Projects

Evaluation of Adapted Exercise Modes

  • Conducting community-based evaluations of adapted exercise modes using wearable sensors and musculoskeletal modeling to assess musculoskeletal safety and participant enjoyment.

  • Data collection combines graded exercise testing on a custom wheelchair ergometer (following ACSM protocols) with personalized workouts performed in community settings.

  • The study recruits participants across a broad age range and varied activity backgrounds (athletic and non‑athletic) to improve generalizability.

  • Primary outcomes include shoulder loading and propulsion metrics derived from musculoskeletal modeling and sensor data, alongside self-reported enjoyment to compare exercise modes.

  • Presentation: Evaluation of propulsion biomechanics during exercise with a suspended-wheel everyday wheelchair (American Society of Biomechanics Annual Meeting, 2024)

Smartwatch-Based Wheelchair Propulsion Monitoring

Developing a novel method to predict manual wheelchair propulsion kinetics using only smartwatch sensor data. This work addresses the limitations of gold-standard devices (e.g., the SMARTWheel), which are no longer widely available and are restricted to laboratory use. The system aims to predict all six components of hand reaction loads (three forces and three moments) and to enable calculation of common wheelchair propulsion metrics. This pilot study is evaluating whether consumer-grade wearables can provide accurate, field-deployable measures of biomechanical loading to support injury prevention and rehabilitation for manual wheelchair users.

  • Presentation: Predicting manual wheelchair propulsion kinetics using smartwatch data: a pilot study (MRC Annual Symposium, Washington University in St. Louis, 2025)

Handcycling Force Prediction from Wearable Sensors

Investigated the feasibility of predicting continuous hand reaction forces during handcycling using only arm segment kinematics. This work compared multiple machine learning architectures including temporal convolutional networks (TCN), residual networks, and ensemble methods against traditional statistical approaches. The TCN model achieved correlation coefficients up to r=0.97 for in-plane forces, enabling replacement of instrumented crank handles in future studies.

  • Publication: Kinematics-Based Predictions of External Loads during Handcycling (Sensors, 2024)

  • Presentation: Estimating hand reaction forces from arm segment accelerations during handcycle propulsion using machine learning (Congress of the International Society of Biomechanics, 2023)

Flexible Multi-Camera Markerless Motion Capture Pipeline

Developing a scalable motion capture system that works with any N≥2 consumer cameras to achieve research-grade 3D motion tracking. The complete pipeline includes:

  • Camera Calibration: Estimating intrinsic parameters for each camera using Zhang’s method with planar calibration targets
  • Lens Distortion Correction: Applying radial and tangential distortion models to correct image artifacts
  • Camera Pose Estimation: Solving the Perspective-n-Point (PnP) problem to determine extrinsic camera parameters
  • Keypoint Tracking: Using deep learning-based keypoint detection (DeepLabCut) to track anatomical landmarks across multiple views
  • 3D Reconstruction: Triangulating 2D keypoints into accurate 3D coordinates using multi-view geometric constraints

Machine Learning for Equine Ground Reaction Force Estimation

  • Estimated equine ground reaction forces using kinematic and anthropometric data with machine learning models.

  • Developed statistical approaches to predict loading during walking and trotting, providing non-invasive methods for veterinary biomechanical assessment.

  • Publication: Statistical approaches for estimating forelimb ground reaction forces in foals during walking and trotting (Journal of Biomechanics, 2025 — Submitted)

Finite Element Analysis of Equine Bone Fracture Risk

  • Analyzed high fracture risk regions of equine bone using subject-specific finite element models.

  • Created detailed models from CT data to understand stress distribution in fracture-prone areas of the third metacarpal.

  • Presentation: Bone quality differences in fracture-prone regions of the equine third metacarpal (Orthopaedic Research Society, 2021)