prediction of applied forces in handrim wheelchair propulsion
TRANSCRIPT
Journal of Biomechanics 44 (2011) 455–460
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/jbiomech
Journal of Biomechanics
0021-92
doi:10.1
n Corr
E-m
www.JBiomech.com
Prediction of applied forces in handrim wheelchair propulsion
Chien-Ju Lin a, Po-Chou Lin a, Lan-Yuen Guo b, Fong-Chin Su a,n
a Institute of Biomedical Engineering, National Cheng Kung University, Tainan, 1 University Road, Tainan City 701, Taiwanb Department of Sports Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan
a r t i c l e i n f o
Article history:
Accepted 27 September 2010Researchers of wheelchair propulsion have usually suggested that a wheelchair can be properly designed
using anthropometrics to reduce high mechanical load and thus reduce pain and damage to joints. A
Keywords:
Wheelchair
Handrim force
Model
90/$ - see front matter & 2010 Elsevier Ltd. A
016/j.jbiomech.2010.09.029
esponding author. Tel.: +886 6 2760665; fax:
ail address: [email protected] (F.-C. Su).
a b s t r a c t
model based on physiological features and biomechanical principles can be used to determine
anthropometric relationships for wheelchair fitting. To improve the understanding of man–machine
interaction and the mechanism through which propulsion performance been enhanced, this study
develops and validates an energy model for wheelchair propulsion. Kinematic data obtained from ten
able-bodied and ten wheelchair-dependent users during level propulsion at an average velocity of 1 m/s
were used as the input of a planar model with the criteria of increasing efficiency and reducing joint load.
Results demonstrate that for both experienced and inexperienced users, predicted handrim contact forces
agree with experimental data through an extensive range of the push. Significant deviations that were
mostly observed in the early stage of the push phase might result from the lack of consideration of muscle
dynamics and wrist joint biomechanics. The proposed model effectively verified the handrim contact
force patterns during dynamic propulsion. Users do not aim to generate mechanically most effective
forces to avoid high loadings on the joints.
& 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Although manual wheelchairs are essential for individuals withmobility impairments, propelling a wheelchair has been shown tobe mechanically inefficient (van der Woude et al., 1986; Veegeret al., 1991) and improper wheelchair design could be a risk ofinjuries after long-term propulsion (Boninger et al., 2000; Mulroyet al., 2005; Wei et al., 2003). The kinematics and kinetics ofpropulsion are highly influenced by the user–wheelchair interface.Thus, a prevailing wheelchair propulsion research issue is howwheelchair fitting affects the propulsion mechanism. An analyticalmodel allows subtle alterations of propulsion technique to bedistinguished from changes of physiological and environmentalfactors (Vanlandewijck et al., 2001). Analytical models have beenwidely utilized in the analyses of human movements such aswalking, jumping, and wheelchair propulsion to improve theunderstanding of how the movements are produced (Pandy, 2001).
Van der Helm and Veeger (1996) first adopted quasi-staticanalysis for predicting muscle forces during wheelchair propulsion.They found that, whereas, the stabilization of the shoulder jointduring wheelchair propulsion required large muscle contributions,the input force of the model deviated from actual forces measuredduring dynamic propulsion. Moreover, they pointed out that the
ll rights reserved.
+886 6 2343270.
impact of the angular velocity of a joint should be considered inpredicting muscle forces. Rozendaal et al. (2003) proposed acriterion to quantify the balance of propulsion forces betweenthe mechanical effect and the biomechanical effort. Their findingssuggested that a deviation of the propulsion force direction fromthe tangential direction might be the best solution that themusculoskeletal system can achieve during the guidedmovement. A planar four-bar linkage model for quasi-staticanalysis of wheelchair propulsion was developed (Guo et al.,2002, 2003; Morrow et al., 2003). Guo et al. examined thepotential propulsion moments and predicted propulsion forcesat five hand positions with simplified static conditions. Propulsionmoments were reported to be greater at the initial and end of thepush phase, which differed from actual moment measured duringdynamic propulsion. Morrow et al. (2003) selected the criterion ofmaximizing the moment at the shoulder joint to assess propulsionforces and joint kinematics under various axle positions andhandrim sizes. Although shoulder and elbow joint angles and thefraction of effective forces (FEF) were determined, the assumptionof a fixed shoulder position in the model is inconsistent withdynamic propulsion.
