Cardiorespiratory Adaptations to Work Volume on an Automobile

Assembly Line

Dania Furk, Lu

ıs Silva

, Mariana Dias

, Phillip Probst

and Hugo Gamboa

Laborat

orio de Instrumentac¸

ao, Engenharia Biom

edica e F

ısica da Radiac¸

ao (LIBPhys-UNL), Departamento de F

ısica,

Faculdade de Ci

encias e Tecnologia, FCT, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal

Keywords:

Cardiovascular, Respiratory, Workload, Occupational Health.

Abstract:

Automobile assembly workers have to perform repetitive tasks with varying workload volumes, according to

their assigned workstation, on a daily basis. With inadequate recovery, this type of occupational activity has

been shown to cause cardiovascular problems. Despite these concerns, cardiovascular and respiratory adapta-

tions to workload variations are often overlooked. This study aims to analyze Electrocardiogram (ECG) and

Respiratory Inductance Plethysmography (RIP) data to understand the evolution of cardiorespiratory adapta-

tions to three speciﬁc work volumes. A sample of sixteen male operators (age = 38±8 years; BMI = 25 ± 3

kg.m

) volunteered from three workstations (H

, H

and H

) with different work cycle durations (1, 3 and

5 minutes, respectively). The results showed that activities with distinct workloads cause different responses

through the data collection in cardiovascular load, heart rate variability (HRV), and respiratory frequency, vari-

ability, and coordination. The workload volume and work phase both inﬂuenced the cardiorespiratory acute

response of the operators on the automobile assembly line, something that could improve individual-speciﬁc

management of tasks assigned to workers.

1 INTRODUCTION

In 2021, a survey (CEDEFOP, 2023) carried out by

the European Union determined that the majority of

workers within the EU are employed in the manu-

facturing sector. On average, employees such as ma-

chine operators and assemblers, work for 39.7 hours

per week (Eurostat, 2023).

For the majority of that time, operators are ex-

posed to numerous occupational hazards such as

repetitive movements, awkward postures, static po-

sitioning, and forceful exertions (Niu, 2010). These

types of activities have been shown to have conse-

quences for worker’s cardiovascular health, due to the

prolonged, low to moderate intensity nature of the

physical activity performed (Holtermann et al., 2018).

Another problem with this type of occupation is the

insufﬁcient recovery of the cardiac system, leading

to continuous stimulation of the body’s inﬂammatory

response, which can lead to the development of car-

diac risks and diseases or their aggravation (Geurts

and Sonnentag, 2006).

https://orcid.org/0000-0001-9811-0571

https://orcid.org/0000-0002-0172-4559

https://orcid.org/0000-0003-3239-9813

https://orcid.org/0000-0002-4022-7424

Speciﬁcally, assembly-line jobs have been pre-

viously linked with high blood pressure, (Pickering

et al., 1996), atherosclerosis (Krause et al., 2007), a

well-known cardiovascular risk, increased mortality

(Krause et al., 2017) and long-term sickness absence

(Holtermann et al., 2012).

Regarding respiration, the monitoring of this

biosignal’s frequency is used to assess effort in sports

(Nicol

o et al., 2017b) and has shown its usefulness

in the identiﬁcation of cognitive load, environmental

stress, and other relevant factors in occupational set-

tings (Massaroni et al., 2019).

Current occupational risk quantiﬁcation tools con-

sider different body parts and key indicators of biome-

chanical load, where an expert ﬁlls in a pre-deﬁned

scoring sheet by watching the workers perform their

tasks. Some of the most common scoring sheets are

the job strain index, OCRA (Occupational Repetitive

Action), the EAWS (European Assembly Worksheet),

and the revised NIOSH (National Institute for Occu-

pational Safety and Health) equation (Andreas and Jo-

hanssons, 2018). The last equation, besides establish-

ing work practice guides for manual lifting deﬁning

limits on workload, recommends that the demands

put on workers should not surpass 30% of their aer-

obic capacity in an 8-hour continuous shift (NIOSH,

1981).

Furk, D., Silva, L., Dias, M., Probst, P. and Gamboa, H.

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line.

DOI: 10.5220/0012587800003657

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 1, pages 71-81

ISBN: 978-989-758-688-0; ISSN: 2184-4305

The incorporation of this risk information into

workplace management has shown positive outcomes

such as a reduction in work-related illnesses, workers’

compensation costs, absenteeism, and increased pro-

ductivity (Goggins et al., 2008; Baraldi and Kamin-

ski, 2011). These methods present multiple shortcom-

ings as they focus only on the biomechanical aspect of

work not accounting for other factors such as individ-

ual speciﬁc differences and physiological load. Fur-

thermore, this analysis relies mainly on observational

methods, that are time-consuming and less accurate

for smaller body parts (Takala et al., 2010), also giv-

ing limited knowledge about internal adaptations to

the task (van der Beek and Frings-Dresen, 1998).

