sensors

Article
Implementation of ISO/IEEE 11073 PHD SpO2 and ECG Device
Specializations over Bluetooth HDP following Health Care
Profile for Smart Living
Alexandra Cristobal-Huerta 1,† , Angel Torrado-Carvajal 1,2,† , Cristina Rodriguez-Sanchez 1 ,
Juan Antonio Hernandez-Tamames 1 , Maria Luaces 3 and Susana Borromeo 1,2, *

1 Electronic Technology Area, Universidad Rey Juan Carlos, 28933 Madrid, Spain;
[email protected] (A.C.-H.); [email protected] (A.T.-C.);
[email protected] (C.R.-S.); [email protected] (J.A.H.-T.)
2 Medical Image Analysis and Biometry Lab, Universidad Rey Juan Carlos, 28933 Madrid, Spain
3 Hospital Universitario Clínico San Carlos, 28040 Madrid, Spain; [email protected]
* Correspondence: [email protected]; Tel.: +34-91-488-46-53
† These authors contributed equally to this work.

Abstract: Current m-Health scenarios in the smart living era, as the interpretation of the smart city
at each person’s level, present several challenges associated with interoperability between different
clinical devices and applications. The Continua Health Alliance establishes design guidelines to
standardize application communication to guarantee interoperability among medical devices. In
this paper, we describe the implementation of two IEEE agents for oxygen saturation level (SpO2)
measurements and electrocardiogram (ECG) data acquisition, respectively, and a smartphone IEEE
Citation: Cristobal-Huerta, A.; manager for validation. We developed both IEEE agents over the Bluetooth Health Device Profile
Torrado-Carvajal, A.; following the Continua guidelines and they are part of a telemonitoring system. This system was
Rodriguez-Sanchez, C.;
evaluated in a sample composed of 10 volunteers (mean age 29.8 ± 7.1 y/o; 5 females) under
Hernandez-Tamames, J.A.; Luaces,
supervision of an expert cardiologist. The evaluation consisted of measuring the SpO2 and ECG
M.; Borromeo, S. Implementation of
signal sitting and at rest, before and after exercising for 15 min. Physiological measurements were
ISO/IEEE 11073 PHD SpO2 and ECG
assessed and compared against commercial devices, and our expert physician did not find any
Device Specializations over Bluetooth
HDP following Health Care Profile
relevant differences in the ECG signal. Additionally, the system was assessed when acquiring and
for Smart Living. Sensors 2022, 22, processing different heart rate data to prove that warnings were generated when the heart rate was
5648. https://doi.org/10.3390/ under/above the thresholds for bradycardia and tachycardia, respectively.
s22155648
Keywords: Bluetooth HDP; Continua Health Alliance guidelines; digital signal processors (DSPs);
Academic Editor: Georg Fischer
electrocardiography; ISO/IEEE standards; SpO2; telemedicine
Received: 27 May 2022
Accepted: 20 July 2022
Published: 28 July 2022

Publisher’s Note: MDPI stays neutral

1. Introduction
with regard to jurisdictional claims in M-Health was first introduced as “Unwired e-med” in the IEEE Transactions on Infor-
published maps and institutional affil- mation Technology in Biomedicine journal in 2000 [1]. Over the past 20 years, the development
iations. and advancements in information and communication technologies (ICTs) and connectivity
has impacted m-Health directly. In this sense, smart living is the interpretation of the
smart city but at the level of each person’s home, creating a new lifestyle thanks to the
inclusion of these ICT improvements in their own personal spaces. This new way of living
Copyright: © 2022 by the authors.
involves, among others, the fields of health, safety, home automation, and smart services,
Licensee MDPI, Basel, Switzerland.
where new technologies are present in the complete cycle of patient treatment. Concretely,
This article is an open access article
new scenarios have emerged where healthcare professionals have access to the Hospital
distributed under the terms and
Information System (HIS) anywhere and anytime [2,3].
conditions of the Creative Commons
In this framework, new technologies should be present in the complete cycle of
Attribution (CC BY) license (https://
patient treatment and ICTs can improve healthcare in developed economies by minimizing
creativecommons.org/licenses/by/
4.0/).
unnecessary visits to a hospital [4–6]. These techniques and devices for early measurement

Sensors 2022, 22, 5648. https://doi.org/10.3390/s22155648 https://www.mdpi.com/journal/sensors

Sensors 2022, 22, 5648 2 of 12

of parameters allow for the continuous and ubiquitous monitoring of patients through
medical decision support applications (alarms, shared diagnosis, DSS (Decision Support
systems), etc.), helping in the diagnosis, monitoring, and intervention of patients. The
deployment of these applications improves the ubiquity of healthcare providers in hospitals,
increasing the time they spend with patients [6–10]. However, the new paradigm of health
care in urban environments brings with it the implementation of ICT-intensive devices
with smaller, higher-performance electronic sensors, and systems with greater interactivity
to improve health management. This requires the optimization of interoperable medical
sensor networks with terminals to fasten responses in emergency situations where a rapid
actuation could be crucial, impacting directly on the efficient use of the infrastructure and
reducing costs related to clinical environments.
In this scenario, where many prototypes of m-health systems based on the Body
Area Network (BAN) are becoming popular, the main drawback these systems face in the
smart living approach is the lack of interoperability among medical devices [11], which
could promote a homogeneous e-health ecosystem [12]. Furthermore, the current situation
presents several challenges associated to the certification and regulation of these devices
and their software to be interoperable [13]. Clinical applications must adapt to a paradigm
in which results are repeatable and reliable. In this context, standards are crucial to
the advancement of smart cities and smart living to help smooth the implementation of
innovative technologies and deliver a reliable framework for practitioners. In addition, the
interfaces must follow these standards to be integrated in a hospital infrastructure without
many changes. Thus, the development and maintenance processes must be agile [14].
ISO/IEEE 11073 “Health informatics—Point-of-care medical device and Personal
Health Device (PHD) communication standards” is a family of standards that enable
communication between medical, healthcare, and wellness devices with external computer
systems. These standards are based on a single framework, ISO/IEEE 11073-20601, and
several device specializations (Figure 1). They provide automatic and detailed electronic
data capture of client-related and vital signs information and of device operational data.
The IEEE 11073 Personal Health Devices (PHD) initiative was designed to accommodate
the capabilities of low-power embedded devices and very low-power wireless technologies.
Additionally, the use of transport technologies was designed to adapt them to this interface,
resulting in the base protocol IEEE 11073-20601. Bluetooth, USB, ZigBee, and NFC each
developed a standard (most recently NFC), and these are profiled by the Continua Alliance
guidelines to apply these standards and guarantee interoperability between medical devices.
This methodology establishes a product certification program, as well as collaboration with
government regulatory agencies.

