Smart Monitoring of User’s Health at Home: Performance

Evaluation and Signal Processing of a Wearable Sensor for the

Measurement of Heart Rate and Breathing Rate

Sara Casaccia, Filippo Pietroni, Andrea Calvaresi, Gian Marco Revel and Lorenzo Scalise

Dipartimento di Ingegneria Industriale e Scienze Matematiche, Università Politecnica delle Marche, Ancona, Italy

Keywords: Physiological Parameters, Heart Rate, Breathing Rate, Wearable Sensor, Signal Processing.

Abstract: Nowadays, the monitoring of users’ health status is possible by means of smart sensing devices at low-cost

and with high measuring capabilities. Wearable devices are able to acquire multiple physiological and

physical waveforms and are equipped with on-board algorithms to process these signals and extract the

required quantities. However, the performance of such processing techniques should be evaluated and

compared to different approaches, e.g. processing of the raw waveforms acquired. In this paper, the authors

have performed a metrological characterization of a commercial wearable monitoring device for the

continuous acquisition of physiological quantities (e.g. Heart Rate - HR and Breathing Rate - BR) and raw

waveforms (e.g. Electrocardiogram - ECG). The aim of this work is to compare the performance of the on-

board processing algorithms for the calculation of HR and BR with a novel approach applied to the raw

signals. Results show that the HR values provided by the device are accurate enough (±2.1 and ±2.8 bpm in

static and dynamic tests), without the need of additional processing. On the contrary, the implementation of

the dedicated processing technique for breathing waveform allows to compute accurate BR values (±2.1 bpm

with respect to standard equipment).

1 INTRODUCTION

The improvement of people’s safety and the control

of their health conditions represent the primary

challenges about Smart Home (Sixsmith A. and

Sixsmith J., 2000) and Smart Cities (Revel et al.,

2014). The Italian project Health@Home: Smart

Communities for citizens’ wellness (H@H) aims to

reach such goal through the integration of domotic

and biomedical devices in Smart residential

environments (Abascal, 2004), (Demiris et al., 2006).

Nowadays, acquiring physical (Catal et al., 2015),

(Bayat et al., 2014), (Abdallah et al., 2015) and

physiological (Pantelopoulos and Bourbakis, 2010),

(Parvaneh et al., 2014), (van Andel et al., 2015)

signals from a person is a quite easy task, thanks to

the rapid growth of low-cost Information and

Communications Technologies (ICT). Wearable

sensors are able to measure even more physiological

waveforms (e.g. ECG) in real-time and are equipped

with on-board algorithms, in order to extract the most

interesting values (e.g. HR, BR, etc.), without the

need of additional processing. However, it is

fundamental to provide a metrological

characterization of such devices, especially if they

will be adopted for health care assistance, e.g. the

remote monitoring of chronic patients in outpatients

settings (Appelboom et al., 2014), (Lowe and

Olaighin, 2014).

Several commercially available systems are

described in literature, with details about their

performances and accuracy estimation through

dedicated tests (Kristiansen et al., 2011), (Parak et al.,

2015), (Vanderlei et al., 2008). One example is the

BioHarness

3.0 (BH3) sensing device, which

allows the user to simultaneously measure five

physiological and physical quantities (HR, BR,

Acceleration, Activity level and Posture). Specific

information about validity and reliability of the

proposed device are discussed in (Johnstone et al.,

2012a), (Johnstone et al., 2012b), through dedicated

laboratory tests. Results from field-based tests, i.e.

discontinuous incremental walk-jog-run protocol as

described in (Johnstone et al., 2012c), suggest that the

accuracy of the collected data decreases with the

entity of movement (high uncertainty for treadmill

Casaccia, S., Pietroni, F., Calvaresi, A., Revel, G. and Scalise, L.

Smart Monitoring of User’s Health at Home: Performance Evaluation and Signal Processing of a Wearable Sensor for the Measurement of Heart Rate and Breathing Rate.