The use of wearables allows the direct measure-

ment of motion and biosignals activity in real-time,

providing means to more individual-speciﬁc plan-

ning and interventions in real occupational settings

(Romero et al., 2016; Goggins et al., 2008)

Previous studies on real assembly lines have put

their main focus on cardiovascular response. Lund-

berg et al. measured self-reports of work charac-

teristics and of perceived physical load, accompa-

nied by the evaluation of objective measures: HR

(Heart Rate), blood pressure, catecholamines, and

cortisol, ﬁnding that perceived stress was associated

with neuroendocrine response and that during work

both HR and blood pressure were signiﬁcantly in-

creased (Lundberg et al., 1989). To study the impact

of minimization of non-productive time during work

activities, to complement biomechanical exposure,

Palmerud et al. and Kazmierczak et al. quantiﬁed

job exposure based on HR monitoring by extracting

the Reserve Heart Rate (RHR) of the workers, where

the ﬁrst one found that mean HR decreased (Palmerud

et al., 2012), and the latter, when adopting this

strategy, increased cardiovascular load.(Kazmierczak

et al., 2005). The modeling of ideal working time

based on energy expenditure of assembly-line work-

ers was also developed, based on moderate workload

tasks (Ayabar et al., 2015). The energy expenditure

was computed based on HR measurements of work-

ers, with smart watches, from three different assembly

lines.

Research on the effect of this kind of work on car-

diorespiratory adaptations has also been made. Nar-

dolillo et al. simulated assembly line tasks and ex-

tracted Heart Rate Variability (HRV) from HR of par-

ticipants, measured with a wearable device. The par-

ticipants included individuals who were either cur-

rently employed in the sector or a related ﬁeld, as

well as students of varying ages and genders. It was

concluded that there were no signiﬁcant differences

in frequency domain metrics between stages of work,

but there were marked differences in some of the

time metrics Mean RRi (intervals between consecu-

tive heartbeats), Standard Deviation (SD) of Normal

to Normal RRi intervals, and RRi Triangular Index)

between some of the trials (Nardolillo et al., 2017).

In our previous work, both Respiratory Induc-

tance Plethysmography (RIP) and ECG were mon-

itored in simulated repetitive tasks under a fatigue-

inducing protocol, where thoraco-abdominal coordi-

nation (Silva et al., 2022) and HRV (Carvalho et al.,

2023) parameters were analyzed from non-worker

participants. In these studies, results showed a de-

crease in correlation and PS between the respiratory

movements of the chest and abdominal walls, and a

decrease in HRV between trials.

It was veriﬁed that current research on assembly

lines focused on exploring RHR (Relative Heart Rate)

as a measure of job exposure, aiming to predict or an-

alyze the effects of interventions made on the line or-

ganization, not considering other physiological mea-

sures. The ones that do, normally were simulated

tasks, not in a real work context and their participants

weren’t workers of the sector.

This study aimed to check for possible health is-

sues related to heart and breathing functions caused

by car manufacturing tasks. We monitored workers’

heart and respiratory signals, ECG, and RIP, respec-

tively, on a real assembly line during their regular

work.

This work is part of the OPERATOR 4.0

project (Zenithwings, Fraunhofer AICOS, LIBPhys-

UNL, Volkswagen Autoeuropa, NST Apparel Lda,

FPCEUP, Controlconsul, Universidade do Minho, In-

stitute for Medical Engineering and Science at MIT,

2020), which has the support of MIT Portugal.

The remaining sections of this document are or-

ganized as follows: in Section 2 the materials and

methods used are described, including a brief descrip-

tion of the study sample, the followed data acquisition

protocol, and adopted statistical analysis. In Section

4 the results are presented, in Section 5 their discus-

sion, and in Section 6 the drawn conclusions and the

proposed future work.

2 MATERIALS AND METHODS

To characterize cardiovascular and respiratory re-

sponses to cyclical work, multiple biosignals of as-

sembly line operators were monitored on the ﬁeld,

during real tasks.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

2.1 Participants

The collected data of 16 subjects from the assembly

workstations were analyzed (age: 38 ± 8 yrs; body

mass index: 25 ± 3 kg.m

; physical activity: 220 ±

135 minutes per week).

To be able to analyze the physiological response to

distinct work volumes, the included participants were

part of three workstations of the assembly processes,

8 from H

, 4 from H

, and 4 from H

. All were right-

handed male workers on automotive assembly lines

and all rotate between workstations.

2.2 Workstations

In this factory, the manufacturing of vehicles relies

on multiple processes. This study focused only on the

activities carried out in the ﬁnal stage of production

as they rely on manual handling: the assembly and

alignment of the last parts to be mounted on the car.

The tasks performed at the studied workstations are

speciﬁed:

• The setting of tail lights and preﬁt alignments

were recorded in H

, these tasks had a mean cycle

time of 3 minutes.

• At H

alignments of the side doors, rear end, and

front end were analyzed, with the mean duration

of the cycle being 5 minutes.

• At last, in the H

area the ﬁnal stages of the as-

sembly are performed: rear-view mirror, cowl top,

boot panel, and the trunk symbol mounting were

monitored, these activities had a cycle-time of 1

minute.

In Figure 1 the dominant positions in which work-

ers perform each of the activities mentioned above are

represented.

2.3 Data Collection

Following the reading and signing of the informed

consent, personal information (age, height, physical

activity habits, and dominant hand) was asked of the

participants. After this, their body mass was mea-

sured with a digital scale.

Next, all devices for the biosignals monitoring

were mounted (further detailed in the following sub-

section), and the volunteers placed the mobile phone

in their uniform pockets, and their activity was moni-

tored for about 50 minutes of normal occupational ac-

tivity. Simultaneously with the beginning of the data

collection, a video of the ﬁrst 4-5 cycles of the per-

formed task was recorded.

Figure 1: Dominant positions of the tasks from each work-

station.

The study was conducted in accordance with the

Declaration of Helsinki, and the protocol was ap-

proved by the Ethics Committee of the University of

Porto.