1.1. Continua Health Alliance Design Guidelines

Continua Health Alliance’s design guidelines are focused on the Interface to Personal
Area Network health devices (PAN-IF) and the Interface between Disease Management
Services (DMS), Wireless Area Network (WAN) devices (xHR Senders), and Electronic
Health Record (EHR) devices (xHR Receivers) (xHRN-IF). Continua follows the Open
Systems Interconnection (OSI) model to establish the protocols to use for interoperability in
the PAN-IF. The protocols used in the different OSI layers can be seen in Figure 1 [15].
Continua has constrained the lower-level protocol standards for Bluetooth and USB
communications with the use of specific profiles for health devices. For the higher layers,
Continua uses the ISO/IEEE 11073 standards family, concretely the IEEE 11073-20601-
Optimized Exchange Protocol and ISO/IEEE 11073-104xx-device specializations to provide
application-level interoperability.
The ISO/IEEE 11073 series of standards is based on an object-oriented system manage-
ment paradigm. The system model is divided into three principal components: the domain
information model (DIM), the service model, and the communication model.
Sensors 2022,22,
Sensors2022, 22,5648
x FOR PEER REVIEW 33 of
of 12
13

Figure1.1. Organization
Figure Organizationand
anddependencies
dependenciesofofthe protocols
the established
protocols by by
established the the
Continua Health
Continua Alli‐
Health
ance, over
Alliance, thethe
over Open Systems
Open Interconnection
Systems (OSI)
Interconnection model.
(OSI) model.

1.1. Continua
The DIMHealth Alliance
describes Design
an agent asGuidelines
a set of objects with attributes, which represent the
behavior of the data that can
Continua Health Alliance’s designbe sent toguidelines
the manager. The communication
are focused on the Interface between the
to Personal
agent and the manager is defined by the ISO/IEEE 11073-20601
Area Network health devices (PAN‐IF) and the Interface between Disease Management standard. The service
model
Servicesdefines
(DMS), theWireless
messagesArea that Network
an agent and (WAN) a manager
devicesinterchange following
(xHR Senders), Abstract
and Electronic
Syntax
HealthNotation
Record (EHR)One (ASN.1).
devices (xHRThese messages(xHRN‐IF).
Receivers) are encoded in Medical
Continua Device
follows the Encoding
Open Sys‐
Rules (MDER), described in the ISO/IEEE 11073-20101 standard. The
tems Interconnection (OSI) model to establish the protocols to use for interoperability in communication
model supports
the PAN‐IF. Thethe topology
protocols of one
used in theor more agents
different OSIcommunicating
layers can be seen overin logical
Figure point-to-
1 [15].
point Continua
connections to a single manager. A connection state machine,
has constrained the lower‐level protocol standards for Bluetooth specified in ISO/IEEE
and USB
11073-20601,
communications defines
withthethe
system
use ofbehavior for each for
specific profiles connection.
health devices. For the higher layers,
Aboveuses
Continua the the
IEEE 11073-20601
ISO/IEEE 11073exchange
standardsprotocol, are the device
family, concretely specializations
the IEEE 11073‐20601‐Op‐ that
describe specific details about how a type of agent works and the objects
timized Exchange Protocol and ISO/IEEE 11073‐104xx‐device specializations to provide and attributes that
itapplication‐level
has. For the system implemented in this paper, the pulse oximeter (IEEE 11073-10404)
interoperability.
and the recently basic ECG (IEEE 11073-10406) specializations were used.
The ISO/IEEE 11073 series of standards is based on an object‐oriented system man‐
agement paradigm. The system model is divided into three principal components: the
1.2. Our Contribution
domain information model (DIM), the service model, and the communication model.
The aim of this paper is to investigate the feasibility of the SpO2 and ECG medical
The DIM describes an agent as a set of objects with attributes, which represent the
devices, following the Continua guidelines [16], for increasing the adherence of patients
behavior of the data that can be sent to the manager. The communication between the
to a cardiac rehabilitation treatment. Cardiac rehabilitation patients have cardiovascular
agent and the manager is defined by the ISO/IEEE 11073‐20601 standard. The service
problems. Vital signs such as heart rate (HR) and oxygen saturation are used as predictors
model defines the messages that an agent and a manager interchange following Abstract
of recovery after surgery or in patients with coronary artery disease (CAD) who underwent
Syntax Notation One (ASN.1). These messages are encoded in Medical Device Encoding
rehabilitation [17,18]. When they are being evaluated or undergo physical training, it is
Rules (MDER), described in the ISO/IEEE 11073‐20101 standard. The communication
essential to monitor their blood pressure and heart rate (HR) both at rest and the increase
model supports the topology of one or more agents communicating over logical point‐to‐
during the exercise they are doing, measuring baseline blood pressure and heart rate and
point connections to a single manager. A connection state machine, specified in ISO/IEEE
peak effort. In addition, SpO2 is measured because if it drops below 90%, training must be
11073‐20601,
stopped. defines
Therefore, thethe systemof
measure behavior
SpO2 and forHReach
is connection.
mandatory for any patient undergoing
Above the IEEE 11073‐20601 exchange
cardiac rehabilitation treatment or high-risk patients [19]. protocol, are
Inthe
the device specializations
next sections, that
we describe
describe
our system specific details consisting
architecture about howof a type of agent works
two modules. and the objects
The hardware module andincludes
attributes
a
description of the signal acquisition, processing, and transmission, corresponding11073‐
that it has. For the system implemented in this paper, the pulse oximeter (IEEE with
10404)
the IEEEand the recently
agent-node thatbasic ECG (IEEE
transmits personal 11073‐10406)
health data.specializations
The software weremodule used.
includes a
description of an Android application that receives processed information, corresponding
1.2. Our
with the Contribution
IEEE manager-node that receives personal health data. Finally, results of first
The
clinical aimaofdiscussion,
trials, this paperand
is toconclusions
investigateare
thepresented.
feasibility of the SpO2 and ECG medical
devices, following the Continua guidelines [16], for increasing the adherence of patients
to a cardiac rehabilitation treatment. Cardiac rehabilitation patients have cardiovascular
problems. Vital signs such as heart rate (HR) and oxygen saturation are used as predictors
Sensors 2022, 22, 5648 4 of 12

