An Android App for Posture Analysis Using OWAS

Christian Lins

1 a

, Franziska Minh-Khai Dang Quang

, Rica Schulze

, Stefanie Lins

Andreas Hein

2 b

and Sebastian Fudickar

3 c

Department of Computer Science, Hamburg University of Applied Sciences, 20099 Hamburg, Germany

Assistance Systems and Medical Device Technology, Carl von Ossietzky University Oldenburg, 26129 Oldenburg, Germany

MOVE Junior Research Group, Institute of Medical Informatics, University of Luebeck, Germany

University of Hagen, 58097 Hagen, Germany

Keywords:

Smartphone App, Postural Assessment, Musculoskeletal Disorders, non-Neutral Postures, UEQ.

Abstract:

In this paper the APA, an App for Posture Analysis, that incorporates the OWAS method (Ovako Working

Posture Analysis System) for assessing potential harmful postures of physically hard working employees,

is presented and evaluated. The app is intended as a tool for occupational safety experts in assessing the

individual postures of workers to identify and prevent harmful working situations in regards to musculoskeletal

hazards. For this, APA incorporates a digitized assessment sheet for the OWAS method and timing support

that helps occupational safety experts structure and simplify the assessment workﬂow. To investigate whether

the protocol sheet can be replaced with the app, a study was conducted in which the inter-rater reliability

among the app and the paper protocol sheet (as control group) was evaluated. In addition, the usability of

the app was determined via the User Experience Questionnaire (UEQ). In the study, the app achieved higher

inter-rater reliability for watching a video recording of postures than the control group. The chi-square test

revealed differences in the use of the app and paper only for leg postures. The UEQ indicated overall above

average results, which indicates a sufﬁcient usability, which was also conﬁrmed by free-textual comments of

the 13 study participants. The results suggest that it would be possible to replace the paper protocol sheet in

the future with an app.

1 INTRODUCTION

Musculoskeletal disorders are a major occupational

health problem, especially in industrialized countries.

Since poor working postures – repeated often or over

a long period of time – are a major risk for mus-

culoskeletal disorders (Hoy et al., 2010; Amell and

Kumar, 2001; Karwowski and Marras, 1998; Matsui

et al., 1997), the postures of employees during work

shifts should be observed regularly to detect and pre-

vent posture damage early. It is important to iden-

tify patterns of work-related musculoskeletal disor-

ders, symptoms, and their risk factors early in the

workplace. In doing so, appropriate work postures not

only counteract the problems, but also result in greater

control of work performance and a reduction in the

https://orcid.org/0000-0003-3714-0069

https://orcid.org/0000-0001-8846-2282

https://orcid.org/0000-0002-3553-5131

number of workplace accidents. Individual worker

strain and corresponding risks are typically assessed

by human observers via questionnaire-based assess-

ment methods such as the Ovako Working Posture

Analysis System (OWAS) (Karhu et al., 1981). It is a

practical method for identifying and assessing poor

working postures. OWAS is known to be easy-to-

use and reliable in terms of high inter-rater reliability

(Lins et al., 2021). In OWAS a human observer rates

the posture of an individual in intervals (e.g. every

few minutes) using three categories: arms, back, and

legs. For each of these categories 3 to 7 different part-

postures can be coded (arms: code 1-3, back: code 1-

4, legs: code 1-7). The codes obtained can be used to

determine both individual holding combinations and

their proportion of the total time. In both cases, val-

ues are classiﬁed into four action classes (from 1: pos-

ture not harmful to 4: posture very harmful) indicat-

ing the severity of the posture and the corresponding

need of improvement. These action classes for postu-

Lins, C., Quang, F., Schulze, R., Lins, S., Hein, A. and Fudickar, S.

An Android App for Posture Analysis Using OWAS.

DOI: 10.5220/0011668300003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 5: HEALTHINF, pages 307-313

ISBN: 978-989-758-631-6; ISSN: 2184-4305

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

307

ral combination are speciﬁed in accordance with risk

assessments for musculoskeletal disorders. If a code

is assigned action class 4, the posture should be cor-

rected immediately (Karwowski and Marras, 1998).

