Summary of Scientific Goals & Features Needed to Achieve Them Wendy Murray Tom Buchanan Brian Garner Jeff Reinholt Ilse Jonkers Allison Arnold Peter Loan

Scientific Goals Model the neuromusculoskeletal system Validate the model (input and output) Simulate (Understand the interaction of components of the biomechanical system) Predict outcomes Adjust the model parameters

Software Features Needed Validation Module that allows the user to compare model performance to experimental data through visualization Minimizes work associated with external data and maximizes the user’s ability to evaluate the model’s performance Features such as scripts, templates, visualizations Subject Scaling and Specificity Complicated issue, but tools for scaling for each subject and inputting specific subject data are important Need to scale segment lengths & inertias, MT paths/parameters, joint kinematics Need an high quality generic model and good scaling algorithms so that we can readily scale generic data. We also need to have a performance measure so that one can argue “because I’ve scaled it I made it better”

Software Features Needed Estimation of muscle activations/forces during movement using dynamic optimization or EMG Compatibility Reverse compatible with SIMM Read imaging data readily MatLab compatible Improved Speed Parallel processing Real-time inputs Easy to use interface Support and maintenance New tools Many listed today…

New tools (1) Tools to import experimental gait data (FA) New & improved scalable full-body model (FA) Better numerical integrators (FA) Fast, accurate methods for modeling muscle path geometry (FA) Dynamics engine with no compile step or restrictions (Simbody?) (FA) Include neuromuscular control (DT) Stability of simulations (DT) More predictive capabilities (DT) Detailed ms models (microscopic) (DT) Efficient models (force enhancement, depression, activation-deactivation, etc.) (RN) Animation-making tools (SP) Bone editing (SP) Amenable to parallelization (RK) True dynamic optimization (end points specified, not path) (RK) Multi-scale modeling (RK) Fast, real-time simulations for VR and patient training (RD/GL) Means for handling contact (SP) Means for processing surfaces (SP) Wrapping surfaces (SP) Fast, accurate methods for modeling muscle path geometry (FA) Incorporation of latest & greatest methods (will only old things be public domain) (TvdB)

New tools (2) Robust excitation patterns (JH) Closed loop kinematics (WM) Kinematic dependence on multiple dof (WM) Improved muscle wrapping including branching (WM) Interactive scaling (WM) Easier user-interface for parameter editing (WM) Contact modeling (WM) Better foot-floor modeling—incompressible hyperelastic solids (SP) Integrated elastic-foundation contact modeling (SP) Contact model with friction (SP) Cartilage models (SP) More sophisticated muscle-tendon models—architecture, force generation, fascia, ligaments (SP) Analysis tools with better user interface (JH) Customizable/usability balance—comment all code and tell user what is changeable (SP) Make users appreciate that SimTK is not a model; generic model may not generalize (SP) Expandability/modularity (TSB) Difficult to calibrate a generic, parametric model to the patient (JR) Quick, easy, clear, concise, intuitive, powerful, versatile, & effective (JR) Developmental environment & math functions (JR) GUI that can be used by researchers and clinicians (RD/GL)