2. Materials and Methods

2.1. Requirements
Medical Requirements
SpO2 and ECG data collection are required in order to evaluate the adherence of
patients to a cardiac rehabilitation treatment. Analysis of these biosignals may produce
alarms to warn about emergency events. For SpO2, the oxygen saturation level must be
higher than a minimum safety threshold (90–95% [18–20]). If the level decreases under this
threshold, a warning must be generated.
For ECG, lead II was registered as it is the best ECG lead in order to record heart
rhythm and heart rate [21]. The lead II has the configuration: II (RA—Right Arm; LA—Left
Arm; and RL—Right Leg). Pediatric ECG electrodes, round fabric, 40 mm diameter, wet gel,
snap electrode, silver chloride (Ag/AgCl) sensor, repositionable. Our clinicians wanted
to track the changes in the heart rate (HR) such as bradycardia (HR < 60 beats per minute
(bpm)) and tachycardia (>100 bpm) [22]. In the case of such an event, we must acquire and
store at least 10 minutes of signal; this corresponds to 5 min before and 5 min after the
warning event.
Technical Requirements
Real-time and wear specifications have resulted in a system with fixed latency, power sup-
ply, and space limitations. Medical requirements have led to the inclusion of storage capacities.
In this work, we proposed the use of a DSP as the processing element. As the architec-
ture of a DSP is optimized specifically for digital signal processing, it is the best solution
regarding the performance and cost in relation to others, such as FPGAs or microprocessors.
For clinical use, the signals should be digitalized with at least 12 bits of resolution.
Moreover, several algorithms should be included to remove noise, extract parameters from
physiological data, and identify warning events.

2.2. System Description

Our system consists of two different modules corresponding with the two IEEE
configuration modes over a Bluetooth connection using the Health Device Profile (HDP)
(Figure 2). These modules were developed following a Service Oriented Architecture (SOA).
Connectivity, semantic interoperability, and security are ensured using ISO/IEEE 11073
Sensors 2022, 22, x FOR PEER REVIEW 5 of 13
over Continua-certified communication modules. Material and methods are described
below for each module.

Figure
Figure 2. IEEE agent
2. IEEE agentfor
forphysiological
physiologicaldata
data acquisition,
acquisition, processing,
processing, andand transmission
transmission (left);(left); and
and IEEE
IEEE manager smartphone application for patient empowerment
manager smartphone application for patient empowerment (right). (right).

2.2.1.
2.2.1. IEEE
IEEE Agent
Agent
HDP defines
HDP defines a transmitter of medical data as aa source;
source; ISO/IEEE-11073-20601 uses
ISO/IEEE‐11073‐20601 uses
the term IEEE agent for this same node. In our framework, this agent corresponds to
the term IEEE agent for this same node. In our framework, this agent corresponds to the the
acquisition, processing,
acquisition, processing, and transmission hardware (Figure 3) [23].
 Figure 2. IEEE agent for physiological data acquisition, processing, and transmission (left); and
IEEE manager smartphone application for patient empowerment (right).

2.2.1. IEEE Agent

HDP defines a transmitter of medical data as a source; ISO/IEEE‐11073‐20601 uses
Sensors 2022, 22, 5648 5 of 12
the term IEEE agent for this same node. In our framework, this agent corresponds to the
acquisition, processing, and transmission hardware (Figure 3) [23].

Figure3.3.Architecture
Figure Architectureofofthe
theIEEE
IEEEagent
agentininour
ourframework
frameworkshowing
showingthe
theacquisition,
acquisition,processing,
processing,and
and
transmission hardware.
transmission hardware.

DataAcquisition.
Data Acquisition. Our Our framework
framework isiscomposedcomposedofofthe theTMDXMDKPO8328
TMDXMDKPO8328and and
ADS1298RFE boards from Texas Instruments (TI) for
ADS1298RFE boards from Texas Instruments (TI) for SpO2 and ECG data acquisition. SpO2 and ECG data acquisition.
Both,the
Both, thepulse
pulseoximeter
oximeterand andthetheECGECGimplementation,
implementation,are areprovided
providedfor forthe
theTMS320C5515
TMS320C5515
DSPasaspart
DSP partofof
thethe Medical
Medical Development
Development KitKit (MDK)
(MDK) intended
intended for for evaluation
evaluation andand devel‐
develop-
opment
ment purposes.
purposes.
TheSpO2
The SpO2acquisition
acquisitionboardboardincludes
includesaalow-power
low‐poweranalog-to-digital
analog‐to‐digitalconverter
converter(ADC),
(ADC),
with 16 bits of resolution and 500 kilo samples per second
with 16 bits of resolution and 500 kilo samples per second (ksps) sampling rate. After (ksps) sampling rate. After
acquiring
acquiring the
the absorption
absorption value
value of of
thethe
redred
(R)(R)andand infrared
infrared (IR)(IR)
raysrays for blood,
for the the blood, the sig‐
the signals
nals
are are conditioned
conditioned by twoby two stages
stages of an amplifier.
of an amplifier.
InInthe
thecase
caseofofthe
theECG,
ECG,the theacquisition
acquisitionboard boardincludes
includesa alow-power
low‐powerADC, ADC,working
workingatat
500
500spsspsdata
datarate
ratewith
with2424bitbitdata
dataresolution.
resolution.AnalogAnalogsignals
signalspasspassthrough
througha adefibrillator
defibrillator
protection
protection(DP) (DP)circuit
circuittotoprotect
protectthe therest
restofofthe
thesystem.
system.The Theboard
boardderives
derives8 8out outofof1212ECG
ECG
leads using differential amplifiers. These leads are filtered by a low pass
leads using differential amplifiers. These leads are filtered by a low pass filter (LPF), which filter (LPF), which
provides
providesanti-aliasing
anti‐aliasingand andremoves
removesfrequencies
frequenciesabove above150150HzHzwithwith6060dBdBattenuation
attenuationand and
digitized.
digitized.Due Duetotoour ourmedical
medicalrequirements
requirements(see (seeSection
Section2.1),
2.1),only
onlylead
leadIIIIisisanalyzed
analyzedby by
the
thesystem.
system.
Both
Bothboards
boardsinterface
interfacewithwiththe theDSP
DSPmodule
moduleusing usingthetheserial
serialperipheral
peripheralinterface
interface(SPI)
(SPI)
and the Inter-Integrated Circuit (I2C)
and the Inter‐Integrated Circuit (I2C) buses. buses.
Bluetooth Communications. Communication between the agent and the manager
is implemented using the ISO/IEEE-11073-20601 optimized exchange protocol over a
Bluetooth stack with HDP. This protocol provides plug-and-play interoperability by using
the IEEE-11073-20601.
According to the established requirements for this proposal, our system includes a
Bluegiga WT12-A Bluetooth module. The Bluegiga WT12 is a fully integrated Bluetooth
2.1 + Enhanced Data Rate (EDR), class 2 module combining an antenna, Bluetooth radio,
and an on-board iWRAP Bluetooth stack including the iWRAP 5.0.0 firmware, which
provides an easy way to configure Bluetooth to work with HDP. This version of iWRAP
includes the implementation of several IEEE agents, which leads to a transparent connection
management and data packaging. Thus, allowing a lower-end host processor to be used
and faster time to market. The agents implemented in this firmware were tested to pass
Continua v.1.5 test cases, which also allows avoiding the certification process, thus, saving
time. We used the standard configuration for a pulse oximeter agent implemented in the
iWRAP firmware.
An example of a connection establishment sequence between a manager and a SpO2
agent implemented in the iWRAP firmware is shown in Figure 4.
 Continua v.1.5 test cases, which also allows avoiding the certification process, thus,
time. We used the standard configuration for a pulse oximeter agent implemented
iWRAP firmware.
An example of a connection establishment sequence between a manager and
Sensors 2022, 22, 5648 6 of 12
agent implemented in the iWRAP firmware is shown in Figure 4.

