RehabWeek 2025 Workshop

Open

RehabWeek Workshop, Chicago, IL, Monday, 12 May 2025 at 08:15 - 09:45

Biomechanical measurements of movement can help predict and prevent injury, monitor disease progression, and inform clinical interventions. Traditionally, these measurements require expensive equipment and trained personnel to process, aggregate, and share these data, which limits access to these methods and studies to a small number of participants. Our team has created two new tools–OpenCap and AddBiomechanics–which have enabled hundreds of researchers to more quickly and easily collect, analyze, and share movement data.

• http://opencap.ai (Uhlrich et al., 2023) measures three-dimensional human movement using smartphone videos.

• http://addbiomechanics.org (Werling et al., 2023) automatically processes motion capture files to generate a scaled musculoskeletal model and compute joint kinematics and kinetics.

In this workshop, participants will learn how these new tools have been validated and how they expedite lab-based and out-of-lab studies of hundreds of participants, with applications to movement screening, injury prevention, and monitoring rehabilitation. Through a series of hands-on examples and demonstrations, workshop participants will learn how to incorporate OpenCap and AddBiomechanics in their research. The workshop will help participants build expertise in common workflows for simulation and biomechanical analysis and provide guidelines for obtaining high quality data.

AddBiomechanics Demo

Follow the steps below if you would like try AddBiomechanics during the hands-on portion of the workshop following the didactic materials.

1. Create an AddBiomechanics account.

   1. Go to the AddBiomechanics website: https://addbiomechanics.org/

   2. Click on “Go to app” in the top right corner.

   3. You will be prompted with a login window. Click “Sign Up” on the bottom to create an account.

2. Create a dataset and subject.

   1. After logging in, you will be directed to your “My Shared Data” page.

   2. Type in a dataset name in the text field “New Dataset Name” and click “Create New Dataset”.

   3. Click on the newly-created dataset name to go to the subjects page.

   4. Type in a subject name in the text field “New Subject Name” and click “Create New Subject”.

   5. Click on the newly-created subject name to begin entering subject information and upload data.

3. Download the sample data:

   1. Unzip the folder into a local working directory of your choice.

   2. It will contain two subdirectories: “Data” and “Results”.

   3. The “Data” folder contains a generic, unscaled OpenSim model file and marker trajectories (.trc) and ground force files (.mot) with matching names for each file. The folder also contains a file subject_metrics.yml, which contains subject metrics. This information is also provided in the table below for your convenience.

4. Fill out the subject metrics and upload data.

   1. Follow the step-by-step instructions on the subject page to fill out the subject metrics. Be sure to add a subject tag (e.g., “Unimpaired”)!

   2. Drag and drop marker trajectory files (.trc) and ground reaction force (.mot) files from the provided sample data to create trials. Any number of trials from the provided sample data may be uploaded.

   3. Add trial tags.

5. Process the data.

   1. Click “Process” to submit the subject’s data trials for processing on Stanford’s Sherlock compute cluster.

   2. The “Status” bar at the bottom of the subject page will cycle through the following states:

      1. “Waiting for server”: waiting for our automated processing server to pick up the trial.

      2. “Queued on Sherlock”: the server has picked up the trial and is now in Sherlock’s processing queue.

      3. “Processing”: the data is now being processed via AddBiomechanics sequential optimization workflow.

      4. “Needs review”: the data has finished processing, but trials with ground reaction forces need review.

      5. “Done”: the data has finished process and all trials have been reviewed (if needed).

   3. If you wish to reprocess the data to include frames that were manually labeled during the dynamics fitting step, click “Reprocess” to resubmit your data for processing.

Subject Metrics

Pre-processed Sample Data

While you are waiting for for your uploaded data to finish processing, you can inspect the subject data which has already been processed. Follow the this link to view and interact with the processed data on the AddBiomechanics subject page: Pre-processed sample data.

The “Results” folder in the sample data above also contains already-processed data files from AddBiomechanics.

Slides

Coming soon!

GaitDynamics: A Foundation Model for Analyzing Gait Dynamics

Data driven-models are a promising solution to quantify gait dynamics with less cost and time compared with traditional lab-based experiments. Typical data-driven models are limited to a specific downstream task with fixed inputs and outputs. GaitDynamics is the first generative foundation model to quantify gait dynamics for different gait patterns. It is a flexible, scalable solution that is able to predict both motions and forces for human walking and running with partial or no input data. Postdoctoral fellow Tian Tan and colleagues from the NIH-funded Restore and Mobilize Centers and the Wu Tsai Human Performance Alliance developed the model and have made it open-source. 

Access GaitDynamics code | Read preprint 

Publications

Uhlrich SD, Falisse A, Kidziński Ł, Muccini J, Ko M, et al. (2023) OpenCap: Human movement dynamics from smartphone videos. PLOS Computational Biology 19(10): e1011462. https://doi.org/10.1371/journal.pcbi.1011462

Werling K, Bianco NA, Raitor M, Stingel J, Hicks JL, et al. (2023) AddBiomechanics: Automating model scaling, inverse kinematics, and inverse dynamics from human motion data through sequential optimization. PLOS ONE 18(11): e0295152. https://doi.org/10.1371/journal.pone.0295152