INSIDabcdef_:MS_0001MS_0001

INSIDabcdef_:MS_0001MS_000

INSIDabcdef_:MS_00 01M

Implementation of interactive healthcare advisor

model using chatbot and visualization
Tae-Ho Hwang, JuHui Lee, Se-Min Hyun, KangYoon Lee*
Department of Computer Engineering
Gachon University
Seongnam, Republic of Korea
{hth7277, juhe2082, tpalsgus0}@gc.gachon.ac.kr, [email protected]

Abstract— In the era of the fourth industrial revolution, Individual’s health history and health-related information is
utilizing the influx of numerous information have significant important for user-centered health care system, allowing
influences on human lives. In particular, the use of artificial continuous monitoring of current health status and for
intelligence data in the medical field will have great potential preventing, diagnosing, and managing the diseases.[2] In
to change and affect the society. Here we report the order to remotely measure a patient's health condition,
components necessary for implementing the interactive
telemetry, ultrasonic sensors, smart sensors, wearable
healthcare advisor model (IHAM) and chatbot-based IHAM.
The biological information of target users used in the research, sensors are being developed in various ways.[3][4] However,
including body temperature, oxygen saturation (SpO2), pulse, there are not enough studies for Interactive healthcare
electrocardiogram (ECG) and etc., was measured and advisor system using chatbots for mutual communication of
analyzed with biological sensors based on oneM2M platform, individual health status, keeping track of health status and
as well as using interactive chatbot to analyze the everyday changes, and providing users with the necessary information.
biophysical conditions. Moreover, the accumulated biological
information in the chatbot and the biological sensors are The rapid development of deep learning and artificial
delivered to the users through the chatbot and further
intelligence technology has expanded incredible academic
provides medical advices to promote the user’s general health.
Through this implementation, we will confirm the
and commercial developments for smart-speakers.
applicability of the health care advisor model in the future and Interactive artificial intelligence services compete in the
present future research directions. smart speaker market and personal assistant services such as
Google ("OK Google"), Apple ("Siri"), Amazon ("Alexa")
Keywords—Interactive Healthcare Advisor, Healthcare and Samsung ("S-Voice"). Due to the development of Deep
chatbot advisor, Chatbot, AI Smart Health, Visualization Learning-based artificial intelligence, the field of voice
recognition has shifted from Natural Language Processing
(NLP) to Natural Language Understanding (NLU) or Natural
I. INTRODUCTION Language Generation (NLG). Voice recognition uses "End-to-
End" speech recognition technology, which uses deep learning
Analyzing and exploiting the information and data technology to simplify the structure of complex voice
provided around us during the 4th industrial revolution is recognition systems, training deep learning neurons only with
directly related to social competitiveness and changes in life. large volume of voice data without specialized domain
Especially in the medical field, highly efficient employing knowledge by language, and converting voice into text.[5]
of biological data using artificial intelligence will bring Interactive health care advisor model has the ability to check
immense changes to the society. Also, this development of the health status of users, analyze health status, predict and
artificial intelligence technology will bring about innovative prevent diseases, and provide users with a method of
changes in the human health care and medical industry. enhancing health conditions.
Continuous health care, such as personal biological signal The objectives of this study are as follows: First, the
information, health checkup information, and life pattern biological signals of the user's temperature, oxygen saturation
management, has many effects on human quality of life and (SPO2) and ECG are measured to analyze the change in
disease prevention. In particular, diseases with high biological information. Second, chatbots are used to analyze
mortality rates and aftereffects of brain diseases such as the health status of the users. Third, the implementation of the
dementia and Alzheimer’s disease need early detection Interactive Healthcare Advisor Model provides health care
through regular medical checkups since earlier detection information to users through chatbots.
leads to more effective treatment.[1] The conventional For this purpose, the user’s biological information is
healthcare advisory system currently serves as a passive acquired from the biological sensor, and the chatbot is
supplementary role in health care, delivering one-sided implemented by using the Kakao i open builder provided by
information. While it is possible to diagnose one’s health Kakao.
through visiting hospital and receiving the medical
examination results, significant difficulties remain for the
residents in the rural area. Interactive healthcare advisor II. RELATED TECHNOLOGIES
chatbot that has studied disease symptoms, risks and
prevention can provide a better opportunity for the patients. A. Chatbot
* : Corresponding author