The second option gives an action class based on the

relationship of the posture of the back, legs, and arms

for an observation period. This refers to the partial

codes and their proportion of time. The highest ac-

tion class of the 4 partial codes indicates how danger-

ous the postures were for the worker. In this regard,

the method should be easy enough to use so that un-

trained people can apply it. It should provide unam-

biguous answers, even if it leads to overconﬁrmation.

The OWAS method fulﬁlls these criteria. See (Lins

et al., 2021; Lins and Hein, 2022) for a detailed dis-

cussion of OWAS.

A digitized assessment holds advantages over

paper-based approaches: the time interval of the ob-

servation can be controlled, the assessment results can

be easily processed and summarized, and potential

coding errors can be avoided. Additionally, it might

be possible to assess several subjects at once by care-

fully timing the different observations, a task that is

very error-prone if using paper sheets. In some coun-

tries such as Germany companies might be required

to legally document the workplace assessment, which

can be automated using digital tools. This is possi-

bly a useful intermediate step to the assessment with

smart workwear, e.g. (Lins and Hein, 2022). As a re-

sult, we propose APA, an Android app which guides

occupational health experts and untrained personnel

through an OWAS observation session. This app is

intended to replace the paper-based questionnaire and

to save time both in entering the information and in

evaluating it. In addition, it is evaluated whether the

inter-rater reliability of the app is comparable to the

one of the paper-based version. Furthermore, the user

experience of untrained observers using the app is in-

vestigated. To summarize the contents of this paper:

• A newly developed app is presented as a tool for

classifying postures using the OWAS method.

• The app is tested with users to determine inter-

rater reliability of the OWAS method when using

the app.

• The observation results are compared to the usual

pen-and-paper application of OWAS.

• The User Experience Questionaire (UEQ) is used

to examine the design of the app for strengths and

weaknesses in use.

The remainder of the paper is organized as fol-

lows: ﬁrst, the app and the evaluation study and its

analysis are described (Section 2), followed by a pre-

sentation of the results in Section 3. A discussion of

the results is provided in Section 4, and the paper con-

cludes with Section 5.

2 METHODS

The APA App and the study design and the statistical

analysis are presented in the following subsections.

2.1 Android App for Posture Analysis

(APA)

APA’s main aim is to enable its users (occupational

safety experts and physiotherapists as well as layper-

sons) to assess the posture of the observed workers

via the OWAS method. The app should support the

user to record and document postures while observed

workers move in work environments. APA was pri-

marily developed for a 1:1 observation, i.e. one ob-

server observes only one worker, although a future

extension to more than one worker is possible. The

APA app (see Figure 1) was developed via a two-step

iterative human-centered design process. Initially, a

click prototype was created and was discussed with

an occupational safety expert resulting in a simpli-

ﬁcation of the user interface and a focus on a one-

by-one assessment instead of the originally intended

support of multiple parallel observations. In a second

iteration the user interface (UI) was implemented ap-

plying the 8 Golden Rules of Shneiderman (Shneider-

man, 2002). Herein, we evaluate the resulting version

of the second design iteration, integrating already the

insights of these expert interviews. Once started, APA

asks for the observer name and the observed worker’s

name and workplace (see Figure 1b). When all en-

tries have been made, a new observation session can

be started. The observation interval is set to 30 sec-

onds in accordance with (Brandl et al., 2017) in order

to leave sufﬁcient room for the risk assessment, while

collecting sufﬁcient samples for an OWAS classiﬁca-

tion. Per observation for this session, the user inter-

face ﬁrst opens with input options for the body pos-

ture, which can be seen in Figure 1c). The observer is

expected to enter individual postures for back, arms,

and legs as well as the weight load. These values

are initially set to code 1 for the ﬁrst observation and

then always set to the last used partial code. Hereby,

observers can use shortcuts for subsequent session-

observations, in case the posture does not change.

Once all codes are set correctly, the user can con-

ﬁrm the entry and then enter a waiting screen. Here

the user can end the session or move on to the next

observation. A special feature is the timer, which pro-

vides an acoustic signal after 30 seconds to remind

HEALTHINF 2023 - 16th International Conference on Health Informatics

308

the observer to continue with the observations. In ad-

dition, the timer time can be tracked visually with a

bar that ﬁlls up.

The APA provides basic information about the

speciﬁc OWAS postures used for assessment (Figure

1a) and allows the export of the assessment results

(Figure 1d). The app implementation is based on An-

droid, Java, SQLite database, the Room and the Hilt

frameworks for depencency injection and the MPAn-

droidChart API for results visualization. Test-driven

development is implemented via JUnit and Mockito.

