IoT-based Health Monitoring System for Intensive Care Units

Adriana Collaguazo

, Rebeca Estrada

and Nestor Arreaga

Escuela Superior Politcnica del Litoral, ESPOL,

Electrical Engineering and Computer Science and Technology Information Center,

Campus Gustavo Galindo Km. 30.5 Via Perimetral, P.O. Box 09-01-5863, Guayaquil, Ecuador

Keywords:

Mobile App, Intensive Care Unit, Smart Environment, IoT Health Application.

Abstract:

Due to the critical condition and unique treatments required by ICU’s patients, their vital signs should be

continuously and reliably monitored. In this paper, we present a system to address this problem including the

design considerations. Our system has three components: i) a mobile application to present the collected data

from the hospitals server, ii) a database server with patients data and iii) a transactional server that manages

and keeps the connection between the other two components. The mobile application implements a vital signs

monitor for each patient including an underlying emergency notiﬁcation system. After several simulations,

we were able to identify that the proposed solution takes around 3 seconds to generate the notiﬁcation after

an emergency alarm occurs. Moreover, the application uses an average of 2.11 Kb/s of trafﬁc data and a

maximum of 250Mb of RAM. This system is a feature that will help to track the ICU patients status, allowing

doctors and ICU managers to work outside the unit and to determine when their presence is required. Results

show that relevant and accurate notiﬁcations can effectively reduce the time response in cases of emergency

and consequently increases the likelihood of the patients’ recovery.

1 INTRODUCTION

The Intensive Care Unit (ICU) is a specialized area

of a hospital that provides care to patients with se-

vere or life-threatening illnesses and injuries. This

area requires constant and close supervision from life

support equipment and medication to ensure normal

bodily functions (National Health Service, ). A con-

tinuous physician supervision is recommended for pa-

tients under these conditions, however, treating physi-

cians cannot remain in the ICU unit all the time to take

care of these patients. Instead, resident doctors re-

main in the area to constantly monitor each patient. In

the event of an emergency, resident physicians should

contact the treating physicians, either by using a com-

munication devices (e.g., phone or beeper) or by per-

sonally searching for them, leading to higher response

time in emergency cases.

This paper proposes the design and implementa-

tion of an automated alert system to tackle the above

problem. For instance, a case study was carried out

in the ICU of Guayaquil Hospital, in Ecuador. In

this hospital, the ICU unit has 21 places for criti-

https://orcid.org/0000-0002-0707-0226

https://orcid.org/0000-0003-3957-9294

cal care, of which 7 places are for isolated patients,

and serves approximately 15 patients (Ministerio de

Salud, 2018). The proposed system have several com-

ponents, such as: 1) a mobile application to present

the collected data from the hospitals server, 2) a

database server with patients data and 3) a transac-

tional server that manages and keeps the connection

between the other two components.

The mobile application shows the vital signs of

patients and each doctor can login into the applica-

tion, then, they can see the area with several win-

dows and the sectors of the ICU and their respec-

tive beds. Accelerometer sensors are used to monitor

the patients’ movement. In the event of any strange

movement, the physicians can get an idea of patients’

position through the report patients’ physical status.

Moreover, alerts are automatically generated to notify

doctors when a patient’s vital signs are outside the es-

tablished range. This type of emergency is displayed

in the application by changing the buttons of the sec-

tor, the bed and the parameter to a red color.

This work implements a solution that is able to:

• Report constantly the vital signs of patients by

means of non traditional systems.

• In case of emergency, alert are sent to the respec-

Collaguazo, A., Estrada, R. and Arreaga, N.

IoT-based Health Monitoring System for Intensive Care Units.

DOI: 10.5220/0011339800003286

In Proceedings of the 19th International Conference on Wireless Networks and Mobile Systems (WINSYS 2022), pages 101-106

ISBN: 978-989-758-592-0; ISSN: 2184-948X

101

tive staff in a hierarchical manner.

• Reduce the current average response time.