2.3.1 Sensor Setup

The followed protocol comprised the measurement

of ECG, RIP, and ACC signals. The sensor placing

started by cleaning the skin areas on which electrodes

were to be attached, to optimize skin-electrode con-

ductivity, hair removal, abrasion, and alcohol clean-

ing were done on each subject.

Firstly, three disposable adhesive Ag/AgCl elec-

trodes (Ambu

), attached to each electrode cable

from the ECG sensor (PLUX WIRELESS BIOSIG-

NALS S.A.), were placed in a conﬁguration to min-

imize arm and chest movement artifacts. On the left

side of the sternum, the positive was positioned at the

level of the manubrium, and the negative was put on

the superior part of the sternum’s body. The ground

electrode was placed on the left anterior superior iliac

spine.

Following this, RIP signals were monitored with

two inductive sensors (PLUX WIRELESS BIOSIG-

NALS S.A.) attached to two elastic belts: one over the

chest passing underarms and the other band at the um-

bilicus level (Gastinger et al., 2014; Silva et al., 2022),

they were adjusted to the participant’s anatomy.

A triaxial ACC (PLUX WIRELESS BIOSIG-

NALS S.A.) was also used and was placed on the cen-

ter of the lower back, secured with an elastic belt.

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line

Figure 2: Sensor setup for signal acquisition.

These three sensors with an acquisition rate of

350 Hz, were all connected to the 8-channel wireless

Hub, PLUX Biosignals (PLUX WIRELESS BIOSIG-

NALS S.A.), that streamed the data from each of

the sensors to the opensignals, (PLUX WIRELESS

BIOSIGNALS S.A.) software, to a smartphone (Xi-

aomi Redmi Note 9, running Android 10).

The right trapezius’ and biceps’ motion data was

also captured with two MuscleBAN, (PLUX WIRE-

LESS BIOSIGNALS S.A.) devices, at a rate of

1000 Hz and also transmitted to opensignals, (PLUX

WIRELESS BIOSIGNALS S.A.).

The ACC data from the lower back, bicep, and

trapezius were acquired for purposes of device syn-

chronization and signal segmentation since the tasks

involve a lot of repetitive arm movements.

A scheme of the sensor conﬁguration is shown in

Figure 2.

2.4 Signal Processing

2.4.1 Electrocardiography

Most of the features that were extracted from the

ECG signal are based on the detection of the R

peaks, so other wave shapes were suppressed to high-

light them. To do that, a Maximum Overlap Dis-

crete Wavelet Transform (MODWT) was used to ﬁl-

ter ECG, given the non-stationary nature of this sig-

nal (Hess-nielsen and Wickerhauser, 1996; Chen and

Tsui, 2020), where the chosen wavelet for each sub-

ject was based on the maximization of the signal en-

ergy to signal Shannon entropy ratio (He et al., 2015),

resulting in the db2 wavelet being the best one. These

time series were decomposed into 6 wavelet levels,

and the signal was reconstructed with the level 4 coef-

ﬁcients by performing the inverse wavelet transform.

These coefﬁcients corresponded to a frequency band

of 10.94-21.88 Hz. Next, an R-peak detection algo-

rithm based on the Shannon Energy Envelope (Xu

and Du, 2022) was applied to the ﬁltered signals. The

ﬁrst step was the normalization of the signal’s ampli-

tude, followed by the computation of its Shannon En-

ergy, and envelope using a moving root mean square

with a window of 70 samples. Like this, the R-peaks

were enhanced and were detected with the scipy’s

ﬁndpeaks function (Virtanen et al., 2020), applying

a minimum distance of 120 samples and a minimum

height of 0.15.

From the detected R-peaks several features were

extracted quantifying both HRV and Cardiovascular

Load as they have been previously used in multi-

ple studies to evaluate cardiovascular load in occupa-

tional context (Dias et al., 2023). The speciﬁc metrics

are indicated in Table 1.

Table 1: ECG extracted features.

HRV (ms) Cardiovascular Load (%)

SDRRi RHR

RMSSD CVL

SD1 CVS

SD2

SDRRi- Standard deviation of consecutive RR

peak intervals; RMSSD - Root Mean Square of

the consecutive RR peak intervals; RHR- Re-

serve Heart Rate; CVS- Cardiovascular Strain;

CVL- Cardiovascular Load; SD2- Poincar

plot standard deviation along the line of iden-

tity.

2.4.2 Respiratory Inductance Plethysmography

Initial ﬁltering was made by a ﬁnite impulse response

bandpass with cut-off frequencies of 0.15 and 0.45Hz

(Silva et al., 2022). Next, the signal was decom-

posed in Intrinsic Mode Functions (IMFs) with a ﬁlter

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

based on Masked Sift Empirical Mode Decomposi-

tion (EMD) (Liu et al., 2013). IMF-4 was chosen to

reconstruct the respiratory signal as it was the one that

presented the most clear respiratory pattern. Both sig-

nals from the chest and abdominal belt were subjected

to this procedure.

From the ﬁltered signal, respiratory rate (RR) in

breaths per minute was determined for both thoracic

and abdominal walls, with a zero-crossing detection

algorithm (R

etory et al., 2016). The rib-cage per-

centage (RC%) deﬁned as the Rib Cage’s (RC) con-

tribution to tidal volume as a percentage of the sum

of both RC and Abdominal (ABD) volume varia-

tion (Ryan et al., 2020) was also computed. Full

Cross-correlation between RC and ABD signals was

extracted, using a moving window of 400 samples

(Makowski et al., 2021). At last, PS between the two

signals from each belt was determined, by applying

the Hilbert Transform to both signals and the subtrac-

tion of their extracted imaginary parts (phase) (Silva

et al., 2022).

