comparison between a computational seated human model and

Applied Bionics and Biomechanics 11 (2014) 175–183DOI 10.3233/ABB-140105IOS Press

175

Comparison between a computational seatedhuman model and experimental verificationdata

Christian G. Olesena,∗, Mark de Zeeb and John Rasmussena

aDepartment of Mechanical and Manufacturing Engineering, Aalborg University, Aalborg, DenmarkbDepartment of Health Science and Technology, Aalborg University, Aalborg, Denmark

Abstract. Sitting-acquired deep tissue injuries (SADTI) are the most serious type of pressure ulcers. In order to investigatethe aetiology of SADTI a new approach is under development: a musculo-skeletal model which can predict forces betweenthe chair and the human body at different seated postures. This study focuses on comparing results from a model developed inthe AnyBody Modeling System, with data collected from an experimental setup. A chair with force-measuring equipment wasdeveloped, an experiment was conducted with three subjects, and the experimental results were compared with the predictionsof the computational model. The results show that the model predicted the reaction forces for different chair postures well. Thecorrelation coefficients of how well the experiment and model correlate for the seat angle, backrest angle and footrest heightwas 0.93, 0.96, and 0.95. The study show a good agreement between experimental data and model prediction of forces betweena human body and a chair. The model can in the future be used in designing wheelchairs or automotive seats.

Keywords: Wheelchair, musculoskeletal model, pressure ulcer, validation, chair reaction forces

1. Introduction

Pressure ulcers, more commonly known as pressuresores, are a frequent complication to spinal cord injury(SCI) patients. The aetiology of the disease is in generalpoorly understood [22].

Statistics show that 24% of all patients with SCIexperience a pressure ulcer during their rehabilitationhospital stay [7]. It is also estimated that 50–85%of all patients with SCI will experience a pressureulcer during their life time [27]. These are very gen-eral prevalences covering a range of different typesof pressure ulcers. The type of pressure ulcers thathas motivated the present investigation is the sitting-acquired deep tissue injury (SADTI) that wheelchair

∗Corresponding author: Christian G. Olesen, AnyBody ResearchGroup, Fibigerstraede 16, DK-9220 Aalborg E, Denmark. Tel.: +4599403355; Fax: +45 98151675; E-mail: [email protected].

users, i.e. paraplegic and quadriplegic patients are par-ticularly susceptible to. [1] The prevalence of deeptissue injury is difficult to assess because it is usuallyonly detected after it has reached the skin surface, atwhich point the injury’s origin is impossible to estab-lish. SADTIs have a tendency to spread under theskin and reach proportions that are difficult to treatand in some cases terminal [11]. For a more generalintroduction to the field of pressure ulcer research hasbeen combined in a comprensive book by Bader, D.et al. [3]

Investigations of the aetiology behind pressureulcers is broad and ranges from the seated postureover global loading of the buttocks to tissue stressesand strains and further on to cell deformation causingnecrosis. It is well acknowledged that pressure ulcersare primarily caused by sustained mechanical loadingof the soft tissues [26, 28]. The types of loading canbe described as pressure, pressure gradients and shear

1176-2322/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved

mailto:[email protected]

176 C.G. Olesen et al. / Comparison between a computational seated human model and experimental verification data

forces [16]. These loads conspire in a complicated fash-ion to generate deformation states varying from point topoint in the soft tissues. An understanding of the inputloads is therefore the first step towards a full under-standing of pressure ulcer formation. The magnitudeand position of the loads are influenced by the patient’sposture and the support conditions of the wheelchair.Several research groups have contributed to the under-standing of the forces between a seated human andits environment, but the difference in aims and exper-imental protocols make them impossible to compare[4, 8, 13–16, 20, 25]. A recent review of the currentliterature within the field has pointed out the need tounderstand the forces acting on the buttocks for dif-ferent seated postures [22]. It is important to realisethat those support forces also partly depend on howthe muscles are activated. It is therefore not enough toestimate the support forces purely base on gravity andthe segment masses. The estimated support forces cansubsequently be used as global boundary conditionsfor FE models of the buttocks, which subsequentlycan calculate the internal strains and stresses in thetissue under the influence of various cushion materi-als. Several papers have described how the mechanicsof the soft tissue on the buttocks react towards loading,but none have investigated how systematic change inposture change the deformation of the tissue [12, 18,19, 29].