Figure 4. Example of connection establishment sequence between a manager and a SpO2 IEEE
Figure 4.agent
Example of connection
implemented establishment
in the iWRAP firmware. Thissequence between
figure shows a manager
the syntax andofathe
and the order SpO2 IEE
implemented
commandsin the iWRAP firmware. This figure shows the syntax and the order of the com
exchanged.
exchanged.
Since the ECG agent is not supported by iWRAP, we developed an ECG agent with
extended configuration in our system. It was explained in the ECG Data Processing subsection.
Since Processing
the ECG System.
agent isIncoming
not supported by iWRAP,
data are processed by the we developed
low-power an ECG agen
TMS320C5515
DSP Medical Development Kit (MDK) evaluation module (EVM) from
extended configuration in our system. It was explained in the ECG Data Processin TI. The implementa-
tion of the EVM only allows connecting one front-end-board at a time.
section. Our software implementations are based on the examples provided by TI. We also
Processing
included RealSystem.
Time Clock Incoming data are
(RTC) and External processed
Memory Interfaceby the peripherals
(EMIF) low‐power fromTMS32
the C55xDevelopment
DSP Medical Chip Support Libraries (CSL). CSLs
Kit (MDK) provide an application
evaluation module (EVM) programming
frominterface
TI. The imp
(API) that shortens development time by providing standardization and portability.
tation of the EVM only allows connecting one front‐end‐board at a time.
In the system, data are received from the acquisition boards through the SPI bus.
Our
Theysoftware implementations
are processed are
to obtain the relevant based on
information the exampleswaveform,
(plethysmographic provided SpO2by TI. W
value, ECG waveform, and HR). Then, data are encapsulated following
included Real Time Clock (RTC) and External Memory Interface (EMIF) periphera the corresponding
IEEE device specialization standard. The most relevant algorithms and new characteristics
included in our system are described in the following subsections.
SpO2 Data Processing. To acquire the SpO2 data, the DSP reads alternatively the R and
IR values from SPI every 1 ms.
A first-order IIR filter is used to remove the DC component. Then, a Hamming
windowed 51st order low-pass FIR filter removes unwanted frequencies above 10 Hz.
The SpO2 percentage value is obtained from the ratio of the Root Mean Square (RMS)
of the R and IR signals using a look-up table (LUT) every three heart beats. The pulse rate
is detected using the filtered IR signal. The values of samples from the plethysmographic
Sensors 2022, 22, 5648 7 of 12

waveform are calculated through a square root algorithm which provides a proportional
value of the peak-to-peak signal. To provide a timestamp for the measure, we used the CSL
to implement the RTC. Further details regarding the actual ECG implementation can be
found in the TI Application Report SPRAB37A—June 2010 [24].
Once the SpO2 percentage value is calculated and its timestamp is obtained, the SpO2
value is compared with the threshold. Hence, if the mean SpO2 value is lower than 95%,
the system generates an alarm. To avoid false alarms, we calculated the mean of the last
10 values of the SpO2 percentage. Moreover, the first four SpO2 values for each monitoring
session were discarded to allow the signal to become steady.
In that moment, and if the connection between the agent and the manager is already
established, the system sends the last value of SpO2 and the timestamp through the UART
peripheral to the Bluetooth, following the iWRAP syntax (Table 1).

Table 1. Example of iWrap syntax to send pulse oximeter data.

Command Description
AGENT UPDATE
Oxygen saturation is 97%
FFFF 1 A4C 97E0
AGENT UPDATE
Time stamp of the measurement
FFFF 1 990 2013-01-30T20:05:2678
AGENT UPDATE
Pulse rate is 74 bpm
FFFF A A4C 74E0