2.4.3 Accelerometer

To denoise the ACC data obtained from the triax-

ial ACC mounted on the center of the lower back

and from the ACCs in the MuscleBAN devices, a

bandpass-ﬁlter with cut-off frequencies of 0.1Hz and

10Hz was used (Silva et al., 2022; Lester et al., 2004).

Next, the signals were smoothed with a window of 0.2

seconds.

2.5 Signal Segmentation

2.5.1 Signal Synchronization

To be able to identify cycles, the ﬁrst step was to syn-

chronize the signals acquired from the multiple de-

vices. This was accomplished by ﬁrst matching the

sampling frequencies of the signals, followed by their

alignment, by computing the full cross-correlation be-

tween the ACC signals of the three devices, as each

volunteer before data recordings was asked to perform

10 jumps, a marked movement to match these signals

in time, leaving all of them in the same initial condi-

tion.

2.5.2 Self-Similarity Matrix

As this study was done in a real automobile assem-

bly line, there were unexpected events, such as line

stops, bathroom breaks, and tasks performed with ad-

ditional movements or in a different way. Despite

that, the monitored activities should have a repeat-

ing pattern, as they are performed with speciﬁc move-

ments in a certain order. As no timestamps or videos

were available for the full acquisition the identiﬁca-

tion of anomalies and cyclic patterns and the segmen-

tation of these signals was accomplished by a Self-

Similarity Matrix (SSM) method. This method has

already been used successfully to segment time se-

ries, in human activity recognition, biosignals seg-

mentation (Rodrigues et al., 2022), and in work-cycle

anomaly and pattern detection (Santos et al., 2021).

2.5.3 Data Cleaning

In the followed protocol, multiple sensors were used

simultaneously, which led some of them to fail to ac-

quire or to disconnect during the acquisition. Also,

in some cases, there were unexpected assembly line

stops, which led us to exclude those recordings from

the analysis, to guarantee that only comparable data

were used. All the signals were cut from the ﬁrst to

the last detected work cycles. This way, the record-

ings were left with 40 minutes to be analyzed.

3 STATISTICAL ANALYSIS

To evaluate how workload affects cardiorespiratory

response throughout the acquisition, the signals were

studied at two-time points: the ﬁrst and last 10 min-

utes, extracting the cardiac and respiratory indicators

at each of those times.

The employed statistical test was a Mixed

ANOVA, as there were two factors: the workstation

and the phase (ﬁrst and last 10 minutes) at which

the metric was extracted. To be able to perform a

more robust test, the sample size was increased to bal-

ance the minority workstations, by cluster-based over-

sampling, i.e., generating artiﬁcial samples with the

SMOTE (Synthetic Minority Over-sampling TEch-

nique) algorithm (Chawla et al., 2002).

To guarantee that the results obtained were reli-

able, 500 simulations changing the random seed of

the over-sampler were performed and the determina-

tion of the p-value was computed with the harmonic

mean combined p-value method, as it is used for de-

pendent tests (Wilson, 2019).

The obtained p-values were corrected for viola-

tion of normality and equality of variance principles,

by using the Yeo-Johnson power transform (Yeo and

Johnson, 2000) and the Welch correction, respec-

tively. The considered level of signiﬁcance was 5%

and when indicators were under it, they were further

analysed by the Tukey post-hoc.

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line

4 RESULTS

4.1 Cardiovascular Response

The most notable results of the mixed test were that

had an evident decrease in SDRRi, SD of RHR,

coefﬁcient of variation of HR, and SD2. The H

workstation had a meaningful change in HR that de-

creased with time. The values of the extracted ECG

metrics of station H

were generally smaller and re-

mained constant through the recording.

The results from the Mixed ANOVA simulation

statistics of the cardiac indicators are presented in Ta-

ble 2 where the p-value of each interaction is shown,

and signiﬁcant results are presented with an *.

Table 2: Mixed ANOVA results for the cardiovascular met-

rics.

Variables p

int

HR 0.432 0.006* 0.043*

Max HR 0.486 0.097 0.038*

SDRRi 0.530 0.018* 0.009*

CV HR 0.273 0.001* 0.017*

RHR 0.229 <0.001* 0.021*

SD RHR 0.274 <0.001* 0.030*

CV RHR 0.071 0.563 0.007*

CVS 0.051 0.008* 0.094

CVL 0.027* 0.006* 0.058

CVL range 0.075 0.198 0.040*

SD2 0.502 0.015* 0.008*

HR- Heart Rate; SDRRi- Standard deviation of con-

secutive RR intervals; RHR- Reserve Heart Rate;

CVS- Cardiovascular Strain; CVL- Cardiovascu-

lar Load; SD2- Poincar

e plot standard deviation

along the line of identity; Max- Maximum; Min-

Minimum; SD Standard Deviation; CV- Coefﬁcient

of Variation; p

- harmonic mean combined p-value

for the between factor; p

- harmonic mean com-

bined p-value for the within factor; p

int

- harmonic

mean combined p-value for the interaction; * sig-

niﬁcant result at a conﬁdence level of 5%.

The post-hoc tests are presented in Figure 3.

4.2 Respiratory Response

The statistical analysis revealed that the mean and

maximal ABD RR was greater, and so was its SD

for the H

workstation. Substantial reduction in res-

piratory correlation and PS was seen for H

and H

stations, respectively. In contrast, H

subjects main-

tained the relationship between both wall motions

through the analyzed time, having the highest mean

values of this metric within the assembly line.

