Evaluating a Multi Depth Camera System to Consolidate Ergonomic
Work in the Education of Caregivers
Conrad Fifelski-von Böhlen
a
, Anna Brinkmann
b
, Sebastian Fudickar
c
,
Sandra Hellmers
d
and Andreas Hein
Assistive Systems and Medical Technologies, Carl von Ossietzky University, Ammerländer Heerstraße 140,
Oldenburg, Germany
sandra.hellmers@uol.de, andreas.hein@uol.de
Keywords: Nursing Education, Ergonomics, Teaching, Optical Devices, Feedback, Case Studies.
Abstract: Through the demographic change in Western European countries the demand for nurses in elderly care rises.
Additionally, constant high physical stresses causing nurses to leave the profession before the retirement age
and having a high number of sick leave days. To reduce the physical strain, focus must be set to ergonomically
correct work in nursing schools. Currently given technical infrastructure in the schools lacks the capability to
provide nursing instructors with analyzable data from simulated care acts. In this work, we present and
evaluate the Multi-Kinect-System, our custom developed depth sensor system for recording and analyzing
care acts. In a study, 13 students of a nursing school performed a simulated transfer tasks under observation
of a nursing instructor and our system. The instructor gives a more in-depth evaluation of the transfer when
using the intuitively analyzable data of our system, regarding the feedback length and information content.
1 INTRODUCTION
1.1 Demographic Background
The percentage of elderly people in Germany
increases in the next decades. In 2060 there will be
22.3 million people over 65 years, what is 33 % of the
German population. In comparison, in the year 2000
they made up only 17 % (Destatis, 2019). This
demographic change leads to an increasing demand
for care services, severing the situation in the nursing
institutions, which are overbooked even today (Kliner
et al., 2017; Weißert-Horn et al., 2014). Additionally,
the everyday work of the caregivers is accompanied
by constant physical and psychological overload. For
example, patient transfers are regarded as one of the
main factors responsible for back pain for elderly care
professionals (Weißert-Horn et al., 2014), leading to
a high amount of sick leave, compared to other
medical professions (Kliner et al., 2017), see figure 1.
a
https://orcid.org/0000-0002-6118-2755
b
https://orcid.org/0000-0001-5228-4947
c
https://orcid.org/0000-0002-3553-5131
d
https://orcid.org/0000-0002-1686-6752
Figure 1: Days of incapacity to work, due to
musculoskeletal diseases for employees in medical
professions (Kliner et al., 2017).
The impact of physical stress in this understaffed
profession is enormous, because a vicious circle
forms. High physical loads results in injuries, which
are the reason for sick leave, what reduces the staff
Böhlen, C., Brinkmann, A., Fudickar, S., Hellmers, S. and Hein, A.
Evaluating a Multi Depth Camera System to Consolidate Ergonomic Work in the Education of Caregivers.
DOI: 10.5220/0010186500390046
In Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 5: HEALTHINF, pages 39-46
ISBN: 978-989-758-490-9
Copyright
c
2021 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
39
and increases the pressure on the remaining staff.
Therefore, a reduction of the physical load for
caregivers is necessary to prepare the profession for
the aging society. A significant reduction of the
musculoskeletal stresses of nurses is achievable by an
ergonomically correct method of transferring patients
(Brinkmann et al., 2020; Jäger et al., 2014).
Furthermore, there is an urgent need for training the
use of care equipment (Rashidi and Mihailidis, 2013).
Even the usage of accessible care aids, like an ant-slip
mat during a manual patient transfer can measurably
reduce the physical load on the lower back (Jäger et
al, 2014).
1.2 Technology-based Approach
Typically, the nursing instructor attends the care acts
and gives feedback afterwards, possibly with visual
data from one standard camera. However, because of
the different poses, nurses are taking during care acts,
choosing different view angles is desirable. The usage
of an optical system that provides multiple viewpoints
and occlusion compensation for the analyzing nursing
instructor is desirable. A combination of multiple
depth cameras provides these features and to change
the point of view in recordings at any time.
Additionally, a three-dimensional scene is captured,
containing more information than a standard camera
image. An in-depth analysis of the care act is possible
with such a system. The care instructor and the
apprentice nurse have a better insight in the
ergonomics of the care transfer and can work together
on improvements. However, the technology-based
approach with the usage of multiple depth cameras
must be evaluated.
