Dear all, I want to know what is the relationship between muscle activity (output of EMG) and force production (output of force platform or analysis by inverse dynamic), and any reference indicated to this relationship.
The relationship between EMG amplitude (e.g., root-mean-square) and muscle force production (or torque about a joint) is often linear but not always. There is often a curvilinear component to it. Most EMG and muscle physiology textbooks discuss this relationship. I am traveling now and don't have access to such textbooks, otherwise I would quote you a specific reference or two.
Gabriel DA, Kamen G, Frost G. Neural adaptations to resistive exercise: mechanisms and recommendations for training practices. Sports Med. 2006;36(2):133-49.
I've actually always had the view that the relationship between EMG and force is highly non-linear. The amount of force a muscle produces when electrically activated by EMG depends on the current (and previous) state of that muscle.
The force a muscle produces depends on its length, velocity of movement, and frequency. All three of those relationships are highly non-linear (just google scholar 'force-length relationship muscle' for instance for references, or you can even just search it on google images which will show you straight away what I'm talking about).
Additionally, muscles are thixotropic. This means that the amount of force they produce depends on the current, and recent, history of movement. When a muscle is moving or has just moved, each unit of input (EMG) will generate a very much larger force compared to when the muscle is in a postural/isometric contraction. Kenneth Campbell has written a very nice review on this topic of short-range mechanical properties of muscles (http://www.ncbi.nlm.nih.gov/pubmed/?term=Short-Range+Mechanical+Properties+of+Skeletal+and+Cardiac+Muscles)
A good starting point is that it is linear. However, this is only when the muscle is activated isometrically at the same length across different intensities. In reality, a true linear relationship is not really recorded due to, but not limited to:
It is actually VERY difficult to achieve a true isometric contraction of the same muscle at the same length, but under different activation intensities. Most would undertake a slow-ramped contraction, but due to the elastic properties of the muscle and tendon, the muscle will change length, and so you will be collecting data at different lengths (which brings in the force-length relationship) and at velocity due to the muscle shortening and elastic structures stretching (which brings in the force-velocity relationship).
in vivo, we normally measure joint torque (or moment; probably another debate) and try and denote muscle force from this (i.e. through division of moment arm). However, this gives the resultant joint torque of all muscles crossing that joint, and not just the torque due to the muscular force of interest. As such, any estimated force we use to relate to the muscle activation is not solely sourced from that muscle.
Muscles are controlled in synergistic groupings. As such, it's very difficult to isolate a single muscle (probably impossible in vivo). This brings into play the requirement to measure all muscles, and to weight the activation accordingly; but even this is not without problems.
How you measure joint torque and moment arm to allow estimation of force will have their own methodological challenges that will affect the outcome of the overall relationship.
Muscle fibre composition will affect the relationship i.e. when the faster twitch muscle fibres (for want of a general term) start activating later in the contraction, there may be an exponential rise in force for activation.
Processing of the EMG signal will massively affect this relationship as well.
These are only a select few reasons why on the global side (i.e. in vivo) of the muscular force-activation relationship, we report a curvilinear relationship between activation (surface EMG) and muscle force. Further, if you are measuring force on a force platform, then you have many muscles crossing many joints that you are then trying to sum (and/or weight accordingly) to develop a relationship.
My thesis can be found here (https://www.researchgate.net/publication/273759936_Mechanical_factors_affecting_the_estimation_of_tibialis_anterior_force_using_an_EMG-driven_modelling_approach), which focuses on this question in detail, so instead of putting a lot of references here for your further research, there are a lot at the back of the thesis.
Thesis Mechanical factors affecting the estimation of tibialis ante...