The results from the Mixed ANOVA simulation

tests of the respiratory indicators are presented in Ta-

ble 3.

Table 3: Mixed ANOVA results for the respiratory metrics.

Variables p

int

Max RC 0.534 0.033* <0.001*

Min RC 0.891 0.533 <0.001*

SD RC 0.391 0.491 0.013*

ABD 0.190 0.468 <0.001*

Max ABD 0.697 <0.001* <0.001*

SD ABD 0.539 0.008* 0.077

Correlation 0.022* <0.001* 0.013*

PS 0.022* <0.001* 0.015*

RC- Ribcage; ABD- Abdominal; PS- Phase syn-

chrony; p

- harmonic mean combined p-value for

the between factor; p

- harmonic mean combined

p-value for the within factor; p

int

- harmonic mean

combined p-value for the interaction; * signiﬁcant

result at a conﬁdence level of 5%.

The post-hoc results are presented in Figure 4.

5 DISCUSSION

This research aimed to better understand cardiores-

piratory adaptations to cyclical work of tasks with

speciﬁc work volumes, and determine if there were

acute changes in the explored indicators extracted

from ECG and RIP signals throughout the data col-

lection. It was expected that the physiological re-

sponse would change according to the different work-

load volumes that are associated with the workstation.

To do so, data was analyzed considering two factors:

phase and workstation.

In terms of the cardiovascular indicators, HR is

used as an indicator of the load put on the cardio-

vascular system (Samani et al., 2012; Umer, 2020),

whereas HRV provides relevant information on au-

tonomic regulation to different stimuli (Mccraty and

Shaffer, 2015). The inﬂuence of the parasympathetic

nervous system on HR prevails over sympathetic ac-

tivity during resting and moderate effort conditions,

with increased variability in inter-beat intervals (Sam-

mito et al., 2015). A decrease in HRV parameters has

been previously linked to fatigue and work stress, but

it is also observed as a reaction to physical activity, in-

tensity, and duration (Tsai, 2017; Tonello et al., 2014;

Brockmann and Hunt, 2023).

When analyzing the values for cardiovascular load

features in Table 2 and looking at the evolution of

HRV in Figure 3, it seems that the H

workstation

tasks place a higher demand on the cardiovascular

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

Figure 3: Descriptive plots with Tukey test results for the signiﬁcant cardiac variables. URQ: Workstation; B: Beginning; E:

Ending. The dashed line box represents a signiﬁcant result in the within factor. The solid line represents signiﬁcant differences

in the between factors, with markers on its tips: B-circle; E-solid circle.

system in terms of both load and cardiovascular regu-

lation mechanisms. Indeed in these groups, the repe-

tition of movements is more frequent, something that

has been associated with increased metabolic, cardiac

and perceived stress (Mang et al., 2022), addition-

ally a lot of movements are performed above-head in

the mentioned workstation, that in similar activities

have shown to increase HR and blood pressure (As-

trand et al., 1968). A signiﬁcant decrease in HRV pa-

rameters has also been found in simulated assembly

line tasks (Carvalho et al., 2023), but here a fatigue-

inducing protocol was applied.

Regarding respiratory adaptations, respiratory fre-

quency is an indicator of physical workload (Nicol

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line

Figure 4: Descriptive plots with Tukey test results for the signiﬁcant respiratory variables. URQ: Workstation; H1- medium

cycle station; H2- long cycle station; H3- short cycle station; B: Beginning; E: Ending. The dashed line box represents a

signiﬁcant result in the within factor. The solid line represents signiﬁcant differences in the between factors, with markers on

its tips: B-circle; E-solid circle.

et al., 2017a). Thoraco-abdominal asynchrony is used

as a measure of respiratory muscle load, deﬁned as

the unmatching movements of the RC and ABD walls,

expressed in phase angle (Hammer and Newth, 2009).

From the observed workstations, the H

’s ABD

respiratory variables present the most distinct evolu-

tion from the initial and ﬁnal 10 minutes, as shown

in Figure 4. This suggests a chaotic movement of the

abdominal muscles, meanwhile, RC parameters did

not change, reﬂected in the diminished correlation be-

tween their motion as conﬁrmed by 3. The tasks per-

formed on the line involve arm movements, a possi-

ble explanation for these results is the fact that the in-

spiratory thoracic muscles are exerting less force for

breathing, leaving ABD expiration muscles and the

diaphragm to compensate (Celli et al., 1988).

The asynchronous motion between walls through

the data collection period was evident in H

and H

workstations and is consistent with the results in Silva

et al., where phase and correlation decreased between

baseline and fatigue trials in a simulated repetitive

task (Silva et al., 2022). The H

subjects maintained

the relationship between both wall motions through

the analyzed time, which suggests different adaption

mechanisms inherent to this station when compared

to the others.

These parameters are easily monitored with dif-

ferent wearable applications, such as smartwatches,

being that this information could be used to leverage

task rotation, by identifying which tasks are harder

for each employee. This type of approach has already

been used in the construction sector to quantify task

demands (Sadat-Mohammadi et al., 2021). Another

possible application is the monitoring and prevention

of health problems of the operators. For instance, the

identiﬁcation of a parameter that is above or under

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

recommended limits, where an alert could be sent to

the worker and supervisor that they need a break. This

has already been applied in Tsai, but in the construc-

tion industry (Tsai, 2017).