1.3 Related Work
Detecting and enhancing the compliance level of
health care professionals like nurses with best
practices in minimizing job-related back pain is a key
research field (Zhao et al., 2016). Studies track
specific activities of nurses that might increase the
risk of lower back pain injuries, like lifting and
pulling (Reichold et al., 2017; Zhao et al., 2016), by
videos or instructors by hand. From a technical point
of view, the depth camera Microsoft Kinect v2 is
evaluated regarding the accuracy of the depth image
and the point cloud data (Yang et al., 2015; Fifelski et
al., 2018; Fankhauser et al., 2015). The LiveScan3D
by Kowalski et. al. (Kowalski et al., 2015) is a system
that combines the depth images of multiple Kinects
V2 to display colored point clouds. Markers and the
Iterative Closest Point (ICP) algorithm (Besl and
McKay, 1992) are used to combine the generated
point clouds. It is shown that this system can be used
to scan a human head precisely enough to create 3D
printed figurines. The idea of training nurses in
specific scenarios or simulation-based is a common
and evaluated practice, for example in resuscitation
(Roh et al., 2013). Furthermore, the application of
technology in the education is not limited to acquire
data or providing scenarios. Approaches to analyze
training with algorithms, processing the gathered data
are possible (Reichold et al., 2017; Lins et al., 2018;
Lins et al., 2019). In order to assure a more distinct
evaluation of physical demanding nursing tasks, we
propose a System for multiview recording, with the
aim to provide high quality data for the nursing
instructor. As most algorithms are just estimating
positions of bodies or poses in the observed scenes,
the analysis of a care expert is necessary anyway. The
Multi-Kinect-System in this work provides higher
frame rates and is capable to process a larger area of
interest, compared to the mentioned approaches.
2 MATERIALS AND METHODS
2.1 Multi-Kinect-System
For the pose and movement analysis an optical
system provides the necessary data. Here the
combination of four Microsoft Kinect v2 depth
cameras is used. In three-dimensional, colored point
clouds it is possible to view the scene from indefinite
adjustable view angles. Because the point clouds are
merely three-dimensional objects, they can be rotated
or translated during visualization in real time or
offline. This is beneficial while observing care acts
like transfers, due to the possibility to see poses and
movements from different sides. A depth camera can
be occluded by obstacles or humans in the scene. A
combination of several depth cameras can
compensate the occlusion. Additionally, the point
clouds of multiple depth cameras can be combined to
a single point cloud. This increases the density of the
data and covers scenes from different sides. The depth
cameras are situated around the bed and are focusing
the center of the bed. All four cameras are installed
on tripods at 1.8 m and cover an area of
approximately 2 m * 2m. They tilted towards the
ground at 45 degrees. Research by Fankhauser et. Al
(Fankhauser et al., 2015) indicates that the best
distance from the point of interest to the camera is
around 0.8 to 1.2 meters. Also, the depth accuracy
deteriorates in the corners of the depth image
(Fankhauser et al., 2015). Therefore, the placement of
HEALTHINF 2021 - 14th International Conference on Health Informatics
40
the cameras needs to follow these findings. Each
Kinect v2 is connected to a mini pc via USB 3.0 that
runs Windows 10. The Kinect v2 Software
Development Kit (SDK) only runs on Windows
machines. Because the whole room network is ROS-
based the minicomputers are communicating to a
master computer running Ubuntu 16.04 and ROS via
private Ethernet connections. A server network PCI
card must be installed to handle the traffic. Contrary
to standard RGB cameras, the Microsoft Kinect v2 is
able of acquiring a depth image alongside with color
information. Although it is also possible to obtain
other data streams like infrared and a coarse body
estimation, we focus on depth and color information.
The depth information is obtained at a rate of 30 Hz
with 512 * 424 pixel, while the color image features
1920 * 1080 Pixel at 30 Hz. The Kinect v2 SDK
provides tools to map the color information to the
depth image. In short, all color pixels are discarded
for regions which are not in the depth image. This
reduces the resolution of the color image. The depth
data must be processed with a static Look-up-Table
(LUT), internal saved by each camera, in order to
calculate point clouds. This is necessary because each
Kinect v2 has slightly different lens distortion
parameters. The mapped color image stream and the
point cloud can be combined to a colored point cloud
on the master computer, see figure 2. The colored
point clouds are displayed for the analysis to the
nursing instructor.