The UART peripheral is configured to work to 115,200 bauds, 8 data bits, no parity,
and one stop bit. When the alarm has been sent, the system waits for the ACK from
the manager.
Previously, to send an alarm, the agent and manager had to establish the connection
following the state machine defined in IEEE 11073-20601. The state machine defines the
states and substates an agent and manager pair pass through. The agent implementation in
the iWRAP 5.0, corresponding to the standard SpO2, manages it.
A successful IEEE association is indicated to the DSP by an event message from the
Bluetooth chip, and the agent will pass to the operating state. The message contains the
channel and configuration identifiers of the IEEE configuration that the manager accepted.
Then, the agent is ready to send data to the manager. If this state is not reached, the
association with the manager has failed and the agent will pass to the unassociated state.
ECG Data Processing. An interruption is generated every 2 ms in order to read ECG
data from the SPI bus. The DSP digitizes nine channels with 24 bits of resolution. The
first channel is the lead-off status. The eight remaining channels are passed through the
same IIR filter as the SpO2 signal, to remove the DC component. Removing unwanted
frequencies is achieved using a multi-band-pass filter, which consists of a FIR filter and
notch filter at 50/60 Hz. Once the data is filtered, the DSP computes the remaining four
leads. Then, a QRS detection algorithm is used to detect the R-peak and compute the
HR. The QRS algorithm is applied over the lead II and consists of a variable threshold
method [25]. The HR is calculated when five R-peaks have been detected. Further details
regarding the actual ECG implementation can be found in the TI Application Report
SPRAB36A—July 2009 [26].
To accomplish the medical requirements, data from the previous five minutes are
stored in an external memory to assess for potential current events. For this purpose, we
used the 64 Mb mobile–SDRAM memory (MT48H8M16LF by Micron). We checked the
HR value to store the data; if it was non-zero, we stored it, as well as the lead status, the
timestamp, the previous five R-R intervals, and the ECG sample related to the lead II.
Based on the analysis of these data, the system decides if constants are out of range
and an alarm must be generated. Given this situation, and given that devices are not
already paired, the communication is established in a similar way as in the case of the SpO2
system, except that the DSP has to implement the state machine for the connection for the
 timestamp, the previous five R‐R intervals, and the ECG sample related to the lead II.
Based on the analysis of these data, the system decides if constants are out of ran
and an alarm must be generated. Given this situation, and given that devices are not
ready paired, the communication is established in a similar way as in the case of the Sp
Sensors 2022, 22, 5648 8 of 12
system, except that the DSP has to implement the state machine for the connection for
ECG. In this case, two data channels are opened, one reliable (for the main informati
and one streaming (for the waveform samples). At that time, the system reads data fr
ECG. In this case, two data channels are opened, one reliable (for the main information)
the SDRAM (from 5 (for
and one streaming minthebefore the samples).
waveform alarm if At it is produced
that later than
time, the system reads 5 min,
data fromor from
startthe
of SDRAM
the monitoring
(from 5 min session
before if
theitalarm
is produced beforelater
if it is produced a 5 than
min5monitoring session). T
min, or from the
start of the monitoring session if it is produced before a 5 min
system reads all the values between two stored HR values. Then, it combines the dmonitoring session). The
system reads all the values between two stored HR values. Then, it combines the data
packet to accomplish the 11073‐10406 standard. The system implements the object clas
packet to accomplish the 11073-10406 standard. The system implements the object classes
withwith
mandatory
mandatoryand recommended
and recommended attributes
attributes includedincluded in the specialization
in the specialization (Figure 5). One(Figure
One data
datapacket
packet is sent with HR, lead‐off status, five R‐R intervals
is sent with HR, lead-off status, five R-R intervals in ms, and the timestamp in ms, and
throughthrough
timestamp the reliablethe
data channel.data channel.
reliable

Figure 5. Representation of the different object classes in the ISO/IEEE 11073‐10406 Basic ECG dev
Figure 5. Representation of the different object classes in the ISO/IEEE 11073-10406 Basic ECG
specialization.
device specialization.

After
After that,
that, packetswith
packets with 20
20samples from
samples leadlead
from II areIIsent
arethrough the streaming
sent through data
the streaming d
channel until there are 20 or fewer samples to be sent. When this happens, the DSP reads
channel until there are 20 or fewer samples to be sent. When this happens, the DSP re
from the SDRAM as before and sends a new data packet. In the same way as in the SpO2
fromsystem,
the SDRAM
data are as before
sent and
through thesends a new data
UART (similarly packet. While
configured). In thethe
same
DSPway as in the Sp
is sending
system, data are sent through the UART (similarly
packets, it is acquiring and storing new data. configured). While the DSP is send
packets, it is acquiring and storing new data.
2.2.2. IEEE Manager
In order to receive and visualize the processed data, a smartphone application was
2.2.2.developed.
IEEE Manager
The MORFEO OpenHealth Project developed the HDP/MCAP implementation
In the
on order to receive
official and visualize
Linux Bluetooth the processed
protocol stack, BlueZ 4.77 data, a smartphone
[27]. This implementationapplication
was w
integrated by Google in Android API 27 and newer versions to
developed. The MORFEO OpenHealth Project developed the HDP/MCAP implemen support HDP through the
Bluetooth Health class.
tion on the official Linux Bluetooth protocol stack, BlueZ 4.77 [27]. This implementat
We used this class to implement an IEEE manager that is able to understand our SpO2
was integrated
and ECG databy Google
through in Android API 27 and newer versions to support HDP throu
HDP.
the Bluetooth
When theHealth
HDP class.
connection is established, the manager will wait for the agent applica-
tion layer (Continua) association request. In this association, the agent and the manager
establish the configuration and the manager starts receiving data.
The application consists of a main menu that leads to two different modes, one for
SpO2 and the other for ECG. In the SpO2 mode, the application provides a visualization of
the connection status, the SpO2 level, and the date and hour of the sent packet. In the same
way, in the ECG mode the application provides the visualization of the connection status,
the HR, the date and hour of the acquisition and the ECG waveform. Figure 6 shows an
example of the ECG application where the waveform from an ECG waveform generator
is displayed. Although the evaluation versions included two screens with ECG and Sp02
data, the clinicians’ recommendation was to simplify with the ECG reading only.
 of the connection status, the SpO2 level, and the date and hour of the sent pac
same way, in the ECG mode the application provides the visualization of the c
status, the HR, the date and hour of the acquisition and the ECG waveform
shows an example of the ECG application where the waveform from an ECG
Sensors 2022, 22, 5648 9 of 12
generator is displayed. Although the evaluation versions included two screens
and Sp02 data, the clinicians’ recommendation was to simplify with the ECG rea

Figure 6. Graphical user interface of the ECG Android application, in which the Blueto
Figure 6. Graphical user interface of the ECG Android application, in which the Bluetooth connection
tion status, the date and hour, heart rate, and ECG waveform are shown.
status, the date and hour, heart rate, and ECG waveform are shown.

