Evaluating a Musculoskeletal Model

Getting Started: Defining your Research Goals for the Model

What results do I need to answer my research question?
What elements of a model might these results be sensitive to?
In what ranges will my study be conducted and in what ranges has the model been validated? If a model is not valid in the ranges of your study, how can you extend those ranges and re-validate the model?

What structures or systems are modeled?
Is it a kinematic model or full dynamic model? Is it 3D?
Is it a model of a new structure or structures? Does it build on an existing model? What models can it be compared to?
What experimental data was used to build and test the model? If human or animal subjects were used, how many and what type of subjects (age, size, etc.)?
What platform(s) is the model implemented in?
Can the model be scaled or otherwise customized?
Has the model been published or otherwise peer reviewed?
Do other studies use/cite the model?
Has the author of the model changed or updated the model since the original publication? Do I have and understand the most recent version?

Bone Geometry
- What bony segments are included in the model?
- What experimental or literature data was used to define bone geometry?
Joint Geometry
- How many joints and how many degrees of freedom?
- How are the joints modeled? Types? Joint axis definitions? Coordinates? Neutral positions?
- Is there coupling of joint motions?
- What is the joint range of motion? Are there other limits or other constraints?
- What experimental or literature data was used to define the joint geometry?
Muscle Geometry
- How many muscles are included in the model?
- How are muscle attachment points defined?
- Are broad muscles represented by multiple segments?
- Does the model have wrapping surfaces or via points? What features do they represent?
- What experimental or literature data was used to define the muscle geometry?
Other Structures
- Does the model include ligaments?
- Does the model include prosthetics, orthotics, or other devices?

Does the model include segment inertial parameters?
What experimental or literature data was used to define the intertial parameters?
Are there segments without intertial parameters? (This is a challenge for creating forward simulations with the model.)

Does the model include muscle actuators?
- What type of model is used for muscle-tendon force generation?
- How is muscle architecture modeled? What parameters define muscle architecture?
- How is activation-contraction dynamics of muscles modeled? What parameters define?
- Was there any fine-tuning/adjustment of muscle model parameters?
Does the model have any other actuators (e.g. powered prosthetics, reserve actuators at the joints)?
Does the model have any other passive force-generating elements (e.g. damping at joint extremes)?

Is a model of foot/floor contact included?
Does the model include any other representations of of contact with external objects?

Were model moment arms compared to experimental data?
Were maximum active joint moments compared to experimental data?
Were passive joint moments compared to experimental data?
Was a sensitivity analysis performed for any of the parameters of the model?
Has the model been used to generate simulations?
- How much time is required to generate simulations of standard motions (e.g. walking, reaching)?
- How do the results compare to previously published data?
- For example, if a lower extremity model, does the model accurately predict knee forces from the grand challenge data set (https://simtk.org/home/kneeloads)?

The webinar “Musculoskeletal Models Deconstructed” provides some specific examples of how to evaluate and develop a model (watch | view slides).