Read on Twitter
Field-changing project looking for a PI: accurate data-driven model of muscle mechanics and energetics. Collect force and heat rate for a large grid of position and activation sequences for a grid of muscle properties. Biomechanists use the resulting black-box model. Why?
With much respect, models of muscle mechanics and energetics used by most biomechanists are inaccurate. Hill's 1938 model is most widely used for mechanics, and we know it misses qualitative features of real muscle. Models of metabolic energy use are more recent, but inaccurate.
Physical chemists have made progress at understanding the underlying phenomena, but we do not yet have straightforward ways of mapping between behavior at the molecular and organismal levels.
Researchers using (neuro-) musculoskeletal simulations to predict or analyze human gait need a simple tool to map stimulation and muscle state, and their recent histories, to muscle force and energy uptake. Otherwise model predictions will be wrong (and we do see this problem).
Enter ML: With enough data from isolated muscle preparations, it should be possible to establish a training set that spans all useful stimulation and length trajectories while measuring force and heat production. Train a network (using states and histories) and we have a plugin.
Of course, it would be laborious and costly. Lots of student time spent on animal surgeries, muscle instrumentation, etc. But it can be parallelized, so calendar time could be brief. And once completed, no isolated muscle experiments would ever be needed again ;-).
But would it be worth it? Judging from the 7K citations to Hill's article, researchers find these models useful. In my conversations with musculoskeletal simulation experts, lack of a sufficiently accurate muscle model is a common complaint. It would be hugely impactful.
So if you or someone you know has a wet lab, a love of muscle and an interest in machine learning, please become the creator of the [your name here] model of muscle mechanics and energetics, revolutionizing biomechanical simulation and analysis.