2.3. Validation
2.3.Methodology
Validation Methodology
The system The was system
evaluatedwasin aevaluated
sample composed of 10composed
in a sample healthy volunteers (mean volunteers
of 10 healthy age
29.8 ± 7.1 y/o; 5 females) under supervision of an expert cardiologist from the Hospital
29.8 ± 7.1 y/o; 5 females) under supervision of an expert cardiologist from th
Universitario de Fuenlabrada (HUF), Madrid, Spain. The research protocol was approved
Universitario de Fuenlabrada (HUF), Madrid, Spain. The research protocol was
by the local Institutional Review Board and written informed consent was acquired from
by the
all participants. local
The Institutional
evaluation Review
consisted Board and
of measuring written
the SpO2 andinformed
ECG signalsconsent
sittingwas acqu
all participants. The evaluation consisted of measuring the
and at rest, before and after exercising for 15 min. A 3-lead configuration was followed SpO2 andforECG sign
and at rest,
the ECG recording with before and after
one electrode exercising
adjacent for 15bone
each clavicle min.onAthe
3‐lead
upperconfiguration
chest (i.e., wa
the left armfor the ECG recording with one electrode adjacent each clavicle bone on the u
electrode (L lead) and the right arm electrode (R lead)), and a third electrode
adjacent the(i.e.,
patient’s lower
the left armleft abdomen(L(i.e.,
electrode theand
lead) left leg
theelectrode
right arm (Nelectrode
lead)). (R lead)), and a
Physiological measurements were assessed and compared against commercial
trode adjacent the patient’s lower left abdomen (i.e., the left leg electrode devices. (N lea
Warning events generated were monitored and received in a local server simulating the
Physiological measurements were assessed and compared against comm
telemonitoring system in the hospital. To check the accuracy of the SpO2 measurement
vices. Warning events generated were monitored and received in a local server s
obtained with our system, we compared it with a commercial Nonin Onyx 9500 device. In
thethe
the same way, telemonitoring
accuracy of thesystem in thewas
ECG signal hospital.
assessed Toagainst
check athe accuracy
General of the
Electric SpO2 me
(GE)
MAC 1200 obtained
ST. with our system, we compared it with a commercial Nonin Onyx 9500
the same way, the accuracy of the ECG signal was assessed against a General El
3. Results MAC 1200 ST.
3.1. Continua Guidelines Compliance
The validation
3. Resultsof the accomplishment of Continua Guidelines was carried out in a
different way for each system.
3.1. Continua Guidelines Compliance
The SpO2 system accomplishes the Continua guidelines due to the use of the agent
implementation The validation
in the of the accomplishment
iWRAP5 firmware ofThe
on the WT12 chip. Continua
firmwareGuidelines
iWRAP5 waswas carrie
designed anddifferent way
tested to meetforthe
each system. of Continua version 1.5.
requirements
As the iWRAP 5.0 firmware does not implement an ECG agent, the validation of the
accomplishment of Continua guidelines for our ECG system was checked by following
the Design Guidelines. Moreover, we made more validations using our own manager
application. Further validations will include performing the Continua certification process.
The Android application can understand both IEEE agents implemented. This appli-
cation is configured to be used in a very simple way.

3.2. Functional Validation

As aforementioned, we compared the measured data with a commercial Nonin Onyx
9500 device (Sp02b ) to check the accuracy of the SpO2 measurement obtained with our
system (Sp02a ). Both the error in accuracy (mean Sp02a –Sp02b difference) and the error in
Sensors 2022, 22, 5648 10 of 12

precision (SD of differences) remained below 1%. This obtained error is within the error
allowed according to medical requirements from the hospital.
In the same way, the accuracy of the ECG signal was compared against a General
Electric (GE) MAC 1200 ST. Our expert physician did not find any relevant differences in
the recorded ECG signal. The results of the physician’s evaluations were accepted for both
SpO2 and ECG signals. Additionally, our system acquired and processed data from an ECG
simulator at different HRs to prove that HR was properly identified and that warnings were
generated when the HR was under or above the thresholds for bradycardia and tachycardia,
respectively. Therefore, using devices and applications based on standards, as the one
reported in this paper, would facilitate the use and integration of different measuring and
mobile devices.

4. Discussion
In this paper, we described the implementation of two IEEE agents for SpO2 and ECG
data acquisition, respectively, and an IEEE manager for validation. We developed both IEEE
agents over Bluetooth HDP following the Continua guidelines. The proposed architecture
is modular, meeting the wireless standards and allowing for real-time processing. Unlike
application frameworks that require structural changes and create complex interdependen-
cies, our model is service-oriented. This solution maintains a reliable, resilient, integrated,
interoperable, and cost-effective system for personalized healthcare, while incorporating
ubiquity. The modular design adopted is intended to provide a future-proofed system,
whose functionality may be upgraded by modifying the hardware or software. The system
features include automatic or on-demand data recording mode and warning state detection,
such as abnormal oxygen levels, bradycardia, or tachycardia. Nevertheless, because of the
modular nature of our implementation and the flexibility of the standard, new algorithms
such as ventricular extrasystole detection algorithms could be implemented in the IEEE
agents, and the warning events generated by such new algorithms could be sent over the
same protocol implementation.
Our expert physician checked the reliability of SpO2 and ECG signals captured by the
system, and reported that both SpO2 values and ECG tracings provided by our system are
reliable and acquired data are feasible for clinical use as this type of information is currently
required in routine clinical practice. In this sense, we hypothesize that this system could
be especially useful in cardiac rehabilitation programs, where a hard control of the health
of the patient is needed to improve adherence. Actually, it has been shown that cardiac
rehabilitation can reduce the morbimortality by practically 50% in patients with heart
disease. Nevertheless, participation and adherence of patients to cardiac rehabilitation
and its associated treatment remains low, especially among the elderly, women, and
low socioeconomic profile patients [28]. In this scenario, the use of smartphone-based
applications has proved an improvement in patient adherence to the treatment; for instance,
Yudi et al. [29] showed that the use of smartphone-based rehabilitation led to a significantly
higher uptake (94% vs. 68%, p < 0.05) and completion (80% vs. 47%, p < 0.05) of the program.
Using devices and applications based on standards, such as the one reported in this paper,
would facilitate the use and integration of different measuring and mobile devices.