A computational approach might contribute fur-ther to an understanding of how the seating postureaffects the mechanical loading of the soft tissue in thebuttock region. Such an understanding might also ben-efit designers of automotive, airline and office seatsbecause of the contact forces’ contribution to the per-ceived discomfort [6, 23, 24].

A validated computational model could be usedto predict the loading on the human body from dif-ferent seated postures without the need for costlyexperiments. Furthermore, an analytical model is freefrom the inevitable experimental noise and statisticalvariations due to individual differences between sub-jects and differences in protocols that may otherwiseocclude the interpretation of results.

However, before using any model it should be com-pared to experimental data in order to verify thepredicted forces [21]. Therefore the objectives of thisstudy are to make an initial verification of a musculo-skeletal model with respect to its ability to predictthe chair reaction forces and how these are changedfor different seated postures. This is accomplished by

comparing force predictions from a model with mea-surements from test subjects.

2. Methods and materials

The study design comprised an experimental anda modelling part. The experimental setup includedmeasurements of reaction forces on a wheelchair mea-sured in synchronization with motion capture data forrecording of the sitting posture of three volunteers.A musculo-skeletal model was created to mimic theseated position and thereby estimate reaction forcesbetween the chair and the model. These reaction forcescould then be compared with the forces measured dur-ing the experiment and in that way make an initialverification of the musculo-skeletal model. For a studyoverview, see Fig. 1.

Fig. 1. Flow chart of the study design. Recording and post pro-cessing of experimental data using a Matlab script, transfer of thekinematic data to the musculo-skeletal model for simulation of forcesand finally comparison between measured and calculated forces.

C.G. Olesen et al. / Comparison between a computational seated human model and experimental verification data 177

2.1. Experimental setup

2.1.1. Force measurementsA custom-built wheelchair, see Fig. 2 (Woltur-

nus A/S, Nibe, Denmark) was mounted withforce-measuring equipment (Advanced MechanicalTechnology, Inc., Watertown, MA, US). The chair wasconstructed by mounting an OR6-7-1000 force plate asa seat, as a footrest an OR6-7-2000 force plate mountedin the floor was used. The backrest consisted of twohorizontal bars that in each side were mounted to twomulti-axis force and torque transducers. In total, fourmulti-axis force transducers (2 × FS6-250 & 2 × FS6-500) were mounted in the two backrest bars. Thebackrest bars were covered with 4 mm foam and theseat was covered with a 2 mm rubber mat.

The wheelchair could be adjusted in a numberof ways. The inclination angle of the seat andbackrest could be changed. The seat could slideforward/backward. The two backrest bars could indi-vidually slide up/down and forward/backwards. Theseadjustments allowed the wheelchair to be adjusted con-tinuously into any position with respect to backrestheight, seat depth, seat angle and backrest angle. Theseat and backrest angles were reported as angles indegrees measured from horizontal seat, turning aroundthe intersection between seat and backrest, measuredpositive rotating from seat to backrest. The forcesapplied to the force plates and force transducers wereamplified using amplifiers from the same manufacturer

as the force measuring equipment. The amplified sig-nals were sampled using Texas Instrument 16 bit A/Dconverters, connected to a computer equipped withdata collection software (Mr.Kick, Aalborg University,Aalborg, Denmark) [17]. The six channels on eachforce-measuring device represented the 3 forces and3 moments, they were all sampled giving a total of 36channels. The sampling rate was 20 Hz over 10 secondsgiving 200 samples per channel.

2.1.2. Posture assessmentThe chair positions were measured using a motion

capture system (Qualisys Proreflex 240, Gothenburg,Sweden) with eight cameras. Passive reflective mark-ers were placed in each corner of the two force platesand on both sides of the two backrest bars. The postureof the subject sitting in the wheelchair was also mea-sured using passive reflective markers. The markerswere placed at the following anatomical landmarks:Forehead (glabella), sternum, the lateral tip of theacromion, the lateral elbow epicondyle), hand (Ossesamoideum of digitus tertius), pelvis (Crista iliaca),greater trochanter, knee (Caput fibulae) and lateralmalleolus.