6 CONCLUSIONS AND FUTURE

WORK

The cardiorespiratory adaptations to different work

volumes were studied using ECG and RIP signals

monitoring during a period of the shift of multiple op-

erators.

The analysis of the evolution of the chosen cardiac

and respiratory indicators showed contrasts among

tasks with distinct work volumes, where cardiac load

was higher in the H

workstation and respiratory dif-

ﬁculty appeared to be higher in the H

and H

work-

stations, revealing distinct strategies of adaptation de-

pending on work-volume.

These ﬁndings are enthusiastic as they corroborate

the usefulness of the integration of respiratory moni-

toring in assembly-line occupational settings and re-

inforce the importance of measuring operators’ car-

diac activity. The found variables could be further

used in task management and interventions at the

workplace.

The time of acquisition in future research should

be longer to obtain a broader vision of the operators’

response to their daily shift and further understanding

of internal compensation mechanisms. Moreover, an

increase in sample size to amplify the representativ-

ity of the studied factory and effect size of statistical

analysis.

This research underscores the potential and sig-

niﬁcance of cardiorespiratory monitoring in enhanc-

ing both task management and worker well-being on

assembly lines. By doing so, it paves the way for a

more human-centered workplace environment. This

should be considered to identify possible degradation

of worker’s long-term well-being.

ACKNOWLEDGEMENTS

This work was supported by Project OPERATOR

(NORTE01-0247-FEDER-045910), co-ﬁnanced by

the European Regional Development Fund through

the North Portugal Regional Operational Program

and Lisbon Regional Operational Program and by

the Portuguese Foundation for Science and Technol-

ogy, under the MIT Portugal Program. M. Dias

and P. Probst were supported by the doctoral Grants

SFRH/BD/151375/2021 and RT/BD/152843/2021,

respectively, ﬁnanced by the Portuguese Foundation

for Science and Technology (FCT), and with funds

from the State Budget, under the MIT Portugal Pro-

gram.

REFERENCES

Andreas, G.-W. J. and Johanssons, E. (2018). Observa-

tional Methods for Assessing Ergonomic Risks for

Work-Related Musculoskeletal Disorders. A Scoping

Review. Revista Ciencias de la Salud, 16:8–38.

Astrand, I., Guharay, A., and Wahren, J. (1968). Circulatory

Responses to Arm Exercise with Different Arm Posi-

tions. Journal of Applied Physiology, 25(5):528–532.

Ayabar, A., De la Riva, J., Sanchez, J., and Balderrama, C.

(2015). Regression model to estimate standard time

through energy consumption of workers in manual as-

sembly lines under moderate workload. Journal of

Industrial Engineering, 2015.

Baraldi, E. C. and Kaminski, P. C. (2011). Ergonomic

planned supply in an automotive assembly line. Hu-

man Factors and Ergonomics in Manufacturing &

Service Industries, 21(1):104–119.

Brockmann, L. and Hunt, K. J. (2023). Heart Rate Variabil-

ity Changes with Respect to Time and Exercise Inten-

sity During Heart-rate-controlled Steady-state Tread-

mill Running. Scientiﬁc Reports, 13:8515.

Carvalho, D., Silva, L., Carvalho, M., Dias, M., Costa, N.,

Folgado, D., Nunes, M., Gamboa, H., Andza, K., and

Edelman, E. (2023). Cardiovascular Reactivity (CVR)

During Repetitive Work in the Presence of Fatigue. In

Ahram, T., Karwowski, W., Bucchianico, P. D., Ta-

iar, R., Casarotto, L., and Costa, P., editors, Intelli-

gent Human Systems Integration (IHSI 2023): Inte-

grating People and Intelligent Systems, volume 69.

AHFE Open Access, AHFE International.

CEDEFOP (2023). Employed Population.

Celli, B. R., Criner, G., and Rassulo, J. (1988). Ventilatory

Muscle Recruitment During Unsupported Arm Exer-

cise in Normal Subjects. Journal of Applied Physiol-

ogy, 64(5):1936–1941.

Chawla, N., Bowyer, K., Hall, L., and Kegelmeyer, W.

(2002). SMOTE: Synthetic Minority Over-sampling

Technique. J. Artif. Intell. Res. (JAIR), 16:321–357.

Chen, C.-C. and Tsui, F. R. (2020). Comparing different

wavelet transforms on removing electrocardiogram

baseline wanders and special trends. BMC Medical

Informatics and Decision Making, 20.

Dias, M., Silva, L., Folgado, D., Nunes, M. L., Cepeda, C.,

Cheetham, M., and Gamboa, H. (2023). Cardiovascu-

lar load assessment in the workplace: A systematic re-

view. International Journal of Industrial Ergonomics,

96:103476.

Eurostat (2023). Hours of work - Annual Statistics.

Gastinger, S., Donnelly, A., Dumond, R., and Prioux, J.

(2014). A Review of the Evidence for the Use of Ven-

tilation as a Surrogate Measure of Energy Expendi-

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line

ture. JPEN. Journal of parenteral and enteral nutri-

tion, 38.

Geurts, S. A. and Sonnentag, S. (2006). Recovery as

an explanatory mechanism in the relation between

acute stress reactions and chronic health impairment.

Scandinavian journal of work, environment & health,

pages 482–492.

Goggins, R. W., Spielholz, P., and Nothstein, G. L. (2008).

Estimating the effectiveness of ergonomics interven-

tions through case studies: Implications for predic-

tive cost-beneﬁt analysis. Journal of Safety Research,

39(3):339–344.