5. Conclusions
In this paper, we described the implementation of two IEEE agents for SpO2 and ECG
data acquisition, and an IEEE manager for validation. We developed both IEEE agents over
Bluetooth HDP following the Continua guidelines. In particular, as far as we know, this
implementation of the extended ECG agent is one of the first available for devices oriented
to m-Health systems. These implementations offer a standardization of medical devices.
The system features include automatic or on-demand data recording mode and warn-
ing state detection, such as abnormal oxygen levels or changes in the HR such as brady-
cardia and tachycardia. Our expert physician checked the reliability of SpO2 and ECG
signals digitized by the system. Due to this, this system can be useful in rehabilitation
Sensors 2022, 22, 5648 11 of 12

programs, when a hard control of the health of the patient is needed to improve adherence
in cardiac rehabilitation.
Future work may improve the current system implementation for concurrent acqui-
sition from different boards. For example, pulse oximeters are often used for estimating
heart rate, obtaining the heart rate from the plethysmography signal; thus, the SpO2 agent
could be extended to include this parameter; thus, we would have several sources to
account for the HR. Additionally, new devices could include more personal health devices
such as weighing scales, respiration rate, insulin pumps, medication monitors, etc. In
particular, as far as we know, this implementation of the extended ECG agent is one of
the first available implementations for devices oriented to m-Health systems. Finally, the
objective of this work was the implementation of ISO/IEEE 11073 PHD SpO2 and ECG
device specializations over Bluetooth HDP following the Health Care Profile. A qualitative
and quantitative evaluation could be interesting in the methodology for our future work
following recommendations from other authors [30–32].

Author Contributions: Conceptualization, A.C.-H., A.T.-C., C.R.-S., M.L. and S.B.; methodology,
A.C.-H., A.T.-C., C.R.-S. and S.B.; software, A.C.-H., A.T.-C. and C.R.-S.; validation, A.C.-H., A.T.-C.,
C.R.-S., M.L. and S.B.; formal analysis, A.C.-H., A.T.-C. and C.R.-S.; investigation, A.C.-H., A.T.-C.,
C.R.-S. and S.B.; resources, J.A.H.-T. and S.B.; data curation, A.C.-H. and A.T.-C.; writing—original
draft preparation, A.C.-H., A.T.-C., C.R.-S., M.L. and S.B.; writing—review and editing, A.C.-H.,
A.T.-C., C.R.-S., J.A.H.-T., M.L. and S.B.; visualization, A.C.-H., A.T.-C. and C.R.-S.; supervision,
J.A.H.-T. and S.B.; project administration, J.A.H.-T. and S.B.; funding acquisition, J.A.H.-T. and S.B.
All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by TSI-020100-2011-0277 project from The Ministry of Industry,
Energy and Tourism of Spain.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki and approved by the local Institutional Review Board.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Not applicable.
Acknowledgments: We would like to thank the volunteers participating in the validation of the
solutions implemented in this project. We would like to thank Worldnet 21 S.L. for their support as
an industrial partner during the development of this project.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Istepanian, R.S.H.; Laxminaryan, S. UNWIRED, the next generation of wireless and interpretable telemedicine systems-editorial
paper. IEEE Trans. Inform. Technol. Biomed. 2000, 4, 189–194.
2. Kyriacou, E.C.; Pattichis, C.S.; Pattichis, M.S. An overview of recent health care support systems for eEmergency and mHealth
applications. In Proceedings of the 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology
Society, Minneapolis, MN, USA, 3–6 September 2009; pp. 1246–1249.
3. Castellano, N.N.; Gazquez, J.A.; Salvador, R.M.G.; Gracia-Escudero, A.; Fernandez-Ros, M.; Manzano-Agugliaro, F. Design of a
real-time emergency telemedicine system for remote medical diagnosis. Biosyst. Eng. 2015, 138, 23–32. [CrossRef]
4. Rathbone, A.L.; Prescott, J. The use of mobile apps and SMS messaging as physical and mental health interventions: Systematic
review. J. Med. Internet Res. 2017, 19, e7740. [CrossRef]
5. Tomasic, I.; Tomasic, N.; Trobec, R.; Krpan, M.; Kelava, T. Continuous remote monitoring of COPD patients—justification and
explanation of the requirements and a survey of the available technologies. Med. Biol. Eng. Comput. 2018, 56, 547–569. [CrossRef]
6. Santos, D.F.; Almeida, H.O.; Perkusich, A. A personal connected health system for the Internet of Things based on the Constrained
Application Protocol. Comput. Electr. Eng. 2015, 44, 122–136.
7. Dwivedi, A.; Bali, R.K.; James, A.E.; Naguib, R.N.G.; Johnston, D. Merger of knowledge management and information technology
in healthcare: Opportunities and challenges. In Proceedings of the IEEE CCECE2002. Canadian Conference on Electrical and
Computer Engineering. Conference Proceedings (Cat. No.02CH37373), Winnipeg, MB, Canada, 12–15 May 2002; Volume 2,
pp. 1194–1199.
8. Mertz, L. Ultrasound? Fetal Monitoring? Spectrometer? There0 s an App for That!: Biomedical Smart Phone Apps Are Taking
Healthcare by Storm. IEEE Pulse 2012, 3, 16–21. [CrossRef]
Sensors 2022, 22, 5648 12 of 12