The positions of the markers were measured with asampling rate of 20 Hz over 10 seconds.

The force measurements and the motion capture sys-tem were started by a trigger in order to synchronizethe force and motion capture recordings.

Fig. 2. The experimental setup with a subject sitting on the instrumented chair. The white dots are the reflective markers used for motion capture.


2.1.3. Conducting the experimentThree able-bodied men of 179 ± 3 cm height and

68 ± 2 kg body mass were included in the study.The subjects’ height, mass and height from floor to

the posterior side of the knee while seated were mea-sured. The posterior knee height was used as a guidefor an appropriate seat height, i.e. the distance fromthe footrest force plate to the front top edge of theseat. This distance (H∗) was kept constant throughoutall experiments, except for the experiment varying theseat height. The subject was seated in the chair and themarkers were attached. The subject was instructed tosit as relaxed as possible with the arms crossed overthe chest in order to make sure they did not affect theresult. The force plates and load cells were reset everytime the posture of the chair was changed to make surethat the baseline was the same for all measurements.The experiment included 12 different seated postures.

The seat angle was varied from −17◦ to 6◦, while thebackrest angle was maintained 108◦ leaned back, andthe footrest height adjusted relatively to the subject’slower leg length (H∗).

The backrest angle was varied from approximately90◦ corresponding to vertical, and reclined to approx-imately 123◦. The seatpan was kept constant in 7◦leaned backwards, and the footrest height adjusted rel-atively to the subject’s lower leg length (H∗).

The distance from the edge of the seat to the footrestvaried from lower leg length plus 2 cm to lower leglength minus 8 cm. while the backrest angle was keptconstant at 108◦ leaned back, and the seat angle waskept constant at 7◦ leaned back. Between each of theposture variations, only one parameter was changedand the subject sat in the chair 4 minutes before 10seconds measurement was started to ensure that theseated position had reached a steady state as describedby Crawford et al. [8] The continuous adjustments andthe fitting of the seat to the different anthropometriesof the test subjects meant that different postures wererecorded for the three subjects.

2.2. Post processing

2.2.1. Motion captureThe motion capture data from the Qualisys cam-

eras were initially post processed in QTM (QualisysTracking Manager V.1.10.282) where the markers wereidentified and named. The markers were then exportedto a TSV file and then processed using a custom-madeMatlab (MatLab R2008B, Mathworks Inc., Ma, US.)

script that based on marker coordinates, defined vec-tors representing the various parts of the chair. Anglesbetween the vectors representing the seat, backrest barsand footrest was also calculated. The output of the Mat-lab scripts was a text file that could be included in theAnyBody model. The text file contained all the markerpositions for driving the model. For an overview of theprocess, see Fig. 1.

2.2.2. Force dataThe collected force data was gathered using Mr.

Kick saving each data file as a Matlab data file. A scriptwas used for averaging the values. A Matlab scriptcombined the motion capture data with the measuredforces.

2.2.3. Musculoskeletal modelThe musculoskeletal analysis was performed in the

AnyBody Modeling System version 4.1 (AnyBodyTechnology A/S, Aalborg, Denmark). The muscu-loskeletal model is based on the “Seated Human”model from the open source AnyBody model reposi-tory [2], which was described in [9, 23]. The AnyBodyModeling System is computer software designed forconstructing musculo-skeletal models of the humanbody and its environment and for determining howthese interact. With this kind of models it is possi-ble to estimate muscle activities, joint reaction forcesand unknown interaction forces with the environmentusing optimization algorithms. The analysis is basedon inverse dynamics, and in order to solve the musclerecruitment problem it is assumed that the muscles areactivated in an optimal way in order to avoid fatigue.Moreover the software is based on multibody dynamicsassuming that the segments are rigid. The mathemat-ics and mechanical theories behind it were describedin detail by Damsgaard et al. [10].