2.2 Study Design

To evaluate the APA app and compare it with the stan-

dard paper assessment sheet, a study was conducted.

Inclusion criteria for participants was the availabil-

ity and experience in using an Android smartphone

with a supported Android version (5 to 10). Due

to COVID-19 restrictions, the study was conducted

in one on one meetings via video calls. Participants

were sent a ﬁle with a brief overview of the OWAS

posture categories and general participant information

in advance. Initially, participants were introduced to

the study aims, the general study procedure and were

shown two introductory videos. The groups were in-

troduced to the protocol-guideline and the groups spe-

ciﬁc methodology (app or paper). Concluding open

questions have been answered. Within the main study,

participants were shown a video of a man adopting

work related human postures. The man models pos-

tures evenly over all OWAS codes. All participants

were shown the same videos and were requested to

rate the posture every 30 seconds according to OWAS.

The study was approved by the Institutional Re-

view Board of the University of Oldenburg (Drs.

24/2017).

2.3 Statistical Analysis

2.3.1 Inter-Rater Reliability

Since no direct comparison with a ground truth is pos-

sible with the OWAS observation method, the test-

retest reliability with multiple observers (inter-rater

reliability) is determined between the two logging

methods (app and paper). Inter-rater reliability indi-

cates the degree of agreement when a subject is as-

sessed by multiple observers. It is usually expressed

by a kappa value (κ < 0: strong disagreement to

κ = 1.0: perfect agreement). The calculation is made

according to Fleiss (Fleiss, 1971).

2.3.2 Test for Independence

The observations (dependent variable) of the two log-

ging groups are tested for their independence from the

respective logging method (app or paper). For this

purpose, the χ

-test for independence is applied. The

null hypothesis H

is: the observations are stochasti-

cally independent of the logging method. The alterna-

tive hypothesis H

: the observations are stochastically

dependent on the logging method.

2.4 Usability Test

The User Experience Questionnaire (UEQ) is used

to evaluate the usability of the app. The UEQ was

also supplemented with free text ﬁelds to collect fur-

ther impressions of the participants about the app. To

compare the quality of the user experience of the app,

the UEQ benchmark is used, which contains the ag-

gregated data from various evaluation studies of di-

verse products. Through the benchmark, the relative

quality of the app compared to other products is esti-

mated. (Laugwitz et al., 2008)

3 RESULTS

3.1 Study Cohort

13 participants with no prior experience in coding

postures were randomized into two groups: The inter-

vention group using the APA (n = 7) and the control-

group using the established paper-protocol (n = 6).

The data of one rater from the APA group were ex-

cluded from the evaluation due to technical problems.

Two raters from the pen-and-paper group were also

excluded, because they completely misunderstood the

OWAS method.

3.2 Statistical Analysis

Table 1 summarizes the inter-rater reliabilities for the

APA use and the control condition per body-part.

Table 1: Inter-rater reliability values (per body-part cate-

gory) for both logging method groups.

Legs Arms Back

APA logging κ = 0.448 1.000 0.383

Paper logging κ = 0.417 0.657 0.249

Table 2 shows the χ

-test values (number of rat-

ings N = 191). According to the test values the H

hypothesis can be provisionally accepted for the legs

category (stochastical dependence between logging

An Android App for Posture Analysis Using OWAS

309

Figure 1: User Interface (UI) for Android of the APA App. The UI can be as well experienced as a screencast video (see

https://youtu.be/BtpdeiYyRGU).

method and rating results), for both arms and back

the H

must be kept.

Table 2: χ

-test values (per body-part category) for both

logging method groups (N = 191), *: H

provisionally ac-

cepted (V = 0.342).