Validated models make it possible to analytically evaluatewheelchair fit and the associated mechanisms of musculoskeletalinjuries resulting from a poor fit. Evaluation of wheelchair fit beforeprescribing a wheelchair may prevent potential injuries. However,above-mentioned studies indicated that predicted forces from astatic condition deviated from actual forces. Besides, effects of
C.-J. Lin et al. / Journal of Biomechanics 44 (2011) 455–460456
velocity were suggested to be taken into consideration to gain moreaccurate predictions. The present study attempts to develop andvalidate a two-dimensional energy model to verify applied han-drim forces during dynamic wheelchair propulsion for novice andexperienced wheelchair users.
2. Materials and methods
This study employed a planar four-bar linkage mechanism, schematized in
Fig. 1, to simulate the movements of the upper limb and the handrim in the sagittal
plane during wheelchair propulsion. This four-bar linkage consists of the vector
from the wheel axle to the shoulder joint, the vector from the wheel axle to the hand
contact point, the forearm, and the upper arm. The wrist joint was assumed to be
fixed in the current model. Given positions of the shoulder joint, wheel axle, and the
wheel angle, kinematic data of the upper extremity joints were obtained using
planar four-bar linkage analysis.
The concept of the model is that users are expected to apply propulsion forces
that maximize efficiency and reduce the required upper extremity joint moments.
Maximizing efficiency is expressed as minimizing the discrepancy between the
work done by the force and the energy change of the whole system. Work W is
defined as the work done on the wheelchair–user system during a time interval:
W¼Moodt, where Mo is the propulsion moment and o is the angular velocity of the
wheel. The energy change includes the kinetic energy of the user–wheelchair
system, the rotational kinetic energy of the two rear wheels, and the energy
consumed by the rolling resistance (FR) between the tires and the floor. Thus, the
formulation of the optimization problem is as follows
Fig. 1. Diagram of the planar four-bar model in the plane of the wheel frame
(S: shoulder joint; E: elbow joint; H: hand position; O: wheel axle; y: wheel angle;
OH,
, SH,
, and EH,
: vectors from the wheel axle, shoulder, and elbow to the hand,
respectively; F: unknown force).
Minimize
w1 � Moodt�½1=2mðv2tþdt�v2
t Þþ2� IDðo2tþdt�o
2t ÞþFRDs�
��
��þw2 � ½ðMsÞ
2þðMeÞ
2�
subject to
Msext oMs oMsflex
Meext oMe oMeflex
where
Mo ¼ 2� OH,� F
FR ¼mg � Crr
Ms ¼ SH,� F
Me ¼ EH,� F
w1 ¼ 1000; w2 ¼ 1
The unknown propulsion force (F) at each instant (t) is determined with the
objective function subject to the physiological constraints of the shoulder and elbow
moments. Ms and Me represent the required shoulder and elbow joint moments,
respectively. Msflex and Msext are the maximum shoulder strengths in flexion and
extension, respectively, and Meflex and Meext are the maximum elbow strengths in
flexion and extension, respectively. Values of the maximum joint strengths were
obtained from previous studies (Guo et al., 2003; Sabick et al., 2004). It was
identified that wheelchair propulsion was an asymmetrical act (Hurd et al., 2008),
but to simplify the model and reduce unknown variables, forces applied to the right
and left wheels were assumed to be symmetrical. OH,
, SH,
, and EH,
represent the
vectors from the wheel axle to the hand, the shoulder joint to the hand, and the
elbow joint to the hand, respectively. The kinetic energy change is computed from
experimental data where ID is the moment of inertia of the rear wheel, m is the mass
of the user–wheelchair system, and v is the linear velocity of the system. The term
DS is the distance that wheelchair has traveled during a time interval. The coefficient
of rolling resistance (Crr) was obtained from experimental results. Two weighting
factors, w1 and w2, are added in our model for solving the multi-objective problem.
The optimization routines were developed in MATLAB.