DOI: 10.5220/0005694901750182

In Proceedings of the 9th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2016) - Volume 4: BIOSIGNALS, pages 175-182

ISBN: 978-989-758-170-0

175

speed > 6 m/s). The good performance of such

equipment makes possible to apply it in several field

of applications, e.g. the continuous monitoring of

sportsmen’s functional state in the conditions of

natural activity (Runova et al., 2012). Other recent

researches underlines the potentials of using this tool

in outpatient settings (Deepika et al., 2015), (Angarita

et al., 2015), (Bakhchina et al., 2014), emergency

department triage (Bianchi et al., 2013), or as a

support tool in the prevention of abnormal events, e.g.

fall detection (Hemalatha and Vaidehi, 2013),

(Sannino et al., 2015). Differently from other

wearable devices, BH3 is able to acquire and store

both raw physiological waveforms (e.g. ECG and

Breathing) and the computed quantities after a

dedicated processing (HR, BR).

The aim of this work is to improve the accuracy

assessment of the BH3, by comparing its

performances in the measure of physiological

quantities (HR, BR) with gold standard techniques

(ECG and respiration belt). In particular, the focus

has been not only in the comparison of the values

computed on-board by the device, but also the ones to

be derived through a dedicated processing of the raw

signals acquired. So, this analysis would allow to

better understand if the values computed by BH3 are

accurate enough without the need of additional

processing or, if needed, to identify the best algorithm

to apply to raw signals.

2 MATERIALS AND METHODS

BH3 is a compact physiological monitoring module,

attached to a lightweight Smart Fabric strap or

garment which incorporates ECG and Breathing

detection sensors. This version has been tuned up

with respect to the previous models, in order to

improve its usability and acceptance, e.g. by older

people (Ehmen et al., 2012). The device is directly

worn on the skin (Figure 1) of the participant via an

elasticated strap positioned around the chest (50g,

50mm width). The monitoring device (weight 18g,

28x7mm), which attaches to the left of the chest strap,

can act as both transmitter and data logger. The

internal memory makes it possible to store up to 500

hours of acquisition and the battery life has been

improved respect to previous versions, up to 35 hours

in logging mode. Five quantities are measured

simultaneously, time stamped and exportable to

Excel.

Electrocardiographic raw signal is acquired

through electrode sensors, housed within the chest

strap and sampled at 250 Hz.

Figure 1: BH3 multi-parametric device.

Breathing waveform is collected using a

capacitive pressure sensor (25 Hz) that detects the

circumference expansion and contraction of the torso.

An embedded algorithm uses the pressure change to

create a sinusoidal waveform. Aberrant pressure

changes are filtered to remove unwanted noise, and

the gain for the waveform is automatically centred at

zero (Bianchi et al., 2013). Tri-axial accelerometer

signal is acquired by using piezoelectric technology

(i.e. cantilever beam set up, sampled at 100 Hz). It is

based on a micro electro-mechanical accelerometer

sensor with a capacitive measurement scheme and is

sensitive along the three orthogonal axes (vertical (x),

sagittal (z) and lateral (y)) (Johnstone et al., 2012a).

Acceleration data is measured in gravitational force

(g) in a range of -3 to +3 g on each single axis, or as

Vector Magnitude Units (VMU), which is an

integrated value over the previous one second epoch:

 





















Finally, the posture of the subject is monitored by

using similar piezoelectric technology. Acting as an

inclinometer, data is reported in angular degrees,

ranges between -180° and +180°, monitoring how far

the device is “off the vertical” (Johnstone et al.,

2012a). In addition to the acquisition of the raw

waveforms, the device is able to perform embedded

processing to compute the instantaneous values, e.g.

HR, BR, posture, with an output frequency of 1Hz.

These data are stored in a formatted Excel file and can

also be transmitted by BH3 through wireless

communication protocol (i.e. Bluetooth SPP 2.1).

Figure 2 shows a dedicated GUI for Windows to

monitor in real-time the subject’s physiological

quantities. In this work, the BH3 has been analysed to

provide the accuracy estimation for the physiological

quantities. In particular, a novel processing technique

has been applied to the raw waveforms to extract

Heart and Breathing Rate (Section 2.3) and the

computed values have been compared to the ones

BIOSIGNALS 2016 - 9th International Conference on Bio-inspired Systems and Signal Processing

176

obtained by signals acquired from standard

instrumentation.

Figure 2: Zephyr GUI to monitor real-time subject health

status.