This paper is organized as follows: Section 2 de-

scribes the prior investigation of the equipment and

materials used to design the system. In section 3 we

presents the proposed system with the connection be-

tween the different modules. Section IV shows the

results of the application tests. Finally, section 5 con-

cludes our research work.

2 RELATED WORK

Researchers of the preliminary studies have managed

the real-time data over networks to be available to

clinicians in the Intensive Care Unit (ICU) anywhere

on the web with appropriate software and privileges.

Considering the IoT applications covers smart en-

vironments in domains such as emergency, health

care, and user interaction(Dr. Ovidiu Vermesan et al.,

2017).

According Thibaud (Thibaud et al., 2018), iden-

tiﬁes that in the near future (within 5 years) several

theoretical or pilot projects will tackle different is-

sues such as better integration of pervasive health-

care services with general health care services in a lo-

cal database environment that ensures data availabil-

ity and that performs intelligent processes to deliver

quicker preliminary information without compromis-

ing the energy-efﬁciency of sensor networks.

In (Lamberti and Montrucchio, 2003), the medical

staff equipped with Personal Digital Assistant (PDA)

devices both inside and outside the hospital have mo-

bile access to the electronic patient’s clinical record.

Using a framework for ubiquitous monitoring in an

ICU, by the bedside monitoring network, on secure

wireless communication channels. Although, the val-

idation of the proposed framework effectiveness is

fundamental, and requires the design of the software

modules needed.

The system called ADSA (Automatic Detection

of risk Situations and Alert) (Ahouandjinou et al.,

2016) is based on a hybrid architecture for a visual

patient monitoring system using a multi-camera sys-

tem and collaborative medical sensors network was

developed. Although this proposal proved to enable

personalization of treatment and management options

targeted particularly to the speciﬁc circumstances and

needs of the individual, it requires the use of cameras

to support the decision process which is not possible

considered the ﬁnancial constraints faced by the hos-

pital used as a case of study.

In (Gupta et al., 2016) the IoT-based health moni-

toring system can provide support in Intensive Care

Units(ICU) using an INTEL GALILEO 2ND. This

system contains a live graph of the patients heart rate

and the temperature is being monitored. However, the

system could have been developed in a mobile appli-

cation for facilitating the access of the users.

In (Chiuchisan et al., 2014) propose the architec-

ture of a health care system for Intensive Care Unit

(ICU) through of bedside monitors to monitor and

record multiple physiological parameters of patients;

Microsoft XBOX Kinect to monitor the movement of

the patients; and sensor board for monitoring of envi-

ronmental parameters such as temperature, humidity,

atmospheric pressure. The system is part of a more

complex system in development and will be improved

by adding new types of sensors like pressure, body

weight.

In (Pickering et al., 2018) using the Continuous

Time Markov Chain (CTMC) in the develop a bot-

tleneck analysis method to identify opportunities for

improvement in the rounding process in an Inten-

sive Care Units at Mayo Clinic. The workﬂow re-

design needs further investigation as the ideal balance

between patient care activities and education is still

largely unknown and is likely to vary depending on

the circumstances.

In (Mahmud et al., 2018) present a Fog-based IoT-

Healthcare solution structure and explore the integra-

tion of Cloud-Fog services in inter-operable Health-

care solutions extended upon the traditional Cloud-

based structure. The mobility of the user and edge-

centric afﬁnity of the applications should be handled

together by the Fog cluster for better performance.

3 SYSTEM DESIGN

In this section, we present the design of the proposed

system taking into account the users requirements to

better asset their needs. Based on this requirements

a top-level architecture is proposed. In addition, the

back end server components are described as well as

the application and node design.

3.1 User Requirements

From a user perspective, collecting and sharing health

information of patients must satisfy at least the fol-

lowing properties:

• Data accuracy: A margin of error should be es-

tablished for the data measuring. In this particular

case, health measures must be very precise since

the slightest difference can have a serious or un-

wanted impact.

WINSYS 2022 - 19th International Conference on Wireless Networks and Mobile Systems

102

• Availability: The system is used by different users

who have different access roles and requirements,

and therefore, must be available to all users any-

time, anywhere.