The effect of Crr on the prediction outcomes was evaluated by testing Crr, Crr+5%,
Crr+10%, Crr�5%, and Crr�10%. In addition, in order to select weighting factors, w1
and w2, several sets of weights, from 1 to 106 in multiples of 10, were evaluated. It
was assumed that conserving energy and reducing the required joint moments had
equal contribution to the applied handrim forces. Due to different units of these two
terms, it was found that when w2 equaled 1 and w1 equaled 1000, minimizing
shoulder and elbow joint loads and minimizing energy consumption had equal
influence on the prediction outcomes. Ratios of w1 to w2 smaller than 1000 increased
the effect of minimizing joint load. Ratios larger than 1000 did not change the
prediction outcomes. Hence, the weighting factors w1¼1000 and w2¼1 were
selected for the multi-objective problems.
Experimental data came from 10 able-bodied individuals and 10 experienced
wheelchair users with spinal cord injuries (Table 1). Subjects with orthopaedic
problems of their upper extremities were excluded. This study was approved by the
National Cheng Kung University Hospital Institutional Review Board. All
participants were informed and gave their consent prior to testing. Distance
from the shoulder to the wheel axle and anthropometric measurements
including body height, weight, and arm length were obtained for all participants.
All users used the same Quickie GP ultralight-weight manual wheelchair with one
instrumented wheel on the right side on level ground paved with a granular canvas
at an average velocity of 1 m/s. Subjects had to complete five successful trials,
defined as stably propelling the wheelchair through a 4-meters pathway in 470.2 s.
The distance between the rear wheels was adjusted according to user’s arm span. An
eight-camera Eagle Motion Analysis System (Motion Analysis Corp., USA)
synchronized with a load cell (Model UFS-4515A, JR3, Inc., Woodland, CA) on the
instrumented wheel was used to collect kinematic and kinetic data during
propulsion. Nineteen reflective markers were attached to the user–wheelchair
system. Detailed marker placements were reported previously (Lin et al., 2009). The
upper extremity was assumed to be a four-segment-linkage system consisting of the
trunk, upper arm, forearm, and hand. Joint kinematics of the upper extremity is
computed using Euler angles to illustrate the orientation of a distal segment
coordinate system relative to a proximal segment coordinate system. The positions
of the shoulder joint, wheel axle, and wheel angle were obtained by projecting the
3-dimensional data on the plane of the wheel frame.
Push phases were defined by exerted force and parameters were analyzed at a
discrete value of 51 of the wheel angle (y) (Fig. 1). The captured hand positions
within a certain range of wheel angles were analyzed together. The number of data
sets varied for each interval of wheel angle. Force directions were evaluated in terms
of the angle (ype) (Fig. 2) between the predicted and experimentally obtained forces,
and the error in force magnitude prediction was determined in absolute values and
as a percentage of the discrepancy within experimentally obtained force magnitude.
Fractions of effective forces (FEF) in novice and experienced wheelchair users were
also analyzed. The FEF is defined as: FEF¼Ft/F; where Ft is the force applied
tangential to the handrim and F is the resultant force.
Table 1Demographic data of subjects (Mean (SD)).
Age (years) Height (cm) Weight (kg) Arm span (cm) Experience of
wheelchair use (years)
Injury level
Inexperienced users (n¼10) 22.1 (1.1) 174.1 (3.8) 70.7 (8.9) 175.3 (6.7) – –
Experienced users (n¼10) 50.1 (14.4) 166.7 (5.8) 65.2 (12.2) 169.6 (6.0) 14.7 (9.4) T3 (n¼1); T5 (n¼1); T10 (N¼1);
T12 (n¼1); L1 (n¼1); L3 (n¼1); L5 (n¼4)
C.-J. Lin et al. / Journal of Biomechanics 44 (2011) 455–460 457
Independent t-test was used to examine the difference in variables between
inexperienced and experienced users and paired t-test was adopted to examine the
difference in FEF between predicted and experimentally obtained values. The
criterion level was set as po0.05.
Fig. 2. Stick diagram illustrating positions of the upper extremities, experimentally
obtained forces, and predicted forces from a representative trial from an experi-
enced wheelchair user (ype: the angle between the predicted force and experimen-
tally obtained force).
3. Results
With the simplified 2-D model, applied handrim forces wereverified at each hand position during propulsion. A stick diagram ofexperimentally obtained forces and predicted forces from a repre-sentative trial is illustrated in Fig. 2. The experimental data showedthat the propulsion angle varied with subject due to differentpropulsion techniques used. Data are shown in 51 increments from1201 to 301 of the wheel angle.
