Gait stabilty and modeling

[Saikat Chakraborty]

Thanks for giving information.

I have an another query; as I am working with gait stability, I observed that many research papers are collecting data using some wearable sensors or camera (Vicon etc.) and through analyzing obtained data by some statistical approach gait stability is concluded. Some papers also used gait modeling approach to analyze stability.

But if we are getting necessary data from sensors (e.g. wearable and vision based) for stability analysis then what is the need gait modeling which is itself a computationally intensive procedure?

[Manvendra Singh]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Modeling? Do you mean simulation?

[Ton van den Bogert]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Saikat,

This is a very interesting discussion topic, and I am sure that some forum members can give useful information.

First, please write another post to summarize a few of those research papers that you mentioned, and provide the references.

That way, we can be sure that we understand the question, and this is also useful as an introduction to the topic. Not everyone will be familiar with it.

--

Ton van den Bogert (moderator)

[Ross Miller]

Re: Best web-site for interaction with doctors engaged with human gait disorder

I really like these studies from Sjoerd Bruijn on stability metrics and gait:

https://www.ncbi.nlm.nih.gov/pubmed/21762919

https://www.ncbi.nlm.nih.gov/pubmed/21856204

They examined correlations between typical metrics for stability in biomechanics and the probability of falls by a passive dynamic walking model. "Falling" was defined as the inability to complete 30 steps after an unexpected change in floor height.

Ross

[Allan Carman]

Re: Best web-site for interaction with doctors engaged with human gait disorder

This is a bit of a general response to the question ‘what is the need for gait modelling <simulation>” and hopefully those working in the area will be able to contribute more detail.

A basic premise when measuring stability is that the greater the variability of movement the greater the instability. In standing balance variability is usually measured as the ant-post and medio-lateral variance of the center of pressure measured from a force platform. In walking the measures may be gait parameters (such as step length, step width, toe clearance …) or gait kinematics/kinetics (joint angles, moments …) or accelerometer/inertia sensor based measurements. I think the choice of approach comes down to what the question is and factors like whether it is lab based or in the field or what equipment is available; either way it is usually the within participant variance of the measure across repeated gait cycles pooled across subjects. With the same assumption that the greater the variance the greater the instability. There are a few other measures used to measure randomness of a large data set such as Approximate Entropy or Lyapunov Exponents. Search Nickolas Stergiou, University of Omaha, his group has done a lot in this area.

An aspect of stability not considered in the above is recovery from a small perturbation. Cerebral Palsy gait can be surprisingly repeatable across repeated gait cycles despite the abnormal joint angles. However, introduce a small perturbation and recovery of gait between an able bodied and CP, I would suspect, will be quite different. Similarly between younger and elderly. I have heard of labs using a support harness and introducing random unexpected movements of a raised floor to perturb participants? But unexpected perturbations is now in the area of simulation and trying to predict neuro-muscular and movement responses following simulated perturbations to predict gait stability or recovery of gait. See Ross’s response for examples of gait simulation.

So to answer the question ‘is simulation necessary’ with a vague answer: it depends on the question you are trying to answer.
Cheers
Allan

[Manvendra Singh]

Re: Best web-site for interaction with doctors engaged with human gait disorder

General answer would be that you can do much more in simulation then in actual conditions, that might not always be easy, or may just be impossible, to recreate. Question would be how closely real conditions can be simulated with reference to a particular problem. In the context of Neuro-Muscular systems affecting Stability/Gait, based on available knowledge base (laws of physics and Research studies), simulation can help test new Hypothesis to get you started. But eventually, it will need to be validated through some sort of field studies.

[Saikat Chakraborty]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Ok,that's mean to analyze the response of neuro-muscular system caused by unexpected perturbation, simulation (i.e. gait modelling) is more beneficial than analyzing gait variable data separately which has been collected through some wearable/non-wearable devices,right?

And you are saying that it is due to "participant variance of the measure across repeated gait cycles pooled across subjects" i.e. measurement of gait variables may be incorrect by observers,right? But observers i.e. researchers are extracting just sensor output data to analyze stability.Can sensor output (i.e. IMU or Kinect) data be incorrect?

Have I rightly clarified your answer? Or are you saying that advantage of gait modelling is mainly in its ability to analyze "unexpected" perturbation i.e. stability can more accurately be measured and justified through gait modelling?

[Manvendra Singh]

Re: Best web-site for interaction with doctors engaged with human gait disorder

I think there is some confusion between what we are referring to by "Modelling" and "Simulation" and analysis of Gait/Balance through sensors. Here is an example. Suppose, your goal is to design a trans-femoral prosthesis that is lower in cost, and also improves Gait and Balance of amputees. If you think through the whole process, you would get clear idea.

[Saikat Chakraborty]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Do you mean that simulation is necessary to enhance accuracy of a model?

[Ross Miller]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Saikat,

In the studies I posted above, the model was used by the researchers because they wanted to record many instances of actual falls without putting human subjects at risk.

Whether results like that are useful for drawing conclusions on fall risk in real human walking will depend on how much confidence you have in the model. I think the important question there is to consider whether the model was missing something present in a real human system that could produce fundamentally different results. The study authors made the argument that if stability metrics (e.g. Lyapunov exponents) cannot predict fall risk even in a simple computer model, this would "render them less useful in assessing human gait stability."

Hope this helps,
Ross

[Saikat Chakraborty]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Thanks, Ross.

So, from the above discussions can I conclude that gait modeling is necessary for clarifying the accuracy of stability than using just some stability metrics upon gait data which we can collect from wearable/non-wearable devices?

For reference, I am providing some articles where gait stability was analyzed without using any specific gait model, just depending on spato-temporal and kinematic data collected from wearable/non-wearable sensors;

https://www.ncbi.nlm.nih.gov/pubmed/22743027

https://www.ncbi.nlm.nih.gov/pubmed/19340662

[Ross Miller]

Re: Best web-site for interaction with doctors engaged with human gait disorder

If the goal is to predict fall risk, modeling and simulation can inform the relationship between particular stability metrics and the fall risk of a model. How or if that relationship can be generalized to predicting fall risk of a human will again depend on how much confidence you have in the model.

I suggest these two recent review articles in JRSI:

https://www.ncbi.nlm.nih.gov/pubmed/23516062

https://www.ncbi.nlm.nih.gov/pubmed/21880615

Ross

[Manvendra Singh]

Re: Best web-site for interaction with doctors engaged with human gait disorder

Look at these this way. It may well be that findings of these studies could be used to design and validate a Gait Simulation model, that in turn could help future researchers to use simulation, rather than making all those recordings on actual subjects for solving a problem that concerns similar subjects in same domain.