Figure 2: The generation of point clouds. First, the depth
image and LUT are combined to a colorless point cloud.
Afterwards the color information is added. The resulting
colored point cloud of 512 * 424 with 4 bytes each point
must be computed 30 times per second per camera.
Another step, that must be done before using the
systems, is to register the four depth cameras to each
other. The 3D registration can be done in different
ways with printed patterns or objects of a special
shape. We use a 15 cm radius Styrofoam sphere. The
sphere is placed on at least three different spots in the
view of all cameras. This allows to calculate the
relative position from one camera to another. In the
visualization of the four aligned point clouds, the user
can adjust the viewpoint. Additionally, we
implemented a player that can be used like a standard
video player for point clouds, see figure 3. Even in the
recordings and when the recorded point cloud is
paused, it is still possible to change the point of view
at any time.
Figure 3: The point cloud player window. Its pixelated
appearance is caused by the depth camera resolution of 512
* 424. It is possible to play (1), stop (2), pause (3), rewind
(8), forward (9) or record (4) point clouds. The live button
(5) switches from recording to live data. To reduce the size
of the recorded files it is possible to apply voxel grid and
statistical outlier filters (6). The coarse body estimation can
be turned on, too (7).
2.2 Case Study
The study "stark" (Studie zur Transferanalyse
rückenschonender Pflegekonzepte) (Study for
Transfer Analysis of back-sparing Care Concepts) is
designed to compare traditional with multiview 3D
imaging feedback rounds. The study is approved by
the ethical board of the Carl von Ossietzky University
Oldenburg (Drs.EK/2019/004).The research question
is whether it is possible to identify the impact of the
Multi-Kinect-System on the second feedback round.
There are 13 study participants involved. To evaluate
a transfer regarding the Kinesthetic care conception
(Maietta and Hatch, 2011) and ergonomic ways of
work, typically, the nursing instructor supervises the
transfer and gives feedback afterwards. This
procedure can be enhanced and digitized through our
complex Multi-Kinect-System. To evaluate our
Multi-Kinect-System in the field, audio is recorded.
The recordings are done in a case study where a
nursing instructor uses the system while educating
nurses. This study compares feedback with and
without the usage of the Multi-Kinect-System. A
nursing instructor articulates which function of the
Multi-Kinect-System point cloud player should be
used, i.e. what the instructor wants to see. The indirect
usage reduces the influence of the usability of the
system. These commands are fulfilled by the study
organizer on the recording computer. The recordings
are containing key words, indicating the use of the
Multi-Kinect-System. The observed keywords are
"stop", "rewind", "forward", "turn view angle" and
words with a similar semantic. Furthermore, the time
Evaluating a Multi Depth Camera System to Consolidate Ergonomic Work in the Education of Caregivers
41
of the feedback rounds is compared, and the criticized
aspects are tracked. These aspects are body balance,
knee and lunge positioning, back, stand, usage of
caretakers movement resources and the whole
movement sequence. The transfer from a care bed to
a wheelchair is the observed task, the study
participant must fulfill. The patient is a 28-year-old
woman weighting 63 kg and already sitting on the
bed's edge. The patient is acting like a movement
impaired elderly person with abdominal stability
while standing. To accomplish the transfer correctly,
the participant must use the movement resources of
the patient, the care equipment and the learned
knowledge how to transfer movement impaired
people. Because the participants are differing in
height, strength and expertise, there is not one pattern
solution. Nevertheless, the participant should choose
the right grips and poses to apply forces according to
the nature of the patient’s body. This study relies on
nurses of the Evangelische Altenpflegeschule e.V.
Oldenburg nursing school. Although the participants
are still in the school courses, they are working in
nursing homes or similar institutions as caregivers.
Some of them are in the courses for further education
to obtain a higher qualification in the profession.
Therefore, they have a lot of practical experience and
are already instructed how to use their body and the
care equipment correctly. Nevertheless, they are
making mistakes during the transfer task, too.
Figure 4: The study setup. A wheelchair is situated beneath
a care bed. The transfer is carried out on the force
measurement plate in front of the bed.
Overall, 13 participants performed the transfer task.
