Ross Miller
UNIVERSITY OF MARYLAND, COLLEGE PARK
Title: Integration of OpenCap and instrumented insoles to improve knee kinetic measures estimation
Collaborators:
- Jae Kun Shim, PhD (University of Maryland)
Abstract:
Kinetics (e.g. joint moments, muscle and joint forces) are related to many health- and performance-related factors in human movement, for example the knee adduction moment in walking in relation to osteoarthritis. The accuracy of joint moment calculations benefits from the inclusion of ground reaction force data. OpenCap software currently does not use ground reaction force measurements for inverse dynamics calculations of joint kinetics, which may limit the accuracy of the calculations. We have developed a prototype instrumented shoe insole with five piezoelectric sensors for estimating ground reaction forces in a portable fashion with a combination of sensor measurements and machine learning. We aim to integrate the ground reaction forces from our wearable product with the smartphone-based kinematics of OpenCap to solve the bottom-up inverse dynamics problem, which we hypothesize will improve the accuracy of knee joint kinetics.
Camila Torriani-Pasin
UNIVERSITY OF TEXAS AT EL PASO
Title: Sit2Stand: Validity in Stroke and Aging Populations in the Clinic and At Home
Collaborators:
- Shashwati Geed, PhD. (University of Texas at El Paso)
Abstract:
This research project aims to validate the Restore Center’s Sit2Stand assessment tool in chronic stroke patients (stroke>6 months prior) and demographically matched older adults. The primary objective is to establish Sit2Stand’s concurrent validity in clinical and home settings and to investigate its potential for predicting community gait. The study is structured into three key objectives:
Aim 1: Establish concurrent validity by comparing Sit2Stand with the Timed-up-and-go (TUG) test in both clinical and home environments. We hypothesize significant correlations between Sit2Stand and TUG, demonstrating its effectiveness in assessing functional capacity in stroke patients. Aim 2: Validate Sit2Stand by comparing it with ActivPAL®, a wearable device measuring sit-to-stand transitions and joint angles. This analysis will provide insights into Sit2Stand’s utility for monitoring mobility and joint kinematics. Aim 3: Explore whether Sit2Stand outcomes can predict community gait, as measured by daily steps. This predictive analysis will provide valuable information on the tool’s applicability as a functional assessment tool.
In both clinical and home settings, participants will perform Sit2Stand and TUG assessments, with joint angles captured using APDM sensors. ActivPAL® devices will record sit-to-stand transitions and daily steps. This research project aims to provide a validated, user-friendly tool for assessing motor function and rehabilitation progress in stroke patients. By enhancing our understanding of Sit2Stand’s utility and its potential for remote functional monitoring, our findings can contribute to the development of more effective stroke rehabilitation interventions, ultimately improving the quality of life for stroke survivors.
Scott Uhlrich
UNIVERSITY OF UTAH
Title: Incorporating data-driven models into musculoskeletal simulations to accurately estimate joint loads in osteoarthritis using OpenCap
Abstract:
Scalable tools for estimating musculoskeletal dynamics could improve the diagnosis and treatment of movement-related conditions. However, assessing these dynamic quantities, like joint loads, has been limited to expensive motion capture laboratories with extensive biomechanics expertise. The purpose of this pilot study is to create a framework for improving the accuracy of OpenCap dynamic estimates for specific rehabilitation applications by incorporating data-driven models into the musculoskeletal simulations (i.e., hybrid models). We will evaluate the utility of these hybrid models by predicting joint loads in individuals with knee osteoarthritis. These loads inform disease progression and intervention design, but the original OpenCap simulation framework could not accurately estimate these load magnitudes. We will first evaluate how accurately physics-based, machine-learning, and hybrid models predict knee contact forces measured with an instrumented implant using kinematic inputs alone. We will then evaluate whether the hybrid models improve the estimation of knee contact forces directly from two smartphone videos. We expect that the hybrid models will be more accurate than an end-to-end machine learning model, especially when trained on smaller datasets. The success of this project will provide a framework for other rehabilitation researchers to leverage high-quality biomechanics data to improve the accuracy of OpenCap dynamic estimates. Furthermore, we will deploy the accurate hybrid model for osteoarthritis on the OpenCap web application, enabling joint loads to be easily measured in future observational and interventional studies.