Predicted forces generally agreed with experimentally obtainedforces in direction and magnitude for both inexperienced andexperienced wheelchair users from the middle to the end of thepush phase. Fig. 3 shows that ype ranged from 0.01 to 7.61 and from12.41 to 35.41 for wheel angles in the range 301–1051 forinexperienced and experienced users, respectively. A largedeviation of predicted force directions was noted when the handwas located at a wheel angle larger than 1051. When the hand waspositioned at a wheel angle larger than 1101, ype increased to avalue larger than 1001. Furthermore, during a propulsion cycle, thepeak FEFs for predicted forces were significantly higher than thosefor actual forces in both inexperienced and experienced users. Fromthe experimental results, peak FEFs were 80% and 74% forinexperienced and experienced users, respectively. However,peak FEFs of predicted forces were 96% and 97%, respectively, forexperienced and inexperienced users.
The discrepancies in force magnitude in absolute values andas a percentage of prediction error are shown in Fig. 4. Throughthe entire push phase, the ensemble averages of magnitudediscrepancy were less than 29.7 N, which accounted for 16.9–88.2%and 11.0–91.5% of actual forces for inexperienced and experiencedusers, respectively. Predicted errors varied with hand positions.Large errors were observed at the very end of a propulsive cycle andin the range 75–1101 of the wheel angle.
An evaluation of the effect of the estimated coefficient of rollingresistance (Crr¼0.034) on prediction outcomes (Table 2)demonstrates that when Crr increased or decreased by 10%, thepredicted force magnitude varied by less than 6.13%. When Crr wasincreased or decreased by 5%, the average change in predicteddirection ranged from 0.57% to 2.61%. Crr had large effect on theprediction outcomes for inexperienced users than it did forexperienced users in both force direction and force magnitude.
4. Discussion
The proposed energy model predicted forces which appeared tobe generally compatible with actual forces in the plane of the wheelframe during dynamic propulsion for both novice and experiencedwheelchair users. In terms of force direction, the angles betweenpredicted and experimentally obtained forces were small exceptfor when the hand held the handrim at a wheel angle larger than1051. Predicted forces were tangential to the handrim whereas
forces measured during actual wheelchair propulsion were in thedownward direction. Similar results were obtained in previousstudies (Guo et al., 2003; Rozendaal and Veeger, 2000; Rozendaalet al., 2003; van der Helm and Veeger, 1996). Upward-pointingforce vectors were obtained during static propulsion for handpositions posterior to the top dead center (Guo et al., 2003; van derHelm and Veeger, 1996) because under static propulsion, it isefficient for elbow flexors to generate propulsive force at thesepositions. Additionally, the upwards propulsion force was the mostefficient in the early push phase based on the effect-cost criterion(Rozendaal and Veeger, 2000; Rozendaal et al., 2003). At the start ofthe push phase, the guided movement, limited degrees of freedomof the upper extremity and muscle contributions determinedhandrim force generation. During the early push phase, theelbow joint went from flexion to extension, resulting in quickswitches of muscles activity. However, force generation capacityand dynamics of muscles were not incorporated in the currentmodel, which might partially explain the initial divergence ofpredicted force directions. Except for the initial stage, predictedforce directions were consistent with experimentally obtained
Fig. 3. Results of predicted force direction, represented as mean angle between the predicted force and experimentally obtained force (ype) in each region of wheel angles.
Fig. 4. Results of predicted force magnitude, represented as (a) a percentage and (b) as absolute values of the discrepancy within experimentally obtained force magnitude in
each region of wheel angles.
C.-J. Lin et al. / Journal of Biomechanics 44 (2011) 455–460458
forces through the rest of the push phase especially forinexperienced users. The model predicted forces showed a moreeffective propulsion method than did actual conditions. Peak FEFsobserved in the experiments were below 80% for both
inexperienced and experienced users, which were in agreementwith previous studies (Kotajarvi et al., 2006; van der Woude et al.,2001). As described in previous studies, FEFs were about 73% andincreased with velocity and intensity in wheelchair-dependent
Table 2Variations in prediction outcomes with coefficient of rolling resistance (Mean (SD)).
