.). More specifically, monitoring includes: order to verify the validity of the ontological model that is
proposed for healthcare monitoring we translated it to JAVA
─ Signal auditing. Personalized assessment of bio-signals classes that were embedded in agents that were implemented
(e.g. respiratory rate, pulse rate, blood pressure, body over the JADE platform.
temperature etc.).
A. Use Case Scenario
─ Mobility and gesture sensing,
Every healthcare monitoring ecosystem consists of patient
─ Sampling of environmental conditions (e.g. O2, CO, bio-signals, environmental and accelerometer/gyroscope
humidity, air temperature). sensors. Although there are two modes of generated data
forwarding (“push” and “pull”) within the scope of this work

43
we study the “pull” one as it is more complicated and operates Finally the “Get the most recent patient's
on demand, gathering all necessary data. Accordingly, the values” collects all the data that are already assigned to a
“push” mode of operation simply transmits all sensed data to patients CPR.
the processing center. The DMT stores the information –
measurement sent from each sensor’s static agent (see Fig. 2).
In parallel, depending on their assigned role, practitioners and
family members have access to medical/health data. Sensed
data may be further processed by a reasoned engine, that is out
of the scope of this work, in order to diagnose, advise, alert or
configure personalized thresholds for each individual patient.
Upon processing the data of the DMT layer, IPT can inform
practitioners or family members about a patient’s medical
condition.
B. System Validation
In order to verify our proposed system validity, we
developed an experimental environment that implements the
main participants of the scenario described above. More
specifically, each sensor (i.e. ambient and vital signal) is
simulated through a device static JADE agent. The device
static agent is also dedicated to collect and transmit the sensed
data to whenever requested. The CPManagerAgent is the
central processing module of our system and in terms of the Fig. 4. The creation of patients CPRs
client-server paradigm is the equivalent of a “server”. At least a
CPManagerAgent has to be active in any of the JADE
platforms that participate in a typical deployment of our
system. The “client” counterpart of our system is named
CGmphoneAgent and acts on behalf of a caregiver. Each
caregiver has its own agent representative. Other individuals
that should participate (e.g. patients, family members etc.) have
to be represented by a similar agent.
The command line menu to the user provides a set of 7
options as provided in Fig. 3.
<<---- PATIENTS- MENU ---->>

0. Terminate program
1. Create a new patient
2. Make a temperature measurement
3. Make a blood pressure measurement
4. Get the most recent patient's values
5. Get the list of temperature measurements
6. Get the list of blood pressure Fig. 5. A snapshot of the interactions among JADE agents
measurements

Fig. 3. The command-line main menu of the proposed system

The first option creates a Computerized Patient Record (CPR)

and contains a patient’s personal health data. Upon its creation
a unique code and a default nick-name is assigned. In future
implantation it will also hold demographic data, such as
address, phone numbers etc. By using this option we created 3
patients (i.e. patient0, patient1 and patient2) as depicted in Fig.
4.
As we mentioned above we simulate the operation of ambient
and vital signal sensors, thus we assign to the patient2 actor
agent the values for temperature and blood pressure. Fig. 6
depicts the creation of blood pressure values through the
option “Make a blood pressure measurement” from the main
menu (i.e. Fig. 3).
Fig. 6. The “Make a blood pressure measurement” menu option