Legs Arms Back

22.349* 0.448 4.762

p < 0.001* 0.799 0.190

3.3 User Experience Questionnaire

Figure 2 summarizes the averaged answers to the indi-

vidual questions of the questionnaire. The sub-items

can be summarized based on coloring into the six

main scales covered by the UEQ: Attractiveness can

be used to determine whether the participant likes the

product or not. Transparent indicates how easy it is

for the user to become familiar with the product. Efﬁ-

ciency describes whether the tasks can be completed

without unnecessary effort. Controllability indicates

whether the user has the impression of being in con-

trol at all times. Stimulation asks whether the user

feels motivated to use the product and originality in-

dicates how interesting the product is and whether it

arouses the interest of the user. The scale ranges from

-3 ”horribly bad” to +3 ”extremely good”. The range

of values between -0.8 and 0.8 encompasses a neutral

rating, values >0.8 represent a positive rating, and

values <-0.8 represent a negative rating. (Schrepp

et al., 2017)

Figure 3 indicates the ”user experience” quality

of the product. In this diagram, above average indi-

cates that 25% of the results were better than the app,

and 50% were worse. Accordingly, good means that

10% were better and 75% were worse (Schrepp et al.,

2017). The results show that the app achieves good re-

sults for Perspicuity, Efﬁciency, Dependability and it

achieves above average results in all remaining ones.

Considering the additional questionnaire-items

provided room for further free-text comments:

• The app users indicated being comfortable with

the RGB Videos.

• While many participants described the app as ap-

pealing and praised individual elements such as

the timer and the intuitive user interface, one par-

ticipant stated that she required an introduction

to use the app. In addition, one user stated that

users do not need the shortcuts, as users found

the search in them slower compared to the already

very fast normal input option.

4 DISCUSSION

Inter-Rater Reliability. The κ-values summarized

in Table 1 show an overall increased inter-rater relia-

bility for the app compared to the pen-and-paper log-

ging group. The app-raters achieved a signiﬁcantly

increased reliability for rating the arms and the back

HEALTHINF 2023 - 16th International Conference on Health Informatics

310

Figure 2: Mean answers to the individual questions of the standardized Usability Questionnaire UEQ (each color is assigned

to an already given range: Attractiveness: purple; Perspicuity: green; Novelty: orange; Stimulation: red; Dependability:

azure; Efﬁciency: dark blue);.

postures, with a perfect reliability for the arms when

using the APA app.

Test for Independence. The χ

test for indepen-

dence revealed no dependence of the observational re-

sults on the logging method, at least for the arms and

back categories. For the legs, the test shows a depen-

dence of the results on the logging method (medium

effect size according to (Cohen, 2013)). This could be

due to the small number of participants or the larger

selection of legs poses, but perhaps also due to the in-

terface design of the app. Smaller devices may require

scrolling the interfaces, so perhaps observers were

more inclined to connect to the icons visible from the

beginning. This potential source of error should be

further investigated in future studies (e.g., with pro-

fessionals) and eliminated if necessary.

Overall, the results of the arms, legs, and back cat-

egories as a whole are inconclusive. The different re-

sults are probably due to the larger number of posture

classes in the legs category (7 vs. 3 or 4). A larger

number of subjects is necessary for a clearer conclu-

sion.

Usability of App. Based on the data obtained from

the UEQ questionnaire, it is possible to determine

which areas need improvement. For the app, it would

be the areas of Originality and Stimulation, which are

not conspicuously bad with 1.0 and 1.25, but lower

compared to the other areas. However, the follow-

ing can be said about the originality area: The app is

not designed to be novel or innovative. Moreover, the

use case is very speciﬁc, while the participants came

from ﬁelds where the OWAS method was not known.

An Android App for Posture Analysis Using OWAS

311

Table 3: Results of the User Experience Questionnaire for the App user group.

UEQ Scales Mean (SD)

Attractiveness (Overall impression of the product. Do users like or dislike it?) 1.524 (0.33)

Perspicuity (Is it easy to get familiar with the product and to learn how to use it?) 1.893 (0.60)

Efﬁciency (Can users solve their tasks without unnecessary effort? Does it react fast?) 1.607 (0.77)

Dependability (Does the user feel in control of the interaction? Is it secure and predictable?) 1.571 (0.66)

Stimulation (Is it exciting and motivating to use the product? Is it fun to use?) 1.250 (0.54)

Novelty (Is the design of the product creative? Does it catch the interest of users?) 1.0 (0.52)

Figure 3: UEQ - Benchmark-Diagram for the App.