Force direction Force magnitude
Inexperienced users Experienced users Inexperienced users Experienced users
Change of coefficient of rolling resistance (%) �10 �5 +5 +10 �10 �5 +5 +10 �10 �5 +5 +10 �10 �5 +5 +10
Effect on prediction results (%) 2.61 1.67 1.34 2.40 0.67 0.66 0.57 0.70 6.13 3.03 2.92 5.84 5.35 2.72 2.70 5.30
(2.67) (1.63) (1.30) (2.11) (1.35) (0.89) (0.92) (0.98) (2.02) (1.06) (1.06) (1.99) (2.50) (1.27) (1.30) (2.58)
C.-J. Lin et al. / Journal of Biomechanics 44 (2011) 455–460 459
individuals (Boninger et al., 1997b; Kotajarvi et al., 2006). In thecurrent model, better predictions were observed in inexperiencedusers, which might be explained by the findings that whencompared to inexperienced users, experienced users propelledthe wheel with forces pointed more downward and the shoulderjoints were subject to lower joint moments. It is supposed that forexperienced wheelchair users, reducing joint demands had a higherpriority than generating effective forces while propelling at aleisure velocity. As previously reported, physiologicallywheelchair users do not aim to produce the most effective forceduring propulsion to avoid high joint contact force (Bregman et al.,2009; Desroches et al., 2008).
From the viewpoint of prediction error, great discrepancies inmagnitude of total forces were noted at the early and end stages ofthe push phase. A partial explanation for this may be that predictedforces deviate more at these phases, consequently creating greaterinconsistencies in force magnitude. Another possible explanationfor the large errors is that during the initial contact and releasingphases, the actual applied forces are relatively small compared tothose during the middle of the push phase. A force discrepancy ofaround 28 N could thus result in a huge error. Very little was foundin the literature regarding the accuracy of predicted force magni-tude. A difference in magnitude in the range 46.9–277 N betweenthe predicted and the experimentally obtained forces was pre-viously reported for static propulsion (Guo et al., 2003). In addition,the wrist joint moment toward radial deviation was reported to bearound 3 Nm during initial propulsion (Boninger et al., 1997a). Thelack of consideration in physiological constraints in wrist jointmoment might also result in the errors. Prediction under dynamicpropulsion is more cumbersome than that under static propulsion.Thus, the decreased discrepancy in the present study suggests thatthe proposed model could better verify applied handrim forces.Various coefficient of rolling resistance (Crr) values have beenreported depending on the material and pressure of the tires andthe surface (Kim and Savkoor, 1997; Martin et al., 1998; Mooreet al., 2000). The effect of Crr on prediction outcomes was tested byevaluating Crr values of 75% and 710%. The prediction outcomesfor force direction and magnitude changed by less than 2.61% and6.13%, respectively, with 710% varies in Crr. Therefore, theestimation of the coefficient of rolling resistance in the currentmodel is believed to be accurate.
The proposed simplified 2-dimensional energy model seems toreflect the major movements of the upper extremities that occur inthe plane of the wheel frame during propulsion. Various limitationsshould be noted in the current model. The essential out-of-planearm orientation and the wrist joint in manual wheelchair propul-sion contribute considerably to upper extremity biomechanics.During wheelchair propulsion, the shoulder joint demonstratesgreat range of motion in the frontal and transverse planes (Koontzet al., 2002; Mercer et al., 2006). Movements in the frontal andtransverse planes during propulsion had impacts on forceapplication. Likewise, the muscles that control more thanone degree of freedom might be of importance in predictingmuscle activity during wheelchair propulsion (van der Helm andVeeger, 1996). Wrist injuries are the second most common type of
injury seen in manual wheelchair. Extreme wrist joint movementsusually lead to a heavy burden on the joints (Shimada et al., 2001;Veeger et al., 1998). Nonetheless, the wrist joint was modeled asfixed in the current study and the physiological constraints of thewrist joint were not taken into consideration. Accordingly, a modelof the wrist joint and the consideration of muscle dynamics isrequired to gain better simulations of applied forces.