The model was scaled in mass to the body mass ofeach test subject. The subjects were approximately thesame anthropometrical size as the generic model in thepublic repository,

The model consists of two parts: a human modeland an environment representing the wheelchair. Theseated model relies on a set of assumptions described in[24]. In addition, a few assumptions were made aboutthe interface between the chair and the human body:

Contrary to real seated persons, rigid multibodymodels are by nature supported on points rather thansurfaces and by reaction forces rather than by pressuredistributions. The model has therefore been equipped


with a number of points through which it can trans-fer reaction forces to the supporting elements, i.e. thebackrest, seat pan and footrest. The contact betweenthe body and the supporting elements is modelled bymeans of contact elements that can provide only com-pressive reactions and shear forces implemented asCoulomb friction in the contact points. The modelcould choose to use any set of the contact points avail-able, In principle this could be compared to modellingthe perfect supporting chair within the boundary con-ditions of the seat.The friction coefficients �, betweenthe body and the seat and footrest was estimated to� = 0.5 based on preliminary tests. For the backrest� = 0.1 was used.

2.3. Comparison

Comparing absolute force values and trends whenchanging an input parameter did the comparisonbetween the experimentally obtained forces and theestimated forces from the musculoskeletal model. Theforce values were plotted as a function of the variableparameter. The forces used for comparison were theseat shear forces and the normal force on the footrest.Seat shear force was considered positive when forceapplied to the skin was in the forward direction, seeFig. 2, The forces on the backrest relates directly to theforces on the seat, therefore these were not reported.

The normal force of the seat was not considered in thisstudy.

3. Results

The results from the forces measured during theexperiment and the estimated forces from the mus-culoskeletal model were compared as absolute valuesand trends of these while changing one parameter atthe time. The seat shear force was considered positivein the frontal direction.

3.1. Effect of changing the backrest angle

Figure 3 illustrates the experimental and modelresults for the three subjects for the backrest anglevariation. The results indicate that backrest inclinationincreases the seat shear force. Linear regression linesfor each of the datasets reveal that the slopes differ by18.8 %. This corresponds to an absolute difference ofmaximally 20 N at the two extremes because the twocurves intersect very closely to 0 N. Please notice, how-ever that despite the addition of the linear regressionlines, nothing in the physics of the problem indicatesthat the behaviour should be linear. The correlationcoefficient, R2-value and confidence interval of howwell the model describes the experiment can be foundin Table 1.

Fig. 3. The effect on the seat shear force from varying the backrest angle. The data show clear correspondence and a trend towards higher seatshear force when the backrest is inclined.


Table 1The seat angle, backrest angle and footrest height for the different experiments. The diagonal of ranges were the parameters changed in each of

the experiments

Seat angle Backrest angle Footrest height

Seat angle variation −17◦– (+6◦) 108◦ H∗Backrest angle variation 7◦ 90◦–123◦ H∗Footrest height variation 7◦ 108◦ (H∗-8 cm) – (H∗+2 cm)

3.2. Effect of changing the seat angle

Figure 4 shows that changing the seat angle fromtilting forward to tilting backwards while keeping theother variables constant has a strong influence on theseat shear force. In Fig. 4 experimental and modellingresults have been pooled for the three subjects, anda linear regression line has been plotted for each ofthe datasets. The slopes differ by 19.8 % leading to amaximum deviation between the linear regressions ofexperimental data versus model data of approximately20 N. The correlation coefficient and confidence inter-val for the experimental and model of the seat anglecan be found in Table 2.

3.3. Effect of changing the height of the footrest

The height of the footrest has an effect on the seatnormal force predicted by the model. Figure 5 showsthe difference between the measured and the predictednormal force on the footrest. Most of the data points arewithin ±20 N and the standard deviation of the error

is 15.5 N. The correlation coefficient and confidenceinterval for the experimental and model of the footrestheight can be found in Table 1.

4. Discussion

The results quantify the correspondence between thecomputational musculoskeletal model and the exper-imentally obtained data. It was found that the seatshear force and the footrest normal force were the onesthat varied the most when changing posture, the otherforces and moments did not vary much not in the exper-iment, nor in the model predictions, therefore theseforces were the main focus point in the result section.