The age of the 10 women and 3 men ranges from 18
to 55 years. There are no participants with physical
restrictions. The participants are entering the study
location and are prepared for the transfer task. In
addition to the examined Multi-Kinect-System, a few
other sensors are used, observing the performance of
the participant for other research interests. These
sensors are an electromyograph, a sensor suit (Motion
Workshop, 2020) and a ground reaction force plate
(AMTI, 2016). None of the used sensors are
impairing the movement of the participant. However,
this work focuses on the evaluation of the Multi-
Kinect-System. The wheelchair is placed on the right
side besides the care bed. Because the force
measurement plate elevating the participant 12 cm
over the ground, the wheelchair is always elevated on
a socket, see figure 4. The nursing instructor
supervises each transfer by visual inspection, but
without the data of the Multi-Kinect-System at first.
This feedback is recorded. Afterwards the same
transfer is examined again, but this time with the
Multi-Kinect-System. To see the data, the participant
and the instructor are gathering around the recording
computer. This second feedback round is recorded
again. The goal is to find and evaluate the differences
between the both feedback rounds. The comparison is
essential to derive how the sensor system extends the
perception of the nursing instructor.
Figure 5: The study procedure. Direct dependencies are
marked with solid lines and indirect ones with dotted lines.
The Multi-Kinect-System operator (MKS Operator)
provides the digital data and is therefore indirectly involved
in the digital feedback. Additionally, the nurse can review
the own performance in the digital feedback round and
therefore understand and react to critics.
Observing the own performance in the scenario
might be educational for the nurse. The study
organizers are just involved as passive participants
(DeWalt and DeWalt, 2011). The three-dimensional
data is presented to the nurse and the nursing
instructor. A study organizer is operating the system,
following their commands by changing the zoom or
the viewpoint and pausing, rewinding or forwarding
the recorded data. Figure 5 depicts the procedure. The
resulting data set, which must be analyzed, consists
of two audio recordings per study participant.
HEALTHINF 2021 - 14th International Conference on Health Informatics
42
3 RESULTS
The data gathered during the feedback rounds with
and without the Multi-Kinect-System is audio data.
Therefore, the task is to designate features, which are
differing in both feedback rounds. The following
features are relevant: the duration, special keywords,
which are indicating the use of the Multi-Kinect-
System and the proposed improvements. Each of the
13 study participants is represented by a three-digit
number. Note, that the order of the numbers in the
following graphs and tables is not the order in which
they performed the transfer task.
3.1 Duration of the Feedback round
Based on the point cloud recordings, the feedback
round with the use of the Multi-Kinect-System must
take at least longer than the feedback without the
system to reason the usefulness of the system. We
assume that a higher duration of a feedback round
indicates a possible presence of additional
information. However, the information of the both
feedback rounds is examined later in addition to the
plain duration here. Table 1 compares the duration of
the audio recording with and without the Multi-
Kinect-System. Figure 6 depicts these differences in
the duration.
Table 1: The duration of the analog (TA) and digital (TD)
feedback rounds for each study participant (SP) in minutes.
TA average is 0:59, TD average is 2:16.
Figure 6: The different time duration of the feedback rounds
in seconds for each Study participant.
3.2 Keywords in the Feedback round
with the Multi-Kinect-System
The audio recordings are indicating the use of the
Multi-Kinect-System. There are keywords like
"stop", "rewind", "forward", "turn view angle" and
words with a similar semantic, which can be
quantified. Table 2 shows the occurrences of the
keywords for each study participants, who are
represented by a three-digit number.
Table 2: The occurrences of keywords (#K) for each study
participant (SP).