44
 [3] P. Sernani, A. Claudi, L. Palazzo, G. Dolcini, and A. F. Dragoni, “A
multi-agent solution for the interoperability issue in health
information systems,” CEUR Workshop Proc., vol. 1099, no. Dii,
pp. 91–96, 2013.
[4] H.-K. Kim, “Convergence agent model for developing u-healthcare
systems,” Futur. Gener. Comput. Syst., vol. 35, pp. 39–48, 2014.
[5] C.-H. Liu, Y.-F. Chung, T.-W. Chiang, T.-S. Chen, and S.-D. Wang,
“A mobile agent approach for secure integrated medical information
systems,” J. Med. Syst., vol. 36, no. 5, pp. 2731–2741, 2012.
[6] V. Vaidehi, M. Vardhini, H. Yogeshwaran, G. Inbasagar, R.
Bhargavi, and C. S. Hemalatha, “Agent Based Health Monitoring of
Elderly People in Indoor Environments Using Wireless Sensor
Networks,” Procedia Comput. Sci., vol. 19, pp. 64–71, 2013.
[7] Y. M. Essa, G. Attiya, and A. El-Sayed, “Mobile agent based new
framework for improving big data analysis,” in Cloud Computing
and Big Data (CloudCom-Asia), 2013 International Conference on,
2013, pp. 381–386.
[8] Q. Sun, W. Yu, N. Kochurov, Q. Hao, and F. Hu, “A multi-agent-
based intelligent sensor and actuator network design for smart house
Fig. 7. The “Get the most recent patient's values” menu option and home automation,” J. Sens. Actuator Networks, vol. 2, no. 3, pp.
557–588, 2013.
V. CONCLUSION AND FUTURE WORK [9] P. Angin and B. K. Bhargava, “An Agent-based Optimization
Framework for Mobile-Cloud Computing.,” JoWUA, vol. 4, no. 2,
The proposed system employs an agent implemented pp. 1–17, 2013.
framework for performing ubiquitous patient monitoring. The [10] I. Garc’\ia-Magariño and C. Gutiérrez, “Agent-oriented modeling
resulting ontology supported information system allows for and development of a system for crisis management,” Expert Syst.
Appl., vol. 40, no. 16, pp. 6580–6592, 2013.
the introduction of sensing technology, devices and next
[11] A. R. D. M. Neves, Á. M. G. Carvalho, and C. G. Ralha, “Agent-
generation services (e.g. decision making and reasoning) based architecture for context-aware and personalized event
through autonomous and intelligent agents. recommendation,” Expert Syst. Appl., vol. 41, no. 2, pp. 563–573,
The system is designed so as to improve the provided 2014.
healthcare quality by encapsulating the evolution in remote [12] W. Ying, P. Ray, and L. Lewis, “A methodology for creating
ontology-based multi-agent systems with an experiment in financial
sensing and knowledge bases. Also, healthcare providers are application development,” in System Sciences (HICSS), 2013 46th
able to actively monitor patients with the help of agents Hawaii International Conference on, 2013, pp. 3397–3406.
technology. [13] S. Li and W. A. Mackaness, “A multi-agent-based, semantic-driven
Our future work includes the development and implementation system for decision support in epidemic management,” Health
of a prototype for automated decision-making and health Informatics J., vol. 21, no. 3, pp. 195–208, 2015.
advising logic through the Jess rule language. Consequently, [14] D. Isern, A. Moreno, D. Sánchez, Á. Hajnal, G. Pedone, and L. Z.
Varga, “Agent-based execution of personalised home care
care plans will be implemented by using caregivers’ treatments,” Appl. Intell., vol. 34, no. 2, pp. 155–180, Apr. 2011.
assessment, technological and medical devices along with [15] Yanqing Ji, Hao Ying, M. S. Farber, J. Yen, P. Dews, R. E. Miller,
relevant decision-making software procedures. and R. M. Massanari, “A Distributed, Collaborative Intelligent
Agent System Approach for Proactive Postmarketing Drug Safety
Surveillance,” IEEE Trans. Inf. Technol. Biomed., vol. 14, no. 3, pp.
826–837, May 2010.
REFERENCES [16] K. Miller and G. Mansingh, “Towards a distributed mobile agent
decision support system for optimal patient drug prescription,” in
[1] M. T. Nguyen, P. Fuhrer, and J. Pasquier, “Enhancing Legacy
Innovative Computing Technology (INTECH), 2013 Third
Information Systems with Agent Technology,” J. Int. J. Telemed.
International Conference on, 2013, pp. 233–238.
Appl. issue Electron. Heal. Arch., vol. 2009, 2009.
[17] F. Bergenti, G. Caire, and D. Gotta, “Agents on the move: JADE for
[2] J. L. Cardoso de Moraes, “A methodology based on openEHR
Android devices,” in Procs. Workshop From Objects to Agents,
archetypes and software agents for developing e-health applications
2014, vol. 2.
reusing legacy systems,” Comput. Methods Programs Biomed.,
2016.

45