Varying the seat angle, backrest angle and footrestheight in general showed good results and the modelpredicted the forces acting between the chair and thesubjects. In each experiment the goal was to vary a sin-gle parameter, but due to (i) the continuous adjustmentof the chair, (ii) the fitting of the chair to each subject’santhropometry, (iii) the small postural adjustments ofthe subject in each trial and (iv) the different choices

Fig. 4. The effect on the seat shear force when the seat angle is changed. The data show clear correspondence and similar trends can be observedas well.


Table 2The correlation coefficients, the confidence interval and the RMS error for the seat normal and shear force prediction for each of the three

experiments

Seat angle Backrest angle Footrest height

Correlation coefficient 0.93 0.96 0.95R2-value 0.87 0.91 0.9195% Confidence interval 0.71–0.99 0.84–0.99 0.82–0.99RMS Seat Fs Error 20.0 N 13.6 N 13.7 NRMS Seat Fn Error 22.3 N 38.4 N 35.4 N

Fig. 5. The difference between the measured and predicted normal force on the footrest. The different data points are for the different experimentswhere the height of the footrest was changed.

of posture in a given chair setup between the subjects,there was small deviation in the chair setup and realizedposture between each of the experiments. However, themajority of this variation was transferred to the modelvia the kinematic input from each of the experimentsto the model.

Other researchers have conducted experiments thatcan be compared with the present results. For exampleGildorf et al. [13] mounted a force plate as a wheelchairseat and measured normal and shear forces for differ-ent backrest angles. They found the average shear forceat the seat for a hard surface at 5◦ incline to be 27 N,which corresponds well with this study, indicating thatthe test subject was typical. In another study conductedby Bush & Hubbard [5] 12 healthy mid-sized maleswere seated in an instrumented chair in different seatedpostures, and normal and shear forces were measured.The experimental setup was somewhat different, butapproximately 15 N seat shear force was measured in

a neutral posture with the backrest reclined 20 degrees.This could be compared with the graph on Fig. 3, pre-dicting between 10 and 20 N seat shear force for thesame backrest angle.

Also Hobson [16] did experimental work on dif-ferent seated postures on spinal cord injured andable-bodied subjects. Their results are difficult to com-pare, since the protocol used in this study, is not similar,however some comparison can be made between seatedpostures where there are only a few degrees differencein seat- backrest angle. Posture “P4” where the seat ishorizontal and the backrest is leaned back to 110◦ issomewhat comparable to the seat angle variation exper-iment in this study. Hobson, D.A. measured a mean70 N shear force, where the experimental results fromthis study was 75–80 N.

The purpose of the seated musculo-skeletal modelis to give detailed information on how different seatedpostures affects reaction forces between the chair and


the human body. It is a model of how parametersinteract therefore verifying trends is an important mea-sure of model quality [21]. The correlation coefficientbetween predicted and measured shear force with back-rest angle variation was 0.9580 with a 95% confidenceinterval between 0.8419–0.9893 indicating that with95% certainty the model can describe at least 84% ofthe experimental result. Similar results were found forvariation of the footrest height, while the correlationcoefficient for variation of the seat angle was 0.9335.Most likely, with more experiments, the confidenceinterval would become smaller and thereby the cer-tainty of how well the model describes the experimentwould be larger. The R2 values for the 3 experimentsshow that the difference between the experimental andmodelling results were relatively small.

The experiment was conducted with three subjectswith a normal body mass index, if for example one ofthe subjects had been over or under weight the modelwould have to be scaled in order to fit the test person.However the study was focused on validating the modelas is, not validating the scaling methods, hence thesmall variation between the three subjects.

The results from varying the footrest showed a biastowards the model predicting lower forces than mea-sured in experiments, the difference is not big, howevermost predictions were lower than the matching experi-mental data. The largest error seen on Fig. 5 is when thefootrest is raised, i.e. the distance from the footrest tothe seat is small. This could be explained by the passiveelastic forces there is in the body when the thorax/thighangle becomes smaller, which is not build into the Any-Body model. The subjects lower legs were 48–50 cmtherefore the most important part of the graph is aroundthis height, where the errors were small. The largeerrors seen on the figure comes from somewhat unnat-ural sitting postures.

