3D Predictive Simulations of Walking

This repository contains code and data to generate predictive simulations of human walking as described in Falisse A, Afschrift M, De Groote F (2022) Modeling toes contributes to realistic stance knee mechanics in three-dimensional predictive simulations of walking. PLoS ONE 17(1): e0256311.

Here is an example of a predictive simulation of walking, based on a complex musculoskeletal model (31 degrees of freedom, 92 muscles, 6 compliant foot-ground contacts per foot), generated with our framework.

The main script is main.py and the easiest is to start exploring the code from there. The code is for use on Windows, but please post an issue if you want support for other platforms as changes are minimal.

Install requirements

Open Anaconda prompt
Create environment: conda create -n 3dpredsim pip spyder
Activate environment: activate 3dpredsim
Navigate to the folder where you want to download the code: eg. cd Documents
Download code: git clone https://github.com/antoinefalisse/predictsim_mtp.git
Navigate to the folder: cd predictsim_mtp
Install required packages: python -m pip install -r requirements.txt

Overview of main files:

main.py: basically everything from loading data, formulating the problem, solving it, and processing the results. Yes I know, it is not fantastic pratice to have everything in one place. Case 42 (default) converges in 772 iterations on my windows machine.
plotResults.py: plots of simulation results against reference data (eg, joint angles and torques, ground reaction forces, and muscle activations).
Figure<>.py: scripts to reproduce figures of the publication.
analyzeResults<>.py: scripts to get key numbers reported in the publication.
The other files are mainly containing classes and functions called in main.py.

Data and Results

To keep this repository small in size, we do not include raw data and results. Instead we provide .npy files with processed data and results. Please find raw data and results on our SimTK project page:

OpenSimModel/new_model/experimentalData.npy contains all processed experimental data and was generated from running extractExperimentalData.py.
Results/optimaltrajectories.npy contains all processed results.
We do provide results from one case: Results/Case_42, which is the default case.

Visualize simulations in OpenSim

Launch OpenSim
Open model, eg OpenSimModel/new_model/Model/new_model_scaled_FK_contacts.osim
Load motion, eg Results/Case_42/motion.mot
Associate Motion Data, eg Results/Case_42/GRF.mot

Citation

Please cite this paper:

Falisse A, Afschrift M, De Groote F (2022) Modeling toes contributes to realistic stance knee mechanics in three-dimensional predictive simulations of walking. PLoS ONE 17(1): e0256311. https://doi.org/10.1371/journal.pone.0256311

antoinefalisse / predictsim_mtp