Professional Documents
Culture Documents
Hinrichs1985 - Momento de Inercia
Hinrichs1985 - Momento de Inercia
IW + a,
< 1985 Pcrprmon Prcrs Lrd
TECHNICAL NOTES
RICHARD N. HINRICHS
Btomechanics Laboratory, Division of Physical Education, North Texas State University. Denton.
TX 76203. U.S.A.
Abstract-A set of regression equations was developed to fully utilize the data of Chandler rf 01. (AMRL
Technical Report 74- 137, Wright-Patterson Air Force Base. 1975) to estimate segmental moments of inertia tn
living subjects. Using anthropometric measurements as predictors, moments of inertia can be computed
about both transverse and longitudinal axes passing through each segment’scenter of mass. Symmetry about
segment long axes is assumed. Because of the small sample size upon which these equations are based. it is
suggested that they be used cautiously, especially to avoid extrapolation to subjects having anthropometrlc
measurements outside the range of sample values.
621
622 Technical Notes
Table 1. Regression equations generated from thedata ofchandler er al. (1975) for
predicting segment moments of inertia from anthropometric dimensions*
RL
Head
I,=f,=24,114(HEADC)-1200.4 0.908
= 25102(HEADC)-6.4805(HEADL)- 1122.6 0.964
Torso
I, = 1098.3 (TORSL) - 59448.0 0.874
= 754.38 (TORSL) + 241.94 (CHSTC) - 59445.0 0.984
= 707.62 (TORSL) + 302.71 (WASTC) - 58455.0 0.990
I, = 173.73 (CHSTC) - 12623.0 0.858
= 206.6 I( WASTC) - I 3339.0 0.909
Upper arm
I, = 13.443 (ACRDL) - 310.99 0.882
= 13.264 (ACRDL) + I .82 I7 (AXARC) - 360.79 0.974
= 12.687(ACRDL)+2.I155(BICPC)-349.16 0.988
= 10.268 (ACRDL) + 5.0655 (ELBOC) - 349.71 0.992
I, = 2,1744(BICPC)-42.810 0.924
= 4.3701 (ELBOC) - 102.63 0.957
= 2.6797 (ELBOC) + 0.94597 (BICPC) - 82.643 0.989
= 1.3707 (ACRDL) + 2.0758 (BICPC) - 85.725 0.996
Forearm
I, = 8.2423 (ELBOC) - I7 I .O9 0.648
= 5.5319fFARMCl+ S.OO66IRDSTLl- 298.94 0.865
= 9.5544(WRISC)+ 10.452(~DSTL)L371.1 I 0.904
I, = 2.2490(ELBOC)-55.691 0.855
= 1,7O60(FARMC)-39.014 0.875
= 1.7258 (FARMC) + 0.92028 (RDSTL) - 63.602 0.964
Hand
I, = 2.7501 (STMCL) - 15.968 0.788
= I.2064 IHANDC) - 19.059 0.824
= 2.7443 (HANDBj - 16.882 0.891
I, = 1.3703(STMCL)-9.2416 0.796
= 1.3387 (HANDB) - 9.4514 0.862
= 0,62016(HANDC)- II.185 0.886
Thigh
I, = 326.80 (KNEEB) - 2279.2 0.598
= 90.039 (THIGL) +45.439 (UTHIC) - 5093.3 0.875
= 78.072 (THIGL) + 37.8OO(MTHIC) - 3960.5 0.88 I
= 80.589 (THIGLl+ 381.74 (KNEEB) - 6525.7 0.944
I, = 16.958 (UTHIC) - 598.00 0.855
= 143.26 (KNEEB) - 1301.2 0.906
= 89.242 (KNEEB) + 7.8926 (UTHIC) - I 108.7 0.962
Calf
I, = 66.879 (ANKLC) - 975.51 0.694
= 30.629 (CALFL) + 40.922 (KNEEC) - 2261.3 0.899
= 265.94 (ANKLC) - 61.797 (CALFC) - 3 I IO.0 0.97 I
I, = 3.5868 (CALFC) - 83.550 0.863
= 8.0795 (KNEEC) - 267.68 0.946
= 12.049 (ANKLC) - 2.7994 (CALFL) - 112.30 0.959
Foot
I r = 3.8538 (BOFTC) - 57.257 0.905
= 3.5538 lFOOTLl+ 2.3132 (ARCHC) - 114.76 0.916
= 6.7508 ~F~OTL~ -4.2725 (LMALH) - 105.42 0.942
I ,_ = 1.2988 (ARCHC) - 26.708 0.855
= 1.2663 (BOFTC) - 22.019 0.926
using the anthropometric dimensions as predictors. Because in Table 1 is kg cm’ rather than the SI unit kg m’ in order to
of the sample size, the equations were restricted to only one or have more manageable coefficients.
two predictors per equation. All possible combinations were The abbreviations used in Table I for the anthropometric
tried first using only a single predictor and then with two. dimensions are defined in Table 2. The reader should refer to
The program also determined an F ratio for each regression Chandler et al. (1975) for full descriptions of each dimension.
coefficient, giving the probability of each coefficient being Also included in Table 2 are ranges of values for each
equal to zero. This gave an indication of the stability of the dimension taken from the Chandler et al. cadavers. These
coefficients. The decision was made to keep an equation if all represent the extreme values in the sample data upon which
its coefficient probabilities were less than 0.125. Among these the equations were generated. It is suggested that the
two or three with the highest values of R2 using a single equations not be used for subjects having anthropometric
predictor and with two predictors were retained. dimensions outside of these ranges. The following is a real
The purpose of having more than a single equation to example of what can happen when an extrapolation is
predict each moment of inertia was to give the best possible attempted: the author used the equations to predict I,for the
estimate of moment of inertia from anthropometric dimen- thigh of a short subject with small bone diameters. His thigh
sions. The small sample size afforded a high risk of overfitting. length was 41.6 cm, and his knee breadth was 8.2 cm, both
Thus while two equations having similar values of RZ may smaller than ‘allowable range’ listed in Table 2. These values
predict similar results for the sample. they may produce predicted a negative moment of inertia, which is, of course, an
markedly diITerent results on a living subject. Which one impossibility. Granted, this subject probably had a relatively
should be used? Because the author could not answer that small thigh moment of inertia, but certainly not a negative one
question, it is suggested that the reader use the two or three or even zero. In cases such as this, the equations should not be
best equations for each segment and then average the results used. The anthropometric dimensions of a given subject
together. should fall within the ranges listed in Table 2 in order to
minimize errors.
Given the limitations of the Chandler et al. data, the work
RESULTS AND DlSCUSSlON represents the most comprehensive study to date in which
BSPs were measured on cadavers. (If only the sample size had
The equations are presented in Table 1. In each case at least been larger .) The accuracy of predicting BSPs of living
one equation with only a single predictor has been included. subjects from cadaver data remains uncertain. If the regres-
In three segments (torso I,, hand Irand I,. and foot I,_) no sion equations presented here are used with caution, however,
two-predictor equations were retained because they did not one should be able to better utilize the Chandler et al. data for
demonstrate stable coefficients. The unit for moment of inertia predicting segment moments of inertia.
624 Technical Notes