• Performance: The system should offer a satisfac-

tory performance to end users. The response time

in all procedures should be as low as possible.

• Data privacy: The privacy of health data must

be protected for all users, especially those under

critical monitoring. The system should provide a

proven mechanism to ensure privacy.

3.2 System Architecture

Figure 1 presents the top-level architecture of the pro-

posed ICU automation system. The architecture con-

sists of a Sink Node that is connected to the back-end

server using the Local Area Network (LAN). The ex-

isting infrastructure provides a WiFi physical layer to

connect the smart devices or phones to the local net-

work. The intranet is also connected to this WiFi net-

work. Thus, these devices provide the primary human

machine interface for the nurses, doctors and other

staff. The server has its own database and application

servers and MySQL is used as the Data Base Manage-

ment System (DBMS) engine while Apache provides

the application server functionality. The application is

Web-enabled, which requires only a standard browser

at the client side.

The ﬁrst module is composed of the sensors de-

ployed on the beds. These sensors are connected to

a gateway designed with an Arduino UNO and an

Ethernet Shield, connected directly to the hospital’s

internal network where the data is transmitted and

stored in the MySQL database. In addition, each pa-

tient has a monitor connected directly to the hospital

network, which provides their vital signs and stores

this information in a proprietary database. The sec-

ond module consists of a PHP-based server based and

a MySQL server. The server contains several scripts

to perform search, add or update information in the

database when a request is received from the mobile

application. The third module is a mobile application

that allows physicians to visualize patient data.

3.3 Sink Node

The sink node consists of an accelerometer sensor and

a wireless transceiver, connected to an Arduino Nano,

on the other side each node is connected to BeneView

T8 vital signs monitor. The motion sensor MMA7361

is used to detect a fall from the bed. This sensor al-

lows us to measure the acceleration of 3 axes (x, y,

Table 1: Connections between Arduino and transceivers.

Transmitter Receiver

nrf24101 Arduino nrf24101 Arduino

nano UNO

VCC 3.3V VCC 3.3V

GND GND GND GND

CSN 10 CSN 8

CE 9 CE 7

MOSI 11 MOSI 11

SCK 13 SCK 13

MISO 12 MISO 12

z), and it was calibrated correctly at the time of use.

Then we connected it to the nano Arduino, together

with a nrf24101 transceiver that allows the connec-

tion between the nodes and the sink node (Table 1).

The Arduino nano is programmed so that when the

severity of an axis changes abruptly, the Arduino will

the sink node to notify the emergency.

The sink node was built over an Arduino UNO, to

which an ethernet shield was added to be able to com-

municate with the server. When the Arduino UNO re-

ceives the signal that the patient fell, he connects with

the server and sends him a request with the number

of the bed to generate an alarm in the database which

the cell phone will notify the doctors. This module

should be improved with equipment with higher ca-

pacity and durability, which can support long periods,

such as a Beaglebone or a Raspberry.

3.4 Transactional Server

The transactional server consists of several PHP

scripts to perform the requests generated by the users

of the mobile application. These scripts receive a re-

quest from the application to search information from

the database and return the requested query in a spe-

ciﬁc format, which is transferred through JSON mes-

sages because of their simplicity and the speed to

transfer the data. To prevent the leakage of sensi-

tive information or problems such as SQL injection,

the data received in each request was sent separately

and parameterized, and no script contains queries that

could reveal data about database structures. More-

over, as an extra security measure, the patient’s in-

formation and vital signs are not stored in the device

running the mobile application.

3.5 Database Server

The management of MySQL database was carried out

by means of PHPMyAdmin software tool. With this

tool, we were able to maintain an up-to-date copy of

IoT-based Health Monitoring System for Intensive Care Units

103

Figure 1: Top-level system architecture of the UCI system.

the database structure and its contents from a hospi-

tal vendor’s proprietary equipment. Thus, we will not

affect the integrity of the actual data of the patients in

the intensive care unit. The patient’s name, location,

blood pressure, heart rate, temperature and oxygena-

tion are some of the information stored as the patient

data. This data can also be consulted through the mo-

bile application.