In addition, a similar analysis has been conducted

for the values provided by the system through on-

board processing. The measurement campaign has

been conducted for both HR and BR during rest

conditions. Moreover, given the importance of Heart

Rate, a dynamic test has also been performed, in order

to quantify the loss of accuracy of the device due to

the movement of the subject (e.g. while the user is

performing daily life routines). This aspect is

extremely important because traditional

instrumentation (i.e. ECG) are not suitable for these

kind of applications (e.g. continuous monitoring of

HR and Heart Rate variability within daily activities,

indoor and outdoor) because of the need of a

continuous power supply and the discomfort for the

user caused by the cables.

2.1 Heart Rate Test Procedure

The HR test procedure has consisted of two trials. The

first part of the work has interested the accuracy

evaluation of the quantities computed from BH3 (i.e.

instantaneous values and HR from the raw

waveforms) during rest condition. As for gold

standard measurement, a 3-lead ECG has been

acquired simultaneously by means of an

ADInstruments board (model ML865 PowerLab

4/25T), with a sampling frequency of 1 kHz and an

uncertainty of ±1 bpm.

Figure 3 shows the setup for the test performed. In

particular, it has consisted of:

• Personal computer to collect and process the data

offline;

• Acquisition board from ADInsruments for the

ECG acquisition (reference system);

• Electrode sensors for the ECG signal acquisition;

• BioHarness 3.0, placed on the chest of the subject

and then connected to PC with a proprietary

cradle (data upload for post-processing).

Figure 3: Measurement setup for Heart Rate test procedure

(rest condition).

Table 1: Participants for the HR rest tests procedure.

Subject M/F

Age

[years]

Weight

[kg]

Height

[m]

BMI

[kg/m

]

1 F 20 55 1.62 21.0

2 F 23 47 1.66 17.1

3 M 23 92 1.81 28.1

4 F 23 48 1.55 20.0

5 M 26 72 1.83 21.5

6 F 23 63 1.64 23.4

7 F 23 47 1.66 17.1

8 F 26 53 1.66 19.2

9 F 21 52 1.64 19.3

10 F 22 70 1.70 24.2

11 F 23 58 1.72 19.6

12 F 23 59 1.62 22.5

13 M 23 74 1.73 24.7

14 F 23 55 1.65 20.2

15 M 24 71 1.72 24.0

16 M 24 71 1.82 21.4

17 M 28 80 1.80 24.7

18 M 23 63 1.72 21.3

19 F 28 64 1.78 20.2

20 F 21 42 1.50 18.7

Mean

23 61.8 1.69 21.4

STD

3.1 12.4 0.09 2.7

Twenty healthy participants (Table 1) have been

recruited for this trial. They have been asked to sit

quietly and breathe normally, while their ECG has

been acquired. Five records of one minute each have

been recorded from both the BH3 and standard ECG.

During the second trial, the subjects have performed

a motion test, which has been implemented to

quantify how motion alters the performance of the

sensors. In this case, four healthy participants (2

males and 2 females, age 25.5±2.5 years, weight

63.5±11.7 kg, height 1.74±0.06 m, BMI 20.77±2.74

kg/m

) have performed a continuous walking exercise

in laboratory using a commercial treadmill. In detail,

the HR motion test procedure has consisted of:

- 5 minutes of initial rest;

- 5 minutes of a gradual speed increasing from 4

Smart Monitoring of User’s Health at Home: Performance Evaluation and Signal Processing of a Wearable Sensor for the Measurement of

Heart Rate and Breathing Rate

177

km/h to 6 km/h

- 5 minutes of a walking period with a speed of 6

km/h;

- 5 minutes of a gradual speed decreasing from 6

km/h to 4 km/h;

- 5 minutes of final rest.

The measurement setup for the motion tests has been

the same for the rest condition, with the only

difference related to the position of the electrode

sensors of the standard ECG (placed in the chest of

the subject). In this case, it has been possible to reduce

the motion artefacts of the gold standard during the

walking experiment.