3.3 Amount of Criticized Aspects
The overall aim is still to improve the caregivers
movements, respecting Kinaesthetics (Maietta and
Hatch, 2011) care conception and ergonomic
approaches. Therefore, it is necessary to compare the
improvements and the critic, the nursing instructor
suggests in the both feedback rounds. The six aspects,
which are the most important for a healthy transfer
regarding the caregiver, are examined. These aspects
were chosen according to the nursing instructor:
body balance (BAL)
knee and lunge positioning (LNG)
usage and form of the back (BAK)
stand on the ground (STD)
usage of the caretakers movement
resources (RES)
movement sequence (MVM)
Body balance is important during the transfer to
control the weight of the caretakers at all time. To
shift the load from the back to the lower extremities a
lunge position alongside with the right knee angle is
desirable. Both, the back and the neck should be
stretched, so that the lifting is not performed with a
bent or twisted spine. The most relevant body part is
the back, as the loss of working time is caused by back
pain. This is also related to the positioning of the feet
to get a stable stand. A too narrow stand reduces the
stability. Because all transfers should take the
caretakers movement resources into account, to avoid
unnecessary loads, this aspect is mentioned during the
feedback rounds. The right movement patterns and
sequences should avoid unfavorable activities like the
torsion of the spine. In the first feedback round, the
audio is recorded, and the criticized aspects can be
counted. Afterwards the transfer is observed using the
Multi-Kinect-System. The following table 3 presents
0 s
50 s
100 s
150 s
200 s
250 s
142
243
424
471
483
107
280
303
350
366
315
437
473
Analog Feedback Digital Feedback
SP 142 243 424 471 483 107 280 303 350 366 315 437 473
TA 0:57 0:43 0:59 0:36 1:00 1:09 1:41 1:29 0:38 1:10 0:38 1:01 0:46
TD 4:06 2:58 2:01 2:09 1:55 1:28 2:07 3:37 2:43 1:27 1:07 2:33 1:20
SP 142 243 424 471 483 107 280 303 350 366 315 437 473
#K 3332151 1 2 1 1 21
Evaluating a Multi Depth Camera System to Consolidate Ergonomic Work in the Education of Caregivers
43
the occurrences of the mentioned critics for each
study participant (SP) for both feedback rounds.
Table 3: The criticized aspects in the analog and digital
feedback. The values are separated by a slash. BAL: body
balance, LNG: knee and lunge positioning, BAK: back,
STD: stand, RES: usage of the caretaker’s movement
resources, MVM: movement sequence. Average of SUM
(sum) of analog feedback round is 2.31 (overall 30) and
average of digital feedback round is SUM is 3.38 (overall
44). SP is the study participant.
SP BAL LNG BAK STD RES MVM SUM
142 1/1 0/2 0/1 1/2 1/0 1/2 4/8
243 0/1 0/0 0/0 0/0 0/2 0/1 0/4
424 0/1 1/0 1/1 0/0 1/1 0/1 3/4
471 1/1 1/0 0/2 0/0 1/0 0/0 3/3
483 1/1 0/0 0/0 0/0 1/1 0/0 2/2
107 1/1 0/1 0/1 0/0 0/0 1/0 2/3
280 2/2 2/2 1/3 0/1 0/0 0/1 5/9
303 0/0 1/0 1/1 0/1 0/0 0/0 2/2
350 0/0 1/0 1/2 0/1 0/0 1/0 3/3
366 0/0 1/0 0/0 0/0 0/0 1/0 2/0
315 0/0 0/0 0/0 0/ 0/1 0/0 0/1
437 1/1 1/1 0/1 0/0 1/1 0/0 3/4
473 0/1 1/0 0/0 0/0 0/0 0/0 1/1
3.4 Information Content
The plain duration of the feedback solely is no prove
regarding the information content. However, when
using the feedback duration as an indicator for a more
in-depth debriefing, while also considering the
amount of criticized aspects in the transfer, denser
information can be assumed while using the Multi-
Kinect-System. The sum of the criticized aspects is
2.31 with analog feedback and 3.39 with digital
feedback on average, while the average analog
feedback time is 59 seconds and digital is 2 minutes
and 16 seconds. For example, for the study
participant 142, the feedback time increased by 3
minutes and 9 seconds between analog feedback and
the feedback with the Multi-Kinect-System. Dividing
the average feedback time by the average amount of
criticized aspects, the result is 25 seconds per
criticized aspect for analog feedback and 40 seconds
per criticized aspect for digital. This can either
indicate that less time per mentioned aspect is needed
in the analog feedback or the digital feedback round
is more in depth through the possibilities of the Multi-
Kinect-System. However, the amount of mentioned
or criticized aspects should always have more priority
than duration concerns. Note that the indirect usage
of the Multi-Kinect-System also consumes a one-
digit number of seconds per participant.
3.5 Discovery of Aspects in the Digital
Feedback
Looking into the data of the point cloud player for
study participant 107, additional aspects regarding the
lunge positioning and one additional aspect regarding
the usage of the back were discovered by the nursing
instructor. The usage of the Multi-Kinect-System
makes additional aspects visible in this and other
cases. Because the analog feedback is always the first
to take place, it is more likely, that issues, discussed
in the analog feedback, are not mentioned again in the
digital feedback round. Indeed, the amount of
criticized aspects in table 3, is higher in some classes
for some of the study participants, than in table 4.