Hammer, J. and Newth, C. (2009). Assessment of thoraco-

abdominal asynchrony. Paediatric Respiratory Re-

views, 10(2):75 –80.

He, H., Tan, Y., and Wang, Y. (2015). Optimal base wavelet

selection for ECG noise reduction using a comprehen-

sive entropy criterion. Entropy, 17(9):6093–6109.

Hess-nielsen, N. and Wickerhauser, M. (1996). Wavelets

and time-frequency analysis. Proceedings of the

IEEE, 84:523–540.

Holtermann, A., Hansen, J. V., Burr, H., Søgaard, K., and

Sjøgaard, G. (2012). The health paradox of occupa-

tional and leisure-time physical activity. British Jour-

nal of Sports Medicine, 46(4):291–295.

Holtermann, A., Krause, N., van der Beek, A. J., and

Straker, L. (2018). The physical activity paradox:

six reasons why occupational physical activity (OPA)

does not confer the cardiovascular health beneﬁts that

leisure time physical activity does. British Journal of

Sports Medicine, 52(3):149–150.

Kazmierczak, K., Mathiassen, S. E., Forsman, M.,

and Winkel, J. (2005). An integrated analysis

of ergonomics and time consumption in Swedish

‘craft-type’ car disassembly. Applied Ergonomics,

36(3):263–273.

Krause, N., Arah, O. A., and Kauhanen, J. (2017). Physi-

cal activity and 22-year all-cause and coronary heart

disease mortality. American Journal of Industrial

Medicine, 60(11):976–990.

Krause, N., Brand, R. J., Kaplan, G. A., Kauhanen,

J., Malla, S., Tuomainen, T.-P., and Salonen, J. T.

(2007). Occupational physical activity, energy expen-

diture and 11-year progression of carotid atheroscle-

rosis. Scandinavian Journal of Work, Environment &

Health, (6):405–424.

Lester, J., Hannaford, B., and Borriello, G. (2004). “Are

You with Me?” – Using Accelerometers to Determine

If Two Devices Are Carried by the Same Person. vol-

ume 3001, pages 33–50.

Liu, S., Gao, R. X., John, D., Staudenmayer, J., and Freed-

son, P. (2013). Tissue Artifact Removal from Respi-

ratory Signals Based on Empirical Mode Decompo-

sition. Annals of biomedical engineering, 41:1003–

1015.

Lundberg, U., Granqvist, M., Hanssonand, T., Magnusson,

M., and Wallin, L. (1989). Psychological and physi-

ological stress responses during repetitive work at an

assembly line. Work & Stress, 3(2):143–153.

Makowski, D., Pham, T., Lau, Z. J., Brammer, J. C.,

Lespinasse, F., Pham, H., Sch

olzel, C., and Chen, S.

H. A. (2021). NeuroKit2: A python toolbox for neu-

rophysiological signal processing. Behavior Research

Methods, 53(4):1689–1696.

Mang, Z. A., Realzola, R. A., Ducharme, J., Bellissimo,

G. F., Beam, J. R., Mermier, C., de Castro Magalhaes,

F., Kravitz, L., and Amorim, F. T. (2022). The Effect

of Repetition Tempo on Cardiovascular and Metabolic

Stress When Time Under Tension is Matched During

Lower Body Exercise. European Journal of Applied

Physiology, 122(6):1485–1495.

Massaroni, C., Nicol

o, A., Lo Presti, D., Sacchetti, M.,

Silvestri, S., and Schena, E. (2019). Contact-Based

Methods for Measuring Respiratory Rate. Sensors,

19(4):908.

Mccraty, R. and Shaffer, F. (2015). Heart Rate Variabil-

ity: New Perspectives on Physiological Mechanisms,

Assessment of Self-regulatory Capacity, and Health

Risk. Global Advances in Health and Medicine,

4(1):46–61. PMID: 25694852.

Nardolillo, A. M., Baghdadi, A., and Cavuoto, L. A. (2017).

Heart Rate Variability During a Simulated Assembly

Task; Inﬂuence of Age and Gender. volume 61, pages

1853–1857.

Nicol

o, A., Marcora, S. M., Bazzucchi, I., and Sacchetti, M.

(2017a). Differential control of respiratory frequency

and tidal volume during high-intensity interval train-

ing. Experimental physiology, 102(8):934–949.

Nicol

o, A., Massaroni, C., and Passﬁeld, L. (2017b). Res-

piratory Frequency During Exercise: The Neglected

Physiological Measure. Frontiers in Physiology,

8:922.

NIOSH (1981). Work Practices Guide for Manual Lifting.

Number 81-122 in DHHS (NIOSH) publication. U.S.

Department of Health and Human Services, Public

Health Service, Centers for Disease Control, National

Institute for Occupational Safety and Health, Division

of Biomedical and Behavioral Science.

Niu, S. (2010). Ergonomics and occupational safety and

health: An ILO perspective. Applied Ergonomics,

41(6):744–753.

Palmerud, G., Forsman, M., Neumann, W. P., and Winkel,

J. (2012). Mechanical exposure implications of ratio-

nalization: A comparison of two ﬂow strategies in a

Swedish manufacturing plant. Applied Ergonomics,

43(6):1110–1121.

Pickering, T. G., Devereux, R. B., James, G. D., Gerin, W.,

Landsbergis, P., Schnall, P. L., and Schwartz, J. E.

(1996). Environmental inﬂuences on blood pressure

and the role of job strain. Journal of hypertension.

