Question about high medio-lateral force gait lower extremity model

Dear all,

Over the last couple of weeks I’ve been working on a lower limb model based on the gait lower extremity example from the AnyBody repository. So far everything works well en the results seem to be feasible, well almost all of them.

I’m looking at the loads and moments on the knee during ADL activities. In the attachment you can find my model and some example data. The c3d file also contains EMG from the following muscles in the following order: Rectus Femoris, Vastus medialis, vastus lateralis, tibialis anterior, gastrocnemius medialis, gastrocnemius lateralis, semimembranosus and biceps femoris. Activity patterns seem to agree with the model.

When I compare different parameters against other published results I see some differences. Most of the them are comparable, but one that is particularly bothering me is the peak medio-lateral force on the knee. Which is 4 times higher that those calculated by Worsley et al. and Taylor (about 0.6 against 0.15*BW) for gait and stair climbing activities. The shape of the results are comparable, but the magnitude is not.

I also ran the model Gait Lower Extremity with the data as supplied by the AnyBody example (GaitNormal0003-processed). In the case of the medio-lateral force the values are almost the same as what i get. (In terms of BW)

I tried different recruitment criteria and muscle types and the most advanced leg model (leg TD) without any remarcable results. I also used the plug-in-gait marker set data for collection.

Can someone explain me where this difference is comming from? is it the markerset or i’m i missing something else? Hopefully you can help me with this problem.

I’m working on the AnyBody version and using the AMMR 1.4

Much appreciated,


  • Predicted knee kinematics and kinetics during functional activities using motion capture and musculoskeletal modelling in healthy older people. Worsley et al.

  • Tibio-femoral loading during human gait and stair climbing. Taylor et al.

Hello Roberto,

the values for M-L forces vary vey much as far as the trials I have analyzed so far. I just run 3 other cases, 2 gait, one stair climbing and I was by 15-40% BW. 60% seems very much, but this might be possible. I have to look at the papers again, I’m not sure if you can compare TKA patients with healthy subjects that easy. I know that my healthy colleage, who was captured for the GaitLowerExtremity model walks with a large rotation.
What is your experience with that?

Hi Amir,

Thank you vey much for your reply. Fisrt of all, I totally agree with you about the fact that you might not completely be able to compare TKA patients with healthy subjects. But the article by Worsley et al. described the kinematics for healthy older subjects (same kind of group i’m working with) using AnyBody. Furthermore Costigan et al. (Knee and hip kinetics during normal stair climbing) also published some results on KCF on healthy young adults with the same order of magnitude.

As explained before, the group I’m working with is composed of healthy older women. As far as i’ve seen, the rotations are not surprisingly high. I was just wondering if I’ve missed something while adapting my model. I changed a few things at the model setup and trial specific files.

When running the standard example form anybody the medio-lateral forces have about the same value as what I get, which is higher than those I’ve found in the literature. I just wanted to know if other people have had the same experiences. I’m still thiking that 60% is kind of high.

Kind regards,

The article by Costigan et al.: Knee and hip kinetics during normal stair climbing. Gait and posture 16(2002) 31-37


I agree also with you, 60% seems also very high for me. I’ll look more into the papers!
Do you have more subjects and are they all the same?

Hi Amir,

I just ran the model for 4 other subjects. Same results, medio-lateral forces around 0.60*BW+/-5.



I analyzed also more Gait Trials. All trials with a “healthy knee” resulted in 50% BW med-lat forces.
However, there are things that can influence the forces. Did you do the Gait measurements yourself? Do you know where the markers have been applied to the subject?
If your marker position was not correct originally in the optimization, you might get wrong loads. I don’t know how much they can vary, but I’ve experienced differences in the range of 100N.
I’ve seen that specially the foot position and of course the knee markers will have an effect.

Hi Amir,

Thanks for your answer. When you say you analysed more gait trials, do you mean with your own data or the one i provided you? 0.5 BW is still much more than the 0.15 described elsewhere.

I didn’t collect the data myself, but I was present during the collection of some of the participants. The biggest differences were at the thigh (RTHI) and shank (RTIB) markers. The offsets I added to the modelsetup file are based on one patient. This doesn’t necessarily mean is the same for all of them. We have different persons in charge of the data collection of the same protocol.

To be completely sure, i would need to collect one data set where i’m definitively sure what the position of the markers is relative to the segments.

Do you think the differences could become as high as 300 N? that’s about the difference needed to match the results elsewhere.

I’ll record a new dataset and come back to you on that one. Any more suggestions so far?


I don’t think it will change the 300N. I would consider contacting Worsley et al. …

otherwise I dont know much more.