3.6 Mobile Application

The mobile application was developed using Android

Studio for phones with Android operating system be-

cause it is a prevalent mobile operating system in

Ecuador. This IDE has available libraries that allow

direct interaction with mobile functions and offers a

suitable graphical environment for view design.

The application generates POST requirements to

communicate with the PHP server. Figure 2 illustrates

different views that the mobile application has. The

ﬁrst is used to allow the user can connect to the sys-

tem using their identiﬁcation number and password.

The second view shows the four sectors of the ICU

area and by selecting any of these, the numbered beds

in the sector is shown. For each bed, the app shows

the patient’s name when there is a patient using the

bed otherwise it shows ”NO DATA”. By choosing the

patient, a parameter view is presented, where the val-

ues of the patient’s vital signs in real time.

Figure 2: Mobile Application Interfaces.

To facilitate the data analysis, the user can choose

any of the parameters and a real-time graph of this

parameter is shown. In addition, the user can choose

to forecast the trend of the parameter, which is done

using the 3rd order moving average of the parameter.

To obtain this data, the system generates requests to

the server every 0.5 seconds sending the correspond-

ing information of the selected parameters, bed and

section.

The mobile application has a persistent service

that keeps the medical staff aware of the patient’s sta-

tus. This service sends a request every 3 seconds and

thus be able to issue timely emergency alerts.

WINSYS 2022 - 19th International Conference on Wireless Networks and Mobile Systems

104

When an event is identiﬁed, a vibrating and sound

alert notiﬁcation is generated. Such notiﬁcations may

not be silenced unless the user voluntarily logs out or

the emergency is resolved.

For the generation of alarms, a hierarchical clas-

siﬁcation is implemented according to users’ role as

follows:

• Physician on shift: instant alert generation.

• Sector supervisor: 5 minutes after the emergency

alert is generated.

• Area supervisor: 10 minutes after the emergency

alert is generated.

To keep control of notiﬁcations, the user’s role

is saved in the internal memory of the mobile phone

while his session is active.

4 RESULTS ANALYSIS

Once the system was fully developed, we perform

several tests to validate the functionality of each com-

ponent. The tool used for testing was Android Proﬁle,

which is part of the Android Studio software package

() [7]. This tool allows the developer to determine the

usage of CPU, memory, power and data. Each test

was performed ﬁfteen times and the results were av-

eraged. The ﬁrst test consisted of measuring the total

time between the generation of an emergency in the

database and the notiﬁcation in the mobile application

to determine the average time to generate emergency

notiﬁcations.

Each sample time period began with the activa-

tion of the emergency trigger and ended with the re-

ceived pop-up notiﬁcation. For better precision, a dig-

ital trigger was used to start the tests. Also, the trigger

system and the alarm were installed on different net-

works and separated by a distance of 9.1 miles.

During the initialization period, the average re-

sponse took around 5.5 seconds with standard devia-

tion of 1.5s. With the system in steady state, the mean

response time was 3 seconds with a standard devia-

tion of 0.5s, as shown in Fig 3

The second test was used to measure mobile data

consumption and to estimate the maximum band-

width consumption during a 20-second period while

keeping the real-time parameters window open, as it

is the most resource-consuming window of the appli-

cation due to the constant checking of the different

vital signs of the patient.

Fig. 4 shows the data rate used for transmission

while performing the parameter analysis in real-time

since it is the most demanding activity of the mobile

Figure 3: Tests of the time interval between the patient’s fall

and the notiﬁcation of the alert.

Figure 4: Data Rate for the functionality of parameter anal-

ysis in real-time.

application. Each test was run for a duration of 20

seconds, and the average data rate was 2.11 Kbps.

The third test allows us to measure the bandwidth

consumption of the notiﬁcation generation service to

determine the minimum bandwidth consumption of

the mobile application when it is in the background.