2.2 Breathing Rate Test Procedure

A preliminary Breathing Rate test procedure has been

implemented with the use of a metronome application

for Android devices (available online at

https://play.google.com/store/apps/details?id=com.gi

smart.metronomefree&hl=it), in order to force the

participants to produce a standardized breathing

waveform. The BH3 has been applied to the subject

in order to acquire both its breathing waveform,

sampled at 25 Hz, and the instantaneous BR values,

(1 Hz). Five healthy participants (3 males and 2

females, age 26.2±1.5 years, weight 66.4±10.7 kg,

height 1.80±0.1 m, BMI 21.2±2.4 kg/m

) have been

recruited for the breathing test. A standard respiratory

belt from ADInstruments (model MLT1132), based

on the extensometer principle, has been used as gold

standard.

Figure 4: Measurement setup for the breathing tests.

As shown in Figure 4, both the belts have been placed

on the subject chest and acquired simultaneously.

Two different timings for the metronome have been

defined, following the frequency of a medium (25

bpm) and a high breathing (40 bpm), and two records

of 90s for each timing have been acquired.

Differently from the HR tests, the instantaneous

BR value provided by BH3 has not been compared to

the reference value, because the length of the test has

been too short to make the stabilization of the value

Figure 5: BR values computed within a trial (truncated 90 s

of signal). The short period of the trial does not allow to

have a stable value for the instantaneous BR computed by

BH3, even if it is possible to obtain the correct BR from the

raw signal.

provided possible. In fact, the instantaneous values

computed by BH3 (referred in Figure 5 as BH -

packet) have not become stable yet (a decreasing

offset is observed within the short time of the trials).

This is because the value provided by the sensor is

calculated after the application of a moving average

window, so several minutes are needed before

performing such test in order to reach a stable value.

Basing on this observation, the performance analysis

has been conducted only between the BR computed

by post-processing BH3 raw waveforms (referred in

Figure 6 as BH - signal) and the quantities computed

from the reference breathing signals. The next section

illustrates the signal processing techniques performed

for all the tests conducted.

2.3 Signals Processing

The application of a processing algorithm is needed

to compute the physiological parameters from the raw

waveforms (ECG, Breathing) of both gold standard

and BH3. In literature there are several approaches to

gather HR values from ECG signals (Pan and

Tompkins, 1985), (Josko, 2007).

In this work, the approach described previously by

authors in (Cosoli et al., 2015) has been applied for

both rest and motion tests, in order to compute the HR

values for both the gold standard and the BH3 raw

signals. The algorithm is based on the computation of

the slope of the signal, which makes the extraction of

relevant features possible, according to the

morphology of the signal (Hu et al., 2014). In

particular, the approach has been used to identify the

R peaks in the ECG waveforms and so calculate the

HR, according to the equation:

BIOSIGNALS 2016 - 9th International Conference on Bio-inspired Systems and Signal Processing

178

 









Before applying the cited procedure, the following

pre-processing steps have been applied, in order to

reduce the noise and issues related to movement

artefacts from the raw signals (Figure 6):

1) Synchronization of both signals (reference ECG,

BH3 waveform) and resampling (250 Hz);

2) Mean removal;

3) Filtering (Butterworth 3

order band pass filter –

0.5÷30 Hz);

4) Normalization with the maximum absolute value

of the waveform.

Figure 6: ECG signals acquired during the dynamic trial for

both sensing devices. (1, 2): raw ECG waveform from gold

standard technique and wearable sensor during rest

condition; (3, 4): raw signals acquired during a walking

condition (6 km/h) with treadmill.

Figure 7: Processing steps for breathing waveform within a

test (25 bpm). 1) Synchronization and resampling (25Hz);

2) Mean removal; 3) Filtering; 4) Normalization and peaks

identification.

The conditioned signal has then been divided in

windows of 30 s, where an average HR value has been

computed. As for the Breathing waveform, the signal

pre-processing procedure has been the same (the steps

are summarized in Figure 7), except for the

application of a different filter (Butterworth 3

order

band-pass filter, 0.5÷1 Hz). In addition, the algorithm

for the feature identification (Hu et al., 2014) has been

tuned, in order to locate the maximum of each

sinusoidal waveform and calculate BR consequently.

3 RESULTS

Figure 8 shows the correlation between the HR values

computed from BH3 raw signals (y-axis) and the ones

from the gold standard (x-axis). A Matlab routine

based on Minimum Volume Ellipsoid (MVE) has

been implemented to remove the outliers (red points)

(Riani et al., 2012).