However, the overall count of criticized aspects
increases significantly between analog and digital
feedback rounds.
4 DISCUSSION
The different findings in the feedback rounds are
significant and must be researched further.
Additionally, the care instructors need training to use
the proposed software on their own, without further
help. A possible impact on the overall usefulness of
the system and the benefit for the nursing instructor
and the student seems to be the usability. Although
there are no complaints about the data quality of the
depth data, it is planned to improve the technical
parameters of our Multi-Kinect-System. Improvable
aspects are resolution, accuracy and usability with the
use of new hard and software. Furthermore,
additional studies must be established with more
participants and specialized Kinaesthetics (Maietta
and Hatch, 2011) trainers. More education facilities
should be involved in testing our system in their
education program. This helps to install this system
in the everyday education of nurses in elderly care,
probably increasing their competences and helping
nursing instructor to detect unfavorable poses and
movements in the education process earlier.
Additionally, the usage of automatic pose evaluation
algorithms (Stoffert, 1985) is planned to support the
instructor’s feedback. More in-depth analysis of
characteristics of unfavorable poses could also lead to
a better understanding why they are common. A
future goal is to integrate our system as an inherent
part in the education of nurses.
HEALTHINF 2021 - 14th International Conference on Health Informatics
44
5 CONCLUSIONS
A case study was executed for the evaluation of our
Multi-Kinect-System to quantify the benefits of
technology-based education enhancements. In this
first approach, we use participating observation and
audio recordings during two types of feedback rounds
after a care scenario, completed by the study
participants. One feedback is given by a nursing
instructor without advanced visual 3D recording data
and one feedback is given with this data. These
feedback rounds are differing significantly in various
parameters. We identified meaningful features, which
are indicating the usefulness of our system in the
education of nurses in elderly care.
ACKNOWLEDGEMENTS
This work is carried out in the research project
ITAGAP, funding number: 02L14A240 of the
Bundesministerium für Bildung und Forschung
BMBF (German Federal Ministry for Education and
Research) at the Evangelische Altenpflegeschule e.V.
Oldenburg (Evangelical Nursing School).
REFERENCES
AMTI, 2016. Accupower force platform, accessed: 2020-
03-18. URL
http://www.accupowersolutions.com/force-platforms.
Besl, P.J., McKay, N.D., 1992. A method for registration of
3-d shapes. In IEEE Transactions on Pattern Analysis
and Machine Intelligence 14 (2), 239-256,
https://doi.org/10.1109/34.12179.
Brinkmann, A., Fifelski, C., Lau, S., Kowalski, C., Meyer,
O., Diekmann, R., Isken, M., Fudickar, S., Hein, A.,
2020. The AAL/Care Laboratory A Healthcare
Prevention System for Caregivers. In Nanomaterials
and Energy 9,1, 27-38,
https://doi.org/10.1680/jnaen.19.00021.
DeWalt, K.M., DeWalt, B.R., 2011. Participant
Observation: A guide for fieldworkers, Alta Mira.
Plymouth.
Fankhauser, P., Bloesch, M., Rodriguez, D., Kaestner, R.,
Hutter, M., Siegwart, R., 2015. Kinect v2 for mobile
robot navigation: Evaluation and modelling. In 2015
International Conference on Advanced Robotics
(ICAR). Istanbul, 388-394,
https://doi.org/10.1109/ICAR.2015.7251485.
Destatis, 2019. Federal office of statistics website,
accessed: 2020-03-18. URL
https://service.destatis.de/bevoelkerungspyramide/#!y
=2050.
Fifelski, C., Brinkmann, A., Ortmann, S.M., Isken, M.,
Hein, A., 2018. Multi depth camera system for 3d data
recording for training and education of nurses. In 2018
International Conference on Computational Science
and Computational Intelligence (CSCI). Las Vegas,
NV, 679-684,
https://doi.org/10.1109/CSCI46756.2018.00137.
Jäger, M., Jordan, C., Theilmeier, A., Wortmann, N., Kuhn,
S., Nienhaus, A., Luttmann, A., 2014. Analyse der
Lumbalbelastung beim manuellen Bewegen von
Patienten zur Prävention biomechanischer
Überlastungen von Beschäftigten im
Gesundheitswesen. In Zentralblatt für Arbeitsmedizin,
64, 98-112,
https://doi.org/10.1007/s40664-013-0010-4
Maietta, L., Hatch, F., 2011. Kinaesthetics Infant Handling,
Hans Huber. Bern, 2
nd
edition.