The model predicted reaction forces between a chairand an able-bodied person, and therefore muscles wereincluded in the model. The muscle activities calculatedby the model were low in all the modelled postures(< 4%). There was, however, a slight increase in theestimated muscle activation when the friction coef-ficient decreases, indicating that in these situationsmuscle activity is necessary to avoid slipping from thechair.

There were a couple of possible error sources in theexperiments, such as marker placement, which couldcause prediction errors. The chosen friction coeffi-cients for the seat and footrest, were investigated in

a sensitivity study. The two coefficients where variedand the shear force at the seat estimated by the modelwas the output. The model turned out to be insensitiveto the friction coefficients if they were above � = 0.15.Thus, if the true friction coefficient if above 0.15, theassumed coefficient of 0.5 does not influence the result.

The study included experimental results from threesubjects, that were chosen to match the standard Any-Body model, this was done to exclude scaling, becausethis study was done in order to verify the model results,not the scaling method, which should be addressed infuture studies. For future patient specific cases, it willbe essential to scale the model to the subject. Espe-cially for disabled subjects, segment masses may haveatypical values and some muscle groups should beremodelled. Three subjects are considered enough forverifying if the interaction between the parameters isthe same in the model compared to reality. Many datapoints have been collected in the experiment for eachsubject. During the experiment the subject had theirarms crossed, which is obviously not a typical seatingposture. The reason was that for normal arm positionsduring sitting, essentials markers on the pelvis werehidden. Pelvis rotation is an important parameter tocapture for estimating the support forces.

The musculo-skeletal model estimates reactionforces between the chair and the human body with thesame trends that can be measured experimentally, andalso the absolute values correspond quite well. Overallthe result is encouraging with respect to the opportuni-ties to use a computational model for seat adjustmentsaimed at controlling the reaction forces in the humanbody.

Acknowledgments

This study is a part of the Seated Human Projectpartially sponsored by RBM A/S. The support is grate-fully acknowledged. The study is also supported by theproject Minimizing the risk of developing a pressureulcer (ERDFN-09-0070), supported by Growth FundNorth Denmark and European Regional DevelopmentFund. The authors also wish to thank Wolturnus A/Sfor their great help with the construction of the custom-built wheelchair for the experiment.

References

[1] L. Agam and A. Gefen, Pressure ulcers and deep tissue injury:A bioengineering perspective, J Wound Care 16(8) (2007),336–342.


[2] AnyScript Community, The AnyScript Community,www.anyscript.org, vol. 2007, no. 01-10-2009.

[3] D. Bader, C. Oomens, C. Bouten, R. Taylor, T. James andJ. Sanders, Pressure Ulcer Research - Current and FuturePerspectives. Springer Berlin Heidelberg, 2005, p. 382.

[4] L. Bennett, D. Kavner, B.Y. Lee, F.S. Trainor and J.M. Lewis,Skin stress and blood flow in sitting paraplegic patients,Archives of Physical Medicine and Rehabilitation 65(4)(1984), 186–190.

[5] T.R. Bush and R.P. Hubbard, Support force measures ofmidsized men in seated positions, Journal of BiomechanicalEngineering 129(1) (2007), 58–65.

[6] T.R. Bush and R.P. Hubbard, Biomechanical evaluation offour office chairs, Human Factors 50(4) (2008), 629–642.

[7] D. Chen, D.F. Apple Jr., L.M. Hudson and R. Bode,Medical complications during acute rehabilitation followingspinal cord injury–current experience of the Model Systems,Archives of Physical Medicine and Rehabilitation 80(11)(1999), 1397–1401.

[8] S.A. Crawford, M.D. Stinson, D.M. Walsh and A.P. Porter-Armstrong, Impact of sitting time on seat-interface pressureand on pressure mapping with multiple sclerosis patients,Archives of Physical Medicine and Rehabilitation 86(6)(2005), 1221–1225.

[9] J. Dahlquist, S.T. Christensen, J. Rasmussen, M.D. Zee andM. Damsgaard, The seated human - biomechanical model-ing/ergonomic design, 2nd Nordic Seating Symposium, 2004.

[10] M. Damsgaard, J. Rasmussen, S.T. Christensen, E. Surmaand M. de Zee, Analysis of musculoskeletal systems in theAnyBody Modeling System, Simulation Modelling Practiceand Theory 14(8) (2006), 1100–1111.