None of the participants had to deal with body pos-

ture analysis before. Therefore, it can be assumed that

the paper protocol sheet was not known and therefore

no comparison could be made. Thus, it is not sur-

prising that the app scored lower on originality. As

described above, the stimulation area expresses how

high the motivation is to use the product. This score

could have been better as well. However, consider-

ing that the test subjects had to watch a video with

repeated 30-second pauses in which they had to wait,

the rather low value is not surprising. Moreover, the

APA app is designed for a work environment in which

the observers would be busy alongside the observa-

tions.

General Considerations. In further studies the re-

sults should be conﬁrmed in a work environment, to

ﬁnally conﬁrm the applicability of the APA app. In

addition, it might be worthwhile to investigate which

effect the level of training brings to the inter-rater re-

liability, as some of the common errors seem to be

well addressed by the app. In order to support such

research questions and enable other groups to bene-

ﬁt from the APA app, we decided to make the app

available as open source software. Thus, the source

code of the app can be found at https://github.com/

fudickar-lab/owas-app.

5 CONCLUSION

We presented an app as a support tool for posture

analysis using the OWAS method. The app was tested

for reliability with human subjects and the usability of

the app was evaluated. The results show that the app

approaches the usual pen-and-paper logging method

in terms of reliability and has higher inter-rater reli-

ability, at least in our cohort. It is suggested to con-

sider digital tools such as apps more for the analysis

of movements and postures. In the future, the app

could be extended to include other assessment meth-

ods apart from OWAS (e.g. REBA, RULA, etc.), and

in addition, different observation strategies could be

implemented and evaluated, e.g., with different sam-

pling frequencies. It may also be worthwile to in-

tegrated a semi-automatic classiﬁcation via the inte-

grated camera of the smartphone.

ACKNOWLEDGEMENTS

The author(s) would like to thank all study partici-

pants for their contribution.

REFERENCES

Amell, T. and Kumar, S. (2001). Work-related muscu-

loskeletal disorders: design as a prevention strategy.

A review. Journal of occupational rehabilitation,

11(4):255–265.

HEALTHINF 2023 - 16th International Conference on Health Informatics

312

Brandl, C., Mertens, A., and Schlick, C. M. (2017). Effect

of sampling interval on the reliability of ergonomic

analysis using the Ovako working posture analysing

system (OWAS). International Journal of Industrial

Ergonomics, 57:68–73.

Cohen, J. (2013). Statistical power analysis for the behav-

ioral sciences. Routledge.

Fleiss, J. L. (1971). Measuring nominal scale agree-

ment among many raters. Psychological Bulletin,

76(5):378–382.

Hoy, D., Brooks, P., Blyth, F., and Buchbinder, R. (2010).

The Epidemiology of low back pain. Best Practice

and Research: Clinical Rheumatology, 24(6):769–

781.

Karhu, O., H

ark

onen, R., Sorvali, P., and Veps

ainen, P.

(1981). Observing working postures in industry: Ex-

amples of OWAS application. Applied Ergonomics,

12(1):13–17.

Karwowski, W. and Marras, W. S. (1998). The Occupa-

tional Ergonomics Handbook -. CRC Press, Boca Ra-

ton, Fla.

Laugwitz, B., Held, T., and Schrepp, M. (2008). Construc-

tion and evaluation of a user experience questionnaire.

In Lecture Notes in Computer Science, pages 63–76.

Springer Berlin Heidelberg.

Lins, C., Fudickar, S., and Hein, A. (2021). OWAS inter-

rater reliability. Applied Ergonomics, 93:103357.

Lins, C. and Hein, A. (2022). Classiﬁcation of body pos-

tures using smart workwear. BMC Musculoskeletal

Disorders, 23(921).

Matsui, H., Maeda, A., Tsuji, H., and Naruse, Y. (1997).

Risk indicators of low back pain among workers in

Japan. Association of familial and physical factors

with low back pain. Spine, 22(11):1242–7; discussion

1248.

Schrepp, M., Hinderks, A., and Thomaschewski, J. (2017).

Construction of a benchmark for the user experience

questionnaire (UEQ). International Journal of Inter-

active Multimedia and Artiﬁcial Intelligence, 4(4):40.

Shneiderman, B. (2002). User Interface Design. - Effek-

tive Interaktion zwischen Mensch und Maschine. Leit-

faden f

ur intelligentes Schnittstellendesign. Was Pro-

grammierer und Designer

uber den Anwender wissen

ussen. mitp-Verlag, Frechen.

An Android App for Posture Analysis Using OWAS

313