Kliner, K., Rennert, D., Richter, M. (Eds), 2017.
Gesundheit und Arbeit – Blickpunkt Gesundheitswesen.
Medizinisch Wissenschaftliche Verlagsgesellschaft.
Berlin.
Kowalski, M., Naruniec, J., Daniluk, M., 2015. Livescan3d:
A fast and inexpensive 3d data acquisition system for
multiple kinect v2 sensors. In 2015 International
Conference on 3D Vision. Lyon, 318-325,
https://doi.org/10.1109/3DV.2015.43.
Lins, C., Eckhoff, D., Klausen, A., Hellmers, S., Hein, A.,
Fudickar, S., 2019. Cardiopulmonary resuscitation
quality parameters from motion capture data using
differential evolution fitting of sinusoids. In Applied
Soft Computing Journal 79, 300-309,
https://doi.org/10.1016/j.asoc.2019.03.023.
Lins, C., Klausen, A., Fudickar, S., Hellmers, S., Lipprandt,
M., Röhrig, R., Hein, A., 2018. Determining
cardiopulmonary resuscitation parameters with
differential evolution optimization of sinusoidal curves.
In Proceedings of the 11th International Joint
Conference on Biomedical Engineering Systems and
Technologies, Oldenburg, 665-670,
https://doi.org/10.5220/0006732806650670.
Motion Workshop, 2020. Motion capture system - shadow
motion sensor suit, accessed: 2020-03-18. URL
http://www.motionshadow.com.
Rashidi, P., Mihailidis, A., 2013. A survey on ambient-
assisted living tools for older Adults. In IEEE Journal
of Biomedical and Health Informatics 17 (3), 579-590,
https://doi.org/10.1109/JBHI.2012.2234129.
Reichold, J., Agrawal, A., Thurlings, M., Cohen, I., Weber-
Fiori, B., Rölle, A., Hassan, M., Dürr, M., Pfeil, U.,
Pape, A.-A., Grünert, G., Schmidt, A., Pfeil, M., Fäßler,
Jauch, V.V., Reiterer, H., Winter, M., Ertel, W.,
Eberhardt, J., 2017. Human-machine interaction in
care-education. In Burghardt, M., Wimmer, R., Wol,
C., Womser-Hacker C. (Eds), Mensch und Computer
2017: Tagungsband. Books on Demand. Norderstedt,
351-356, https://doi.org/10.18420/muc2017-ws08-
0311.
Roh, Y.S., Lee, W.S., Chung, H.S., Park, Y.M., 2013. The
effects of simulationbased resuscitation training on
nurses' self-efficacy and satisfaction. In Nurse
Evaluating a Multi Depth Camera System to Consolidate Ergonomic Work in the Education of Caregivers
45
Education Today 33 (2), 123-128,
https://doi.org/10.1016/j.nedt.2011.11.008.
Stoffert, V., 1985. Analyse und Einstufung von
Körperhaltungen bei der Arbeit nach der Owas-
Methode. In Zeitschrift für Arbeitswissenschaft 1, 31-
38.
Weißert-Horn, M., Meyer, M.D., Jacobs, M., Stern, H.,
Raske, H.-W., Landau, K., 2014. Ergonomisch richtige
Arbeitsweise beim Transfer von
Schwerpflegebedürftigen. In Zeitschrift für
Arbeitswissenschaft 68 (3), 175-184,
https://doi.org/10.1007/BF03374443.
Yang, L., Zhang, L., Dong, H., Alelaiwi, A., Saddik, A.E.,
2015. Evaluating and improving the depth accuracy of
kinect for windows v2. In IEEE Sensors Journal 15 (8),
4275-4285,
https://doi.org/10.1109/JSEN.2015.2416651.
Zhao, W., Lun, R., Gordon, C., Fofana, A.B., Espy, D.D.,
Reinthal, M.A., Ekelman, B., Goodman, G.,
Niederriter, J., Luo, C., Luo, X., 2016. A privacyaware
kinect-based system for healthcare professionals. In
2016 IEEE International Conference on Electro
Information Technology (EIT), Grand Forks, ND, 205-
210, https://doi.org/10.1109/EIT.2016.7535241.
HEALTHINF 2021 - 14th International Conference on Health Informatics
46