Wheelchair propulsion is a complex movement involving aman–machine interface. In order to make the proposed modelapplicable under clinical conditions in the future, kinematicsparameters and anthropometric measurements of users are pro-posed as the inputs to the model. Inputting kinematic data, appliedhandrim forces were obtained. Joint kinetics was also acquired.Handrim force prediction in both direction and magnitude duringdynamic propulsion was achieved in this study. Prior studieshighlighted the differences in propulsion techniques, upper extre-mity kinematics, trunk movement, and applied handrim forcesbetween able-bodied and wheelchair-dependent users (Dubowskyet al., 2009; Rodgers et al., 2003; Veeger et al., 1992). However, thecriteria used in this study verified propulsive forces for bothinexperienced and experienced users. The model could befurther applied to examine how changes in wheelchair designand velocity affect applied forces and joint kinetics in propulsion.
Conflict of interest statement
The authors declare that there is no conflict of interest inthis study.
Acknowledgements
This work was supported by the National Health ResearchInstitute, Taiwan, under Grant NHRI-EX98-9617EI. The authorswould like to thank Hsiao-Feng Chieh for her technical support andassistance with data collection. Foreign Language Center inNational Cheng Kung University is appreciated for its editorialassistance.
References
Boninger, M.L., Baldwin, M., Cooper, R.A., Koontz, A., Chan, L., 2000. Manualwheelchair pushrim biomechanics and axle position. Archives of PhysicalMedicine and Rehabilitation 81, 608–613.
Boninger, M.L., Cooper, R.A., Robertson, R.N., Rudy, T.E., 1997a. Wrist biomechanicsduring two speeds of wheelchair propulsion: an analysis using a local coordinatesystem. Archives of Physical Medicine and Rehabilitation 78, 364–372.
Boninger, M.L., Cooper, R.A., Robertson, R.N., Shimada, S.D., 1997b. Three-dimen-sional pushrim forces during two speeds of wheelchair propulsion. AmericanJournal of Physical Medicine and Rehabilitation 76, 420–426.
Bregman, D.J., van Drongelen, S., Veeger, H.E., 2009. Is effective force application inhandrim wheelchair propulsion also efficient? Clinical Biomechanics 24 13–19.
Desroches, G., Aissaoui, R., Bourbonnais, D., 2008. Relationship between resultantforce at the pushrim and the net shoulder joint moments during manualwheelchair propulsion in elderly persons. Archives of Physical Medicine andRehabilitation 89, 1155–1161.
Dubowsky, S.R., Sisto, S.A., Langrana, N.A., 2009. Comparison of kinematics, kinetics,and EMG throughout wheelchair propulsion in able-bodied and persons with
C.-J. Lin et al. / Journal of Biomechanics 44 (2011) 455–460460
paraplegia: an integrative approach. Journal of Biomechanical Engineering 131,021015.
Guo, L.Y., Su, F.C., An, K.N., 2002. Optimum propulsion technique in different wheelchairhandrim diameter. Journal of Medical and Biological Engineering 22, 1–10.
Guo, L.Y., Zhao, K.D., Su, F.C., An, K.N., 2003. Moment generation in wheelchairpropulsion. Proceedings of the institution of mechanical engineers. Part H.Journal of Engineering in Medicine 217, 405–413.
Hurd, W.J., Morrow, M.M., Kaufman, K.R., An, K.N., 2008. Biomechanic evaluation ofupper-extremity symmetry during manual wheelchair propulsion over variedterrain. Archives of Physical Medicine and Rehabilitation 89, 1996–2002.
Kim, S.J., Savkoor, A.R., 1997. The contact problem of in-plane rolling of tires on a flatroad. Vehicle System Dynamics 27, 189–206.
Koontz, A.M., Cooper, R.A., Boninger, M.L., Souza, A.L., Fay, B.T., 2002. Shoulderkinematics and kinetics during two speeds of wheelchair propulsion. Journal ofRehabilitation Research and Development 39, 635–649.
Kotajarvi, B.R., Basford, J.R., An, K.N., Morrow, D.A., Kaufman, K.R., 2006. The effect ofvisual biofeedback on the propulsion effectiveness of experienced wheelchairusers. Archives of Physical Medicine and Rehabilitation 87, 510–515.
Lin, C.J., Lin, P.C., Su, F.C., 2009. Preferred elbow position in confined wheelchairconfiguration. Journal of Biomechanics 42, 1005–1009.
Martin, J.C., Milliken, D.L., Cobb, J.E., McFadden, K.L., Coggan, A.R., 1998. Validation ofa mathematical model for road cycling power. Journal of Applied Biomechanics14, 276–291.