[11] A. Gefen, The biomechanics of sitting-acquired pressureulcers in patients with spinal cord injury or lesions, Inter-national Wound Journal 4(3) (2007), 222–231.

[12] A. Gefen, N. Gefen, E. Linder-Ganz and S.S. Margulies, Invivo muscle stiffening under bone compression promotes deeppressure sores, Journal of Biomechanical Engineering 127(3)(2005), 512–524.

[13] P. Gilsdorf, R. Patterson and S. Fisher, Thirty-minute continu-ous sitting force measurements with different support surfacesin the spinal cord injured and able-bodied, Journal of Reha-bilitation Research and Development 28(4) (1991), 33–38.

[14] P. Gilsdorf, R. Patterson, S. Fisher and N. Appel, Sittingforces and wheelchair mechanics, Journal of RehabilitationResearch and Development 27(3) (1990), 239–246.

[15] R.H. Goossens, C.J. Snijders, T.G. Holscher, W.C. Heerensand A.E. Holman, Shear stress measured on beds

and wheelchairs, Scandinavian Journal of RehabilitationMedicine 29(3) (1997), 131–136.

[16] D.A. Hobson, Comparative effects of posture on pressure andshear at the body-seat interface, Journal of RehabilitationResearch and Development 29(4) (1992), 21–31.

[17] K. Larsen, Mr. Kick, http://www.smi.hst.aau.dk/∼knl/mk/,vol. 2010, no. 31-01-2008.

[18] E. Linder-Ganzand and A. Gefen, Stress analyses coupledwith damage laws to determine biomechanical risk factorsfor deep tissue injury during sitting, Journal of BiomechanicalEngineering 131(1) (2009), pp. 011003.

[19] E. Linder-Ganzand and A. Gefen, Mechanical compression-induced pressure sores in rat hindlimb: Muscle stiffness,histology, and computational models, Journal of AppliedPhysiology (Bethesda, Md.: 1985) 96(6) (2004), 2034–2049.

[20] C.L. Maurerand and S. Sprigle, Effect of seat inclination onseated pressures of individuals with spinal cord injury, Phys-ical Therapy 84(3) (2004), 255–261.

[21] B.M. Niggand and W. Herzog, Biomechanics of the Musculo-Skeletal System. Wiley, Chichester, 1999.

[22] C.G. Olesen, M. de Zee and J. Rasmussen, Missing links inpressure ulcer research - an interdisciplinary overview, Jour-nal of Applied Physiology (2010).

[23] J. Rasmussen and M. de Zee, Design optimization of air-line seats, SAE International Journal of Passenger Cars -Electronic and Electrical Systems (2009), pp. 2009.

[24] J. Rasmussen, S. Tørholm and M. de Zee, Int J Ind Ergon39(1) (2008).

[25] R. Roaf, The causation and prevention of bed sores, Journalof Tissue Viability 16(2) (2006), 6–8.

[26] M. Romanelli, M. Clark, G.W. Cherry, D. Colin and T.Defloor, Science and Practice of Pressure Ulcer Management,Springer, 2005.

[27] C.A. Salzberg, D.W. Byrne, C.G. Cayten, P. van Niewer-burgh, J.G. Murphy and M. Viehbeck, A new pressure ulcerrisk assessment scale for individuals with spinal cord injury,American Journal of Physical Medicine and Rehabilitation /Association of Academic Physiatrists 75(2) (1996), 96–104.

[28] T. Schroeder, Basisbog i Medicin Og Kirurgi. Munksgaard,Kbh., 2005.

[29] A. Stekelenburg, C.W. Oomens, G.J. Strijkers, K. Nicolay andD.L. Bader, Compression-induced deep tissue injury exam-ined with magnetic resonance imaging and histology, Journalof Applied Physiology (Bethesda, Md.: 1985) 100(6) (2006),1946–1954.

www.anyscript.org

http://www.smi.hst.aau.dk/~knl/mk/

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2010

RoboticsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal ofEngineeringVolume 2014

Submit your manuscripts athttp://www.hindawi.com

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks


comparison between a computational seated human model and

Documents