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

978-1-7281-6758-9/20/$31.00 ©2020 IEEE 452 ICTC 2020
zed licensed use limited to: MINISTERE DE L'ENSEIGNEMENT SUPERIEUR ET DE LA RECHERCHE SCIENTIFIQUE. Downloaded on July 22,2024 at 09:18:55 UTC from IEEE Xplore. Restrictions
 INSIDabcdef_:
INSIDabcdef_:MS_00 01M

Chatbot is a compound word of chat and robot, which

refers to an artificial intelligence-based communication
system that understands and answers questions through
conversations with humans based on artificial intelligence.
The initial model of the chatbot was developed in 1966, and
the recent development of artificial intelligence technology
has allowed deep learning technology to be applied to the
chatbot to understand the user's questions and to present
appropriate answers, and to perform complex tasks.[5]

Chatbots can be classified as "rules-based chatbots" that Fig. 1. Chatbot Working Process
operate based on predefined rules according to their operating
methods and "machine-learning-based chatbots" that
understand natural languages and generate or select B. Healthcare Advisor by Chatbot
appropriate answers. The machine learning algorithm consists
of the Natural Language Process (NLP) module that acts as a According to a study by the cognitive health advisor model
natural language decomposition, the Natural Language that reflects the concept of CPS (Cyber-physical systems), [8]
Understanding (NLU) module that identifies the meaning of the composition of the cognitive health advisor platform
the natural language, and the Natural Language Generation consists of as follows: First, an open platform should be
(NLG) module that selects or generates the defined answers. supported. Second, it should be composed of an open platform
Furthermore, chatbots are divided into one-time question-and- so that information collected and analyzed can be interfaced
answer chatbots that provide a set answer to frequently asked with other systems through various interfaces. Moreover, the
questions, and continuous chatbots that collect information to functional composition of Cognitive health advisor is as
achieve a specific purpose. Among the continuous interactive follows: First, lifelogging and behavioral analysis, biometric
chatbots, chatbots that operate according to a set scenario is analysis and prediction, second, the management of health
known as frame-based system. In addition, chatbots can be records related to the EMR data, third, a conversational
classified as "search model chatbots" that answer users' interface.[7][8][9]
questions correctly in Knowledge DB and "creation model Healthcare chatbots aim to solve health-related problems
chatbots" that generate new, previously undefined answers to and provide efficient medical information. Major global
users' questions, which require a lot of learning data.[6] Key healthcare companies in chatbot market include Sensely,
technologies in chatbot include pattern recognition Buoy Health, Your.MD, and Florence, providing users with
technology that identifies shapes, characters, voices, etc., medical help from chatbots on any platform in both the web
natural language processing technology that recognizes and and app. (Fig.2)
processes human language on computers, Semantic Web Sensely uses the bot to allow patients to talk about
where computers can understand the meaning of information symptoms and receive appropriate self-treatment advice, and
resources, text mining to find new and useful information in to communicate using Chat mode and Voice mode. Sensely
unstructured text data, and context aware computing uses a variety of characters to build a unique brand persona.
technology that utilizes information in virtual space. [5][6][7]. Your.MD is available as a web and app, and provides medical
information tailored to the user's gender and age. Your.MD
In this paper, we used Kakao i Open Builder as chatbot acts as a personal health helper, checking the user's symptoms
builder. Kakao i Open Builder is a chatbot design platform and giving health-related quizzes. Florence receives data from
developed by Kakao. Open Builder enables interactive bot the user and collects health information over time. It also
designs that allow users to interact quickly, and to create sends an alarm to the user about the time he or she takes the
chatbots with interactive interfaces provided through Kakao medicine.
Talk channels. Implementing chatbots using Kakao i
Openbuilder requires writing entities, intents, and scenarios.
First, entity is a systematic arrangement of terms that chatbots
can understand. If entities are defined, chatbots can extract
key data from user's input to identify users' intentions.
Medical information such as types of biological data, related
exercise, and applicable first aid treatments were established
as entities. basic unit of the user's intention. According to the
user's input, the user's intention, the bot's action, and the
content to respond are designed through blocks. Blocks are
designed as units of bot's performance missions, such as "This
week's guidance on blood glucose changes, good lifestyle for
the heart, and exercise good for high blood pressure." Third, a
scenario is a unit of service that users can experience within a Fig. 2. Healthcare Service company by chatbot
bot, in which one scenario comprises of various blocks. The
bot operator can group and manage a number of blocks by the
desired service on a per scenario basis. The chatbot reacts to
As such, healthcare chatbots are used on various platforms.
user’s input through the process shown in in Fig.1, identifying
Finally, we expect the role of chatbots in the medical industry
the intention of the user and delivers the appropriate
to become more important, requiring services such as real-
information.