9. Rodriguez-Sanchez, M.C.; Torrado-Carvajal, A.; Borromeo, S.; Hernandez-Tamames, J.A.; Luaces, M. Novel Applications for
M-Health and Free Messaging. IEEE Pervasive Comput. 2012, 11, 74–75. [CrossRef]
10. Kuroda, T.; Sasaki, H.; Suenaga, T.; Masuda, Y.; Yasumuro, Y.; Hori, K.; Ohboshi, N.; Takemura, T.; Chihara, K.; Yoshihara, H.
Embedded Ubiquitous Services on Hospital Information Systems. IEEE Trans. Inf. Technol. Biomed. 2012, 16, 1216–1223. [CrossRef]
[PubMed]
11. Sevin, A.; Bayilmis, C.; Kirbas, I. Design and implementation of a new quality of service-aware cross-layer medium access
protocol for wireless body area networks. Comput. Electr. Eng. 2016, 56, 145–156. [CrossRef]
12. Lemlouma, T.; Laborie, S.; Rachedi, A.; Santos, A.; Vasilakos, A.V. Special Issue on Selected Papers from e-Health Pervasive
Wireless Applications and Services 2017. Information 2019, 10, 52.
13. Kumari, A.; Tanwar, S.; Tyagi, S.; Kumar, N. Fog computing for Healthcare 4.0 environment: Opportunities and challenges.
Comput. Electr. Eng. 2018, 72, 1–13. [CrossRef]
14. Rezaei, F.; Hempel, M.; Sharif, H. A survey of recent trends in wireless communication standards, routing protocols, and energy
harvesting techniques in E-health applications. In Wearable Technologies: Concepts, Methodologies, Tools, and Applications; IGI Global:
Hershey, PA, USA, 2018; pp. 1479–1502.
15. American Thoracic Society; American College of Chest Physicians. ATS/ACCP Statement on cardiopulmonary exercise testing.
Am. J. Respir Crit. Care Med. 2003, 167, 211–277. [CrossRef]
16. Carroll, R.; Cnossen, R.; Schnell, M.; Simons, D. Continua: An interoperable personal healthcare ecosystem. IEEE Pervasive
Comput. 2007, 6, 90–94. [CrossRef]
17. Canet, J.; Gallart, L. Predicting postoperative pulmonary complications in the general population. Curr. Opin. Anaesthesiol. 2013,
26, 107–115. [CrossRef] [PubMed]
18. Evrengul, H.; Tanriverdi, H.; Kose, S.; Amasyali, B.; Kilic, A.; Celik, T.; Turhan, H. The relationship between heart rate recovery
and heart rate vari- ability in coronary artery disease. Ann. Noninvasive. Electrocardiol. 2006, 11, 154–162. [CrossRef]
19. Ambrosetti, M.; Abreu, A.; Corrà, U.; Davos, C.H.; Hansen, D.; Frederix, I.; Iliou, M.C.; Pedretti, R.F.E.; Schmid, J.-P.; Vigorito,
C.; et al. Secondary prevention through comprehensive cardiovascular rehabilitation: From knowledge to implementation. 2020
update. A position paper from the Secondary Prevention and Rehabilitation Section of the European Association of Preventive
Cardiology. Eur. J. Prev. Cardiol. 2020, 28, 460–495. [CrossRef] [PubMed]
20. Understanding Pulse Oximetry SpO2 Concepts. In Philips Medical Systems; Global Information Center: Eindhoven, The
Netherlands; Available online: https://www.documents.philips.com/doclib/enc/fetch/586262/586457/Understanding_Pulse_
Oximetry.pdf (accessed on 1 May 2022).
21. Meek, S.; Morris, F. ABC of clinical electrocardiography: Introduction. I—Leads, rate, rhythm, and cardiac axis. BMJ: Br. Med. J.
2002, 324, 415–418. [CrossRef]
22. Hampton, J.R. The ECG Made Easy; Elsevier Health Sciences: Amsterdam, The Netherlands, 2008.
23. Health Device Profile Implementation Guidance Whitepaper Medical Devices Working Group; Bluetooth Special Interest
Group. 2009. Available online: https://www.bluetooth.com/wp-content/uploads/2019/03/HDP-Implementation_WP_V10.pdf
(accessed on 1 May 2022).
24. Markandey, V. Pulse oximeter implementation on the TMS320C5515 DSP medical development kit (MDK). Texas Instruments
Application Report Jun-2010. Available online: https://www.ti.com/lit/an/sprab37a/sprab37a.pdf?ts=1658835461222&ref_url=
https%253A%252F%252Fwww.google.com%252F (accessed on 1 May 2022).
25. Hamilton, P.S.; Tompkins, W. Quantitative investigation of QRS detection rules using the MIT/BIH arrhythmia database. IEEE
Trans. Biomed. Eng. 1986, 12, 1157–1165. [CrossRef] [PubMed]
26. Markandey, V. ECG Implementation on the TMS320C5515 DSP Medical Development Kit (MDK). Texas Instruments Application
Report Jul-2009. Available online: https://www.itfind.or.kr/COMIN/file29190-Pulse%20Oximeter%20Implementation%20on%
20the%20TMS320C5515%20DSP%20Medical%20Development%20Kit%20MDK.pdf (accessed on 1 May 2022).
27. Barrón-González, H.G.; Martínez-Espronceda, M.; Trigo, J.D.; Led, S.; Serrano, L. Recent Advances in mHealth: An Update to
Personal Health Device Interoperability Based on ISO/IEEE11073. In Mobile Health; Springer: Cham, Switzerland, 2015; pp.
375–397.
28. García-Bravo, S.; Cuesta-Gómez, A.; Campuzano-Ruiz, R.; López-Navas, M.J.; Domínguez-Paniagua, J.; Araújo-Narváez, A.;
Barreñada-Copete, E.; García-Bravo, C.; Flórez-García, M.T.; Botas-Rodríguez, J.; et al. Virtual reality and video games in cardiac
rehabilitation programs. A systematic review. Disabil. Rehabil. 2021, 43, 448–457. [CrossRef]
29. Yudi, M.B.; Clark, D.J.; Tsang, D.; Jelinek, M.; Kalten, K.; Joshi, S.; Phan, K.; Nasis, A.; Amerena, J.; Arunothayaraj, S. SMARTphone-
based, early cardiac REHABilitation in patients with acute coronary syndromes [SMART-REHAB Trial]: A randomized controlled
trial protocol. BMC Cardiovasc. Disord. 2016, 16, 170. [CrossRef]
30. Bland, J.M.; Altman, D.G. Measuring agreement in method comparison studies. Stat. Methods Med. Res. 1999, 8, 135–160.
[CrossRef]
31. Guber, A.; Gali, E.S.; Sarah, K.; David, S. Wrist-Sensor Pulse Oximeter Enables Prolonged Patient Monitoring in Chronic Lung
Diseases. J. Med. Syst. 2019, 43, 230. [CrossRef] [PubMed]
32. Sarmento, A.; Carlo, V.; Stefania, P.; Carolina, L.; Alessandra, S.; Flavia, N.; Massimo, M.; Leonardi, A.; Ossola, D.; Rigoni, A.; et al.
Qualitative and Quantitative Evaluation of a New Wearable Device for ECG and Respiratory Holter Monitoring. Int. J. Cardiol.
2018, 272, 231–237. [CrossRef] [PubMed]