Figure 8: Linear correlation between the HR values of the

BH3 raw ECG and the HR values of the standard ECG for

the rest condition of 20 participants.

Figure 9: Uncertainty estimation for the HR values of the

BH3 raw ECG for the rest condition.

Smart Monitoring of User’s Health at Home: Performance Evaluation and Signal Processing of a Wearable Sensor for the Measurement of

Heart Rate and Breathing Rate

179

Then, a residual analysis has been performed to

verify the Gaussian distribution and to calculate the

final accuracy of the measurement. For example,

Figure 9 shows a deviation of ±2.8 bpm (coverage

factor k=2). If combining this result with the

uncertainty of the gold standard (±1.0 bpm) a final

accuracy of ±3.0 bpm is obtained.

The same analysis has been conducted for all the

tests performed. In particular, the four cases reported

in Table 2 are related to:

1) the HR computed from the BH3 raw signal

during the rest condition;

2) the HR values provided (on-board processing)

for the rest condition;

3) the HR computed from the BH3 raw signal

during the motion test;

4) the HR values provided (on-board processing)

for the motion test.

Table 2: Results of the tests conducted. R

= coefficient of

determination. STD = standard deviation of residuals.

Case

Sensitivity

[]

Bias

[bpm]

[]

STD (2σ)

[bpm]

Combined

STD [bpm]

1 0.986 1.485 98.1 ±2.8 ±3.0

2 1.008

0.247

99.1 ±1.9 ±2.1

3 0.820 15.05 81.7 ±5.4 ±5.5

4 0.992 1.044 99.0 ±2.6 ±2.8

If looking at Table 2, it can be observed that the

uncertainty of HR measurement increases with the

level of activity performed and similar results have

been found in the state of the art (STD = ±6 bpm from

0 to 12km/h in (Johnstone et al., 2012a)).

Besides, it can also be noted that the HR values

computed from BH3 with a proper internal processing

(case 2 and 4) are more accurate than the ones

obtained by processing the raw ECG and Breathing

signals (case 1 and 3). This suggests that the on-board

algorithms are robust enough to provide a good

measurement, even in presence of motion artefacts

(improvement of accuracy from ±5.5 to ±2.8 bpm).

As concerning Breathing Rate, the computed BR

values are highly correlated (R

= 98.3%) to the ones

of gold standard instrumentation (Figure 10), with a

deviation of ±2.1 bpm (Figure 11).

These results are in agreement with the ones

discussed in (Johnstone et al., 2012a), where an

accuracy < ±3.0 bpm has been found. Within these

trials, it hasn’t been possible to assess the accuracy

for the instantaneous BR values due to short time of

the acquisition. This suggests that future experiments

are needed to quantify clearly the time needed to the

measurement to become stable and reliable.

Figure 10: Linear correlation between the BR values of the

BH3 raw signal and the BR values of the respiratory belt

for the frequency of a slow breathing (25 bpm) and a high

breathing (40 bpm) of five participants.

Figure 11: Uncertainty estimation for the BR values of the

BH3 raw signal within the tests performed.

4 CONCLUSIONS

Wearable devices are even low at low-cost and with

high computing capabilities, allowing them to

measure different physiological signals and extract

useful quantities (e.g. HR, BR). However, it is

fundamental to provide metrological characterization

of such measurements and, if needed, to identify the

best processing technique to get reliable values.

The aim of this work is to evaluate the accuracy

of the BioHarness 3.0 commercial device in the

monitoring of physiological parameters.

The experiments conducted for Heart Rate have

demonstrated that the values computed on-board by

the device are accurate (deviation of ±2.1 bpm for the

static condition and ±2.8 bpm concerning the test with

BIOSIGNALS 2016 - 9th International Conference on Bio-inspired Systems and Signal Processing

180

the treadmill) and no further processing of the raw

signals is required.

On the contrary, during the tests conducted for the

Breathing Rate, it has been observed that the BR

values computed require several minutes to become

stable. However, the proposed signal processing of

the breathing waveform allows to compute BR

values, which are strongly correlated (R

= 98.3%) to

the gold standard and with a deviation of ±2.1 bpm.

Future works will be focused on a deeper analysis

of the breathing signal coming from the BH3.

Particular attention will be paid on the identification

of the time interval needed for the instantaneous BR

value to become stable.