453
zed licensed use limited to: MINISTERE DE L'ENSEIGNEMENT SUPERIEUR ET DE LA RECHERCHE SCIENTIFIQUE. Downloaded on July 22,2024 at 09:18:55 UTC from IEEE Xplore. Restrictions
 INSIDabcdef_:
INSIDabcdef_:MS_00 01M

time monitoring health conditions and early diagnosis of The development of sensors and communication
diseases. technologies enables users to monitor their health status and
behavior with various biological sensors.[10] These
biomonitoring devices comprises a sensor device for
III. INTERACTIVE HEALTHCARE ADVISOR converting the biological signal into a digital signal and a
controller module for collecting and processing the biological
A. Advisor System Configuration signal. Recently, various technologies such as physical
condition warning [11] and Chronic Disease detection [12]
The data of the Interactive Healthcare Advisor have been developed.[13]
implemented in this study were divided into the steps of
acquiring biological information and health status information,
the biological information analysis and health information
analysis stage, the health status classification stage, and the
educational stage for providing health information according
to the health status and promoting health.

Fig. 5. Bio-Sensor Interface Process

Body temperature, blood pressure, pulse, oxygen

saturation (SPO2) and electrocardiogram (ECG) sensors were
used to measure users' biological signals. Each sensor
measures the biological signal and sends the acquired
Fig. 3. Data Processing Model biological signal to the server according to the data process.

Step 1) During the information acquisition phase, C. Health Advisor service by Chatbot
biological information is measured and acquired using
The chatbot, created with Kakao Open Builder, is managed
biosensors such as body temperature, blood pressure, pulse
in a scenario in which various blocks responding to users'
rate, spO2 and ECG. User’s health status information is
intentions are assembled.
acquired through Interactive bilateral conversation using a
chatbot.
Step 2) The "health-caution" and "health-danger" sections
were established according to the level of risk and the criteria
for prevention according to the characteristics of each
biological signal.
Step 3) Using the results of the analysis stage, a normal
section, a caution section, and a danger section were divided
to classify the user's health condition.
Step 4) Depending on the classified health status, the
chatbot informs users of their health status, and provides
information or recommends exercises necessary for
improving health.
The Interactive Healthcare Advisor, comprising of sensing
platform and chatbots, consists of a data layer that acquires
and manages biometric data, an information layer that Fig. 6. Chatbot Scenario Table
manages health information, a knowledge layer that manages
knowledge acquired using information, and a service layer The types of biological information measured through the
that delivers health information to users. controller vary, and the normal range and useful exercise and
food are all different for each individual. Accordingly, in order
to respond an answer suitable for the user's input, it is
important that the correct information of the biological data to
be obtained and the user's intention.

Fig. 4. “Interactive Healthcare Advisor” System Configuration

B. Bio-signal Sensor Interface Fig.7. Health Data Interface Process

454
zed licensed use limited to: MINISTERE DE L'ENSEIGNEMENT SUPERIEUR ET DE LA RECHERCHE SCIENTIFIQUE. Downloaded on July 22,2024 at 09:18:55 UTC from IEEE Xplore. Restrictions
 INSIDabcdef_:
INSIDabcdef_:MS_00 01M

biological signals could be supervised, and that information

The scenario was written for Health Advisor chatbot work could be provided to the user using a chatbot system. However,
as shown in Fig. 8. The user's biological information can be this study served as a simple healthcare advisor based on the
checked by a controller, and information such as the amount provided biological information and scenarios, and had
of exercise can be obtained from the user through a chatbot. limitation for disease prediction and diagnosis using medical
information. For the proposed Interactive Health Care Advisor
to be more efficient in healthcare applications, an artificial
intelligence algorithms that reflects and judges the health
abnormalities from the biological information, lifestyle
patterns need to be developed. As a future direction, it is
necessary to conduct research such as "health condition
determination algorithm" corresponding to biological
information and individual life patterns, "health condition
change prediction algorithm" based on biological information
changes and life patterns, and "Healthcare Advisor Process
Algorithm" using chatbots.