Mercer, J.L., Boninger, M., Koontz, A., Ren, D., Dyson-Hudson, T., Cooper, R., 2006.Shoulder joint kinetics and pathology in manual wheelchair users. ClinicalBiomechanics 21, 781–789.
Moore, S.W., Rahman, K.M., Ehsani, M., 2000. Effect on vehicle performance ofextending the constant power region of electric drive motors. SAE Transactions108, 2187–2191.
Morrow, D.A., Guo, L.Y., Zhao, K.D., Su, F.C., An, K.N., 2003. A 2-D model of wheelchairpropulsion. Disability and Rehabilitation 25, 192–196.
Mulroy, S.J., Newsam, C.J., Gutierrez, D.D., Requejo, P., Gronley, J.K., Haubert, L.L.,Perry, J., 2005. Effect of fore-aft seat position on shoulder demands duringwheelchair propulsion: part 1. A kinetic analysis. Journal of Spinal CordMedicine 28, 214–221.
Pandy, M.G., 2001. Computer modeling and simulation of human movement.Annual Review of Biomedical Engineering 3, 245–273.
Rodgers, M.M., McQuade, K.J., Rasch, E.K., Keyser, R.E., Finley, M.A., 2003. Upper-limbfatigue-related joint power shifts in experienced wheelchair users and
nonwheelchair users. Journal of Rehabilitation Research and Development 40,27–37.
Rozendaal, L.A., Veeger, D.E., 2000. Force direction in manual wheel chair propul-sion: balance between effect and cost. Clinical Biomechanics 15 (Suppl 1),S39–41.
Rozendaal, L.A., Veeger, H.E., van der Woude, L.H., 2003. The push force pattern inmanual wheelchair propulsion as a balance between cost and effect. Journal ofBiomechanics 36, 239–247.
Sabick, M.B., Kotajarvi, B.R., An, K.N., 2004. A new method to quantify demand on theupper extremity during manual wheelchair propulsion. Archives of PhysicalMedicine and Rehabilitation 85, 1151–1159.
Shimada, S.D., Cooper, R.A., Boninger, M.L., Koontz, A.M., Corfman, T.A., 2001.Comparison of three different models to represent the wrist during wheelchairpropulsion. IEEE Transactions on Neural Systems and Rehabilitation Engineer-ing 9, 274–282.
van der Helm, F.C., Veeger, H.E., 1996. Quasi-static analysis of muscle forces in theshoulder mechanism during wheelchair propulsion. Journal of Biomechanics 29,39–52.
van der Woude, L.H., de Groot, G., Hollander, A.P., van Ingen Schenau, G.J., Rozendal,R.H., 1986. Wheelchair ergonomics and physiological testing of prototypes.Ergonomics 29, 1561–1573.
van der Woude, L.H., Veeger, H.E., Dallmeijer, A.J., Janssen, T.W., Rozendaal, L.A.,2001. Biomechanics and physiology in active manual wheelchair propulsion.Medical Engineering and Physics 23, 713–733.
Vanlandewijck, Y., Theisen, D., Daly, D., 2001. Wheelchair propulsion biomechanics:implications for wheelchair sports. Sports Medicine 31, 339–367.
Veeger, H.E., Meershoek, L.S., van der Woude, L.H., Langenhoff, J.M., 1998. Wristmotion in handrim wheelchair propulsion. Journal of Rehabilitation Researchand Development 35, 305–313.
Veeger, H.E., van der Woude, L.H., Rozendal, R.H., 1991. Within-cycle characteristicsof the wheelchair push in sprinting on a wheelchair ergometer. Medicine andScience in Sports and Exercise 23, 264–271.
Veeger, H.E.J., Lute, E.M.C., Roeleveld, K., van der Woude, L.H.V., 1992. Differences inperformance between trained and untrained subjects during a 30-s sprint test ina wheelchair ergometer. European Journal of Applied Physiology and Occupa-tional Physiology 64, 158–164.
Wei, S.H., Huang, S., Jiang, C.J., Chiu, J.C., 2003. Wrist kinematic characterization ofwheelchair propulsion in various seating positions: implication to wrist pain.Clinical Biomechanics 18, S46–52.