This test consists of having the alarm system running

in background mode considering a time 20-seconds

interval. In this case, the average data rate was 0.51

Kbps, as shown in Fig. 5

The purpose of the paper was to analyze the

amount of resources required on the mobile phone to

run the application in a controlled environment. How-

ever, it is important to remark that the used sensor was

a low cost sensor, which means that it is device with

a time-consuming maintenance. In addition, the ac-

celerometer starts to overheat after a few hours, which

might affect the response time and the accuracy. This

should be taken into account when implementing a

sustainable solution for a real environment.

IoT-based Health Monitoring System for Intensive Care Units

105

Figure 5: Data Rate required for notiﬁcation generation ser-

vice.

5 CONCLUSIONS

In this paper, we presented an innovative system that

enables the medical staff to check and analyze the vi-

tal signs of ICU’s patients from distance, having a

constant control of them. Pop-up notiﬁcations alert

the response team if one or more patients have a sig-

niﬁcant change of vital signs. This allows the med-

ical team to react quickly and apply the appropri-

ate treatment. The database access table helps the

hospital management team to control and protect pa-

tient data. This also helps the system provide a hi-

erarchically controlled alert. The resources required

for mobile application allows it to be run in low-mid

range and higher phones. The application design fa-

cilitates the training period its users, which helps the

users’ work environment. As future work, we plan

to make the system more functional, allowing to es-

tablish relationships between several parameters and

obtain more accurate information about the patient’s

condition. Patients’ data from the system could be an-

alyzed with the machine learning algorithm to predict

their status and reduce the generation of emergency

notiﬁcations to the cases where the presence of the

main doctor is really necessary.

REFERENCES

Ahouandjinou, A., Assogba, K., and Motamed, C. (2016).

Smart and pervasive ICU based-IoT for improving

intensive health care. In 2016 International Con-

ference on Bio-engineering for Smart Technologies

(BioSMART), pages 1–4. IEEE.

Chiuchisan, I., Costin, H. N., and Geman, O. (2014).

Adopting the internet of things technologies in health

care systems. In EPE 2014 - Proceedings of the 2014

International Conference and Exposition on Electri-

cal and Power Engineering, pages 532–535. Institute

of Electrical and Electronics Engineers Inc.

Dr. Ovidiu Vermesan, SINTEF, N., Friess, D. P., and EU, B.

(2017). Internet of Things: Converging Technologies

for Smart Environments and Integrated Ecosystems.

Gupta, P., Agrawal, D., Chhabra, J., and Dhir, P. K. (2016).

IoT based smart healthcare kit. In 2016 International

Conference on Computational Techniques in Infor-

mation and Communication Technologies, ICCTICT

2016 - Proceedings, pages 237–242. IEEE.

Lamberti, F. and Montrucchio, B. (2003). Ubiquitous real-

time monitoring of critical-care patients in intensive

care units. In Proceedings of the IEEE/EMBS Region

8 International Conference on Information Technol-

ogy Applications in Biomedicine, ITAB, volume 2003-

Janua, pages 318–321.

Mahmud, R., Koch, F. L., and Buyya, R. (2018). Cloud-Fog

Interoperability in IoT-enabled Healthcare Solutions.

In Proceedings of the 19th International Conference

on Distributed Computing and Networking - ICDCN

’18, pages 1–10, New York, New York, USA. ACM

Press.

Ministerio de Salud, E. (2018). Rendici

on de cuentas 2018.

National Health Service. Intensive care - NHS.

Pickering, B., Li, J., Dong, Y., Gajic, O., and Lee, H. K.

(2018). Bottleneck Analysis to Improve Multidisci-

plinary Rounding Process in Intensive Care Units at

Mayo Clinic. IEEE Robotics and Automation Letters,

3(3):2678–2685.

Thibaud, M., Chi, H., Zhou, W., and Piramuthu, S. (2018).

Internet of Things (IoT) in high-risk Environment,

Health and Safety (EHS) industries: A comprehensive

review. Decision Support Systems, 108:79–95.

WINSYS 2022 - 19th International Conference on Wireless Networks and Mobile Systems

106