Supplement : ofﬁcial journal of the International So-

ciety of Hypertension, 14(5):S179–85.

etory, Y., Niedzialkowski, P., de Picciotto, C., Bonay, M.,

and Petitjean, M. (2016). New Respiratory Inductive

Plethysmography (RIP) Method for Evaluating Ven-

tilatory Adaptation during Mild Physical Activities.

PLoS ONE, 11.

Rodrigues, J., Liu, H., Folgado, D., Belo, D., Schultz,

T., and Gamboa, H. (2022). Feature-Based Informa-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

tion Retrieval of Multimodal Biosignals with a Self-

Similarity Matrix: Focus on Automatic Segmentation.

Biosensors, 12(12).

Romero, D., Stahre, J., Wuest, T., Noran, O., Bernus, P.,

Fasth, Fast-Berglund,

A., and Gorecky, D. (2016). To-

wards an Operator 4.0 Typology: A Human-Centric

Perspective on the Fourth Industrial Revolution Tech-

nologies. pages 29–31.

Ryan, L., Rahman, T., Strang, A., Heinle, R., and Shaf-

fer, T. H. (2020). Diagnostic differences in respira-

tory breathing patterns and work of breathing indices

in children with Duchenne muscular dystrophy. PLoS

ONE, 15.

Sadat-Mohammadi, M., Shakerian, S., Liu, Y., Asadi, S.,

and Jebelli, H. (2021). Non-invasive physical demand

assessment using wearable respiration sensor and ran-

dom forest classiﬁer. Journal of Building Engineering,

44:103–279.

Samani, A., Holtermann, A., Søgaard, K., Holtermann,

A., and Madeleine, P. (2012). Following er-

gonomics guidelines decreases physical and cardio-

vascular workload during cleaning tasks. Ergonomics,

55(3):295–307. PMID: 22409167.

Sammito, S., Thielmann, B., Seibt, R., Klussmann, A.,

Weippert, M., and B

ockelmann, I. (2015). Guideline

for the application of heart rate and heart rate vari-

ability in occupational medicine and occupational sci-

ence. ASU Int, 2015(06):1–29.

Santos, A., Rodrigues, J., Folgado, D., Santos, S., Fuj

ao,

C., and Gamboa, H. (2021). Self-Similarity Matrix of

Morphological Features for Motion Data Analysis in

Manufacturing Scenarios. pages 80–90.

Silva, L., Dias, M., Folgado, D., Nunes, M., Namburi, P.,

Anthony, B., Carvalho, D., Carvalho, M., Edelman,

E., and Gamboa, H. (2022). Respiratory Inductance

Plethysmography to Assess Fatigability during Repet-

itive Work. Sensors, 22.

Takala, E.-P., Pehkonen, I., Forsman, M., Hansson, G.-

A., Mathiassen, S. E., Neumann, W. P., Sjøgaard, G.,

Veiersted, K. B., Westgaard, R. H., and Winkel, J.

(2010). Systematic evaluation of observational meth-

ods assessing biomechanical exposures at work. Scan-

dinavian Journal of Work, Environment & Health,

(1):3–24.

Tonello, L., Rodrigues, F., Souza, J., Campbell, C., Leicht,

A., and Boullosa, D. (2014). The role of physical ac-

tivity and heart rate variability for the control of work

related stress. Frontiers in Physiology, 5.

Tsai, M.-K. (2017). Applying Physiological Status Mon-

itoring in Improving Construction Safety Manage-

ment. KSCE Journal of Civil Engineering, 21:2061–

2066.

Umer, W. (2020). Sensors based physical exertion monitor-

ing for construction tasks: Comparison between tra-

ditional physiological and heart rate variability based

metrics. In Proc., Joint CIB WO99 & TG59 Conf.

van der Beek, A. J. and Frings-Dresen, M. H. (1998).

Assessment of mechanical exposure in ergonomic

epidemiology. Occupational and Environmental

Medicine, 55(5):291–299.

Virtanen, P., Gommers, R., Oliphant, T. E., Haberland, M.,

Reddy, T., Cournapeau, D., Burovski, E., Peterson, P.,

Weckesser, W., Bright, J., van der Walt, S. J., Brett,

M., Wilson, J., Millman, K. J., Mayorov, N., Nel-

son, A. R. J., Jones, E., Kern, R., Larson, E., Carey,

C. J., Polat,

I., Feng, Y., Moore, E. W., VanderPlas,

J., Laxalde, D., Perktold, J., Cimrman, R., Henriksen,

I., Quintero, E. A., Harris, C. R., Archibald, A. M.,

Ribeiro, A. H., Pedregosa, F., van Mulbregt, P., and

SciPy 1.0 Contributors (2020). SciPy 1.0: Fundamen-

tal Algorithms for Scientiﬁc Computing in Python.

Nature Methods, 17:261–272.

Wilson, D. J. (2019). The harmonic mean p-value for com-

bining dependent tests. Proceedings of the National

Academy of Sciences, 116(4):1195–1200.

Xu, W. and Du, F. (2022). A Robust Qrs Complex Detec-

tion Method Based On Shannon Energy Envelope And

Hilbert Transform. Journal of Mechanics in Medicine

and Biology, 22(03):2240013.

Yeo, I.-K. and Johnson, R. A. (2000). A New Family

of Power Transformations to Improve Normality or

Symmetry. Biometrika, 87(4):954–959.

Cardiorespiratory Adaptations to Work Volume on an Automobile Assembly Line