ACKNOWLEDGEMENTS

The research work has been developed within the

framework of the Health@Home Italian project,

financed by MIUR (Italian Ministry of Research).

The authors would like to thank Mr. Fabio Padiglione

(ADItech srl) and Mr. Marco Domizio (Eidos srl) for

their technical support.

REFERENCES

Abascal, J. Ambient intelligence for people with disabilities

and elderly people. ACM's Special Interest Group on

Computer-Human Interaction (SIGCHI), Ambient

Intelligence for Scientific Discovery (AISD)

Workshop, Vienna, 2004.

Abdallah, Z. S., Gaber, M. M., Srinivasan, B. &

Krishnaswamy, S. 2015. Adaptive mobile activity

recognition system with evolving data streams.

Neurocomputing, 150, 304-317.

Angarita, G. A., Natarajan, A., Gaiser, E. C., Parate, A.,

Marlin, B., Gueorguieva, R., Ganesan, D., Malison, R.

T., 2015. A remote wireless sensor

network/electrocardiographic approach to

discriminating cocaine use. Drug and Alcohol

Dependence in press.

Appelboom, G., Camacho, E., Abraham, M. E., Bruce, S.

S., Dumont, E. LP., Zacharia, B. E., D’amico, R.,

Slomian, J., Reginster, J. Y., Bruyère, O., 2014. Smart

wearable body sensors for patient self-assessment and

monitoring. Archives of Public Health, 72, 28.

Bakhchina, A. V., Shishalov, I. S., Parin, S. B., Polevaya,

S. A., 2014. The dynamic cardiovascular markers of

acute stress. International Journal of

Psychophysiology, 2, 230.

Bayat, A., Pomplun, M. & Tran, D. A. 2014. A study on

human activity recognition using accelerometer data

from smartphones. Procedia Computer Science, 34,

450-457.

Bianchi, W., Dugas, A. F., Hsieh, Y., Saheed, M., Hill, P.,

Lindauer, C., Terzis, A., Rothman, R. E., 2013.

Revitalizing a vital sign: improving detection of

tachypnea at primary triage. Annals of emergency

medicine, 61, 37-43.

Catal, C., Tufekci, S., Pirmit, E. & Kocabag, G. 2015. On

the use of ensemble of classifiers for accelerometer-

based activity recognition. Applied Soft Computing.

Cosoli, G., Casacanditella, L., Pietroni, F., Calvaresi, A.,

Revel, G. M., Scalise, L., A novel approach for features

extraction in physiological signals. In: IEEE, ed.

Memea - Medical Measurements and Applications,

2015 Turin. 380-385.

Deepika, A., Baruah, S., Shukla, D. P., Sathyaprabha, T. N.,

Devi, B. I., 2015. Demonstration of subclinical

autonomic dysfunction following severe traumatic

brain injury using serial heart rate variability

monitoring. Autonomic Neuroscience: Basic and

Clinical in press.

Demiris, G., Skubic, M., Rantz, M., Keller, J., Aud, M.,

Hensel, B., He, Z., 2006. Smart home sensors for the

elderly: a model for participatory formative evaluation.

human-computer interaction, 6, 7.

Ehmen, H., Haesner, M., Steinke, I., Dorn, M., Gövercin,

M., Steinhagen-Thiessen, E., 2012. Comparison of four

different mobile devices for measuring heart rate and

ECG with respect to aspects of usability and acceptance

by older people. Applied ergonomics, 43,

582-587.

Hemalatha, C. S., Vaidehi, V., 2013. Frequent bit pattern

mining over tri-axial accelerometer data streams for

recognizing human activities and detecting fall.

Procedia Computer Science, 19, 56-63.

Hu, X., Liu, J., Wang, J., Xiao, Z., Yao, J., 2014. Automatic

detection of onset and offset of QRS complexes

independent of isoelectric segments. Measurement, 51,

53-62.

Johnstone, J. A., Ford, P. A., HugheS, G., Watson, T.,

Garrett, A. T., 2012a. BioHarness™ multivariable

monitoring device: part. I: validity. Journal of sports

science & medicine, 11, 400.