Fig. 8. Chatbot Scenario Diagram ACKNOWLEDGMENT

This research was supported by the MSIT(Ministry of
Science and ICT), Korea, under the ITRC support
In the flow chart of the data process of the chatbot (Fig. 9) program(IITP-2020-2017-0-01630) and was supported by the
the user measures health data using a controller. Measured National Research Foundation of Korea(NRF) grant funded
information is uploaded to ChatBot Cloud Server, which (NRF-2019R1F1A1057663).
delivers appropriate medical information to users. Users can
self-diagnose and obtain information through chatbots.
REFERENCES
[1] The Relationship between Alzheimer's Disease and Diabetes: Type 3
Diabetes Kroner, “Alternative Medicine Review” Vol. 14, Issue 4, Pages
373-379, December 2009
[2] A. Lymberis, "Smart wearable systems for personalized health
management: current R&D and future challenges," Proceedings of the
25th Annual International Conference of the IEEE Engineering in
Medicine and Biology Society (IEEE Cat. No.03CH37439), Vol.4, pp.
3716-3719, 2003
[3] A. Lmberis and A. Dittmar, "Advanced Wearable Health Systems and
Applications - Research and Development Efforts in the European
Union," in IEEE Engineering in Medicine and Biology Magazine, vol. 26,
Fig. 9. Chatbot Data Process Diagram no. 3, pp. 29-33, May-June 2007
[4] Fabrice Axisa, P. M. Schmitt, C. Gehin, G. Delhomme, E. McAdams and
A. Dittmar, "Flexible technologies and smart clothing for citizen
After visualizing the health information provided by medicine, home healthcare, and disease prevention," in IEEE
chatbots,(Fig. 10) it was provided to users so that they could Transactions on Information Technology in Biomedicine, vol. 9, no. 3,
more effectively identify and manage their health conditions. pp. 325-336, Sept. 2005
[5] S.K. Kim, M.C.Sin, J.Y.Kang, “Chatbot technology introduction and case
analysis“, Korea information & communication society, 35-2-8 P21-28,
2018
[6] Chung, K., Park, R.C. Chatbot-based heathcare service with a knowledge
base for cloud computing. Cluster Comput 22, 1925–1937, 2019
[7] K. Lee, “Reference Model and Architecture of Interactive Cognitive
Health Advisor based on Evolutional Cyber-physical Systems”, KSII
Transactions on Internet and Information Systems Vol. 13, no. 8, Aug.
2019
Fig. 10. Chatbot Communication Visualization [8] E. Lee, “The past, present and future of cyber-physical systems: A focus
on models.” Sensors, 15(3): p.4837-4869
[9] Muller, H.A., “The Rise of Intelligent Cyber-Physical Systems”.
Computer, p. 7-9, 2017(12)
IV. CONCLUSION [10] M. Chen et al., “Narrow Band Internet of Things: A Survey,” IEEE
Access, vol. 5 , pp. 20557-20577, 2017
In this study, we implemented the Interactive Healthcare
[11] Y Zhang et al., “Home M2M Networks: Architectures, Standards, and
Advisor System to measure the basic biological information QoS Improvement,” IEEE Commun. Mag., vol. 49, no. 4, pp. 44–52,
such as body temperature, blood pressure, pulse, oxygen Apr. 2011
saturation (SpO2) and electrocardiogram (ECG) and check the [12] M. Chen et al., “Disease Prediction by Machine Learning over Big
individual user’s health conditions using a chatbot. In addition, Healthcare Data,” IEEE Access, vol. 5, pp. 8869–79, June 2017
we established a Healthcare Advisor Use Case scenario and [13] M. Chen, Y. Zhang, M. Qiu, N. Guizani and Y. Hao, "SPHA: Smart
implemented chatbots using the open builder provided by Personal Health Advisor Based on Deep Analytics," in IEEE
Communications Magazine, vol. 56, no. 3, pp. 164-169, March 2018
Kakao. We have successfully confirmed that changes in user's
455
zed licensed use limited to: MINISTERE DE L'ENSEIGNEMENT SUPERIEUR ET DE LA RECHERCHE SCIENTIFIQUE. Downloaded on July 22,2024 at 09:18:55 UTC from IEEE Xplore. Restrictions