Johnstone, J. A., Ford, P. A., Hughes, G., Watson, T.,

Garrett, A. T., 2012b. BioHarness™ Multivariable

Monitoring Device: Part. II: Reliability. Journal of

sports science & medicine, 11, 409.

Johnstone, J. A., Ford, P. A., Hughes, G., Watson, T.,

Garrett, A. T., 2012c. Field based reliability and

validity of the BioHarness™ multivariable monitoring

device. Journal of sports science & medicine, 11, 643.

Josko, A. Discrete wavelet transform in automatic ECG

signal analysis. Instrumentation and Measurement

Technology Conference Proceedings, 2007. IMTC

2007. IEEE, 2007. IEEE, 1-3.

Kristiansen, J., Korshøj, M., Skotte, J. H., Jespersen, T.,

Søgaard, K., Mortensen, O. S. & Holtermann, A. 2011.

Comparison of two systems for long-term heart rate

variability monitoring in free-living conditions-a pilot

study. Biomedical engineering online, 10, 27.

Lowe, S. A., Ólaighin, G., 2014. Monitoring human health

behaviour in one's living environment: a technological

review. Medical engineering & physics, 36, 147-168.

Smart Monitoring of User’s Health at Home: Performance Evaluation and Signal Processing of a Wearable Sensor for the Measurement of

Heart Rate and Breathing Rate

181

Pan, J. & Tompkins, W. J. 1985. A real-time QRS detection

algorithm. Biomedical Engineering, IEEE Transactions

on, 230-236.

Pantelopoulos, A. & Bourbakis, N. G. 2010. A survey on

wearable sensor-based systems for health monitoring

and prognosis. Systems, Man, and Cybernetics, Part C:

Applications and Reviews, IEEE Transactions on, 40,

1-12.

Parak, J., Tarniceriu, A., Renevey, P., Bertschi, M.,

Delgado-Gonzalo, R. & Korhonen, I. Evaluation of the

beat-to-beat detection accuracy of PulseOn wearable

optical heart rate monitor. Engineering in Medicine and

Biology Society (EMBC), 2015 37th Annual

International Conference of the IEEE, 2015. IEEE,

8099-8102.

Parvaneh, S., Grewal, G. S., Grewal, E., Menzies, R. A.,

Talal, T. K., Armstrong, D. G., Sternberg, E. & Najafi,

B. 2014. Stressing the dressing: Assessing stress during

wound care in real-time using wearable sensors. Wound

Medicine, 4, 21-26.

Revel, G. M., Pietroni, F., Zarli, A., Anfosso, A.,.

Enhancing the built environment for ageing at home

through ICTs and advanced sensing. . The Cities of

Tomorrow: the Challenges of Horizon 2020, 2014

Torino, Italy.

Riani, M., Perrotta, D., Torti, F., 2012. FSDA: A MATLAB

toolbox for robust analysis and interactive data

exploration. Chemometrics and Intelligent Laboratory

Systems, 116, 17-32.

Runova, E. V., Parin, S. B., Nekrasova, M. M., Bakhchina,

A. V., Kovalchuk, A. V., Shyshalov, I. S., Polevaya, S.

A., 2012. Monitoring and distant diagnostics of

sportsmen's functional state based on information

technologies and telemetry in the conditions of natural

activity. International Journal of Psychophysiology,

85, 420-421.

Sannino, G., De Falco, I., De Pietro, G., 2015. A supervised

approach to automatically extract a set of rules to

support fall detection in an mHealth system. Applied

Soft Computing.

Sixsmith, A., Sixsmith, J., 2000. Smart care technologies:

meeting whose needs? Journal of telemedicine and

telecare, 6, 190-192.

Van andel, J., Ungureanu, C., Aarts, R., Leijten, F. &

Arends, J. 2015. Using photoplethysmography in heart

rate monitoring of patients with epilepsy. Epilepsy &

Behavior, 45, 142-145.

Vanderlei, L., Silva, R., Pastre, C., Azevedo, F. M. D. &

Godoy, M. 2008. Comparison of the Polar S810i

monitor and the ECG for the analysis of heart rate

variability in the time and frequency domains. Brazilian

Journal of Medical and Biological Research, 41, 854-

859.

BIOSIGNALS 2016 - 9th International Conference on Bio-inspired Systems and Signal Processing

182