This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Lecture 4 SystemsSystems-Level Modeling Approaches EML 5595 Mechanics of the Human Locomotor System Outline Assign journal articles to review Similarities and differences in the two articles for today First journal article review Erdemir et al. (2007) Second j journal article review Buchanan et al. (2004) Article Similarities Both seek to "provide insight neural control and provide tissue loading" and thus to "contribute to improved diagnosis and management of both neurological and orthopaedic conditions" conditions (Erdemir et al, 2007). Both discuss the use of musculoskeletal models and optimization optimization. Both discuss how to model muscle and tendon as well as musculoskeletal geometry. Both discuss forward and inverse dynamics approaches for predicting muscle forces. Both discuss the limitations of the approaches they present. Article Differences Erdemir et al. (2007) focuses on "optimization approaches" for predicting muscle forces, while Buchanan et al (2004) focuses on "EMG-driven al. "EMGEMGEMG approaches." Erdemir et al. (2007) use EMG data for model evaluation, evaluation, while Buchanan et al (2004) use al. EMG data for model calibration. calibration. Buchanan et al. (2004) spend more time describing modeling of muscle activation and contraction dynamics than do Erdemir et al. (2007). Erdemir et al (2007) Overview al. Musculoskeletal dynamics Equations of motion Musculoskeletal geometry g y Muscle modeling Forward vs. inverse dynamic solutions Muscle force estimation Inverse dynamics optimization methods Forward dynamics optimization methods Optimal control methods Alternate methods (mainly EMG-driven) ( y EMG) Challenges Musculskeletal Dynamics Musculoskeletal Geometry Muscle Moment Arms Muscle Modeling Forward vs. Inverse Dynamic Solutions Forward vs. Inverse Dynamic Solutions Muscle Force Estimation Muscle Force Estimation Alternate Methods Leave discussion of EMG-driven approaches EMGto Buchanan et al. (2004). Challenges Predicting temporal characteristics of EMG p g p profiles does not guarantee accurately calculated muscle forces. Accurate prediction of muscle forces depends on p p correct values of model parameters that are difficult or impossible to measure. Description of joint mechanics has a significant impact on calculated muscle forces. Why? Choice of optimization cost function can have a significant impact on calculated muscle forces. What is the right cost function for normal and impaired subjects? Direct validation of predicted muscle forces is impossible. Buchanan et al (2004) Overview al.
Four Step P F -St Process to Estimate Muscle Forces Fourt E ti t M l F Step 1: Neural signal to muscle activation Step 2: Muscle activation to muscle force Step 3: Muscle force to muscle moment Step 4: Muscle moment to skeletal motion
External Forces and Moments Neural Command C 1 , 2 Musculotendon d Dynamics F1 F2 Musculoskeletal Geometry G t (moment arms) T1 T2 Eqns 1 of 2 Motion 1 2 1 2 Buchanan et al (2004) Overview al.
Four Step P F -St Process to Estimate Muscle Forces Fourt E ti t M l F Step 1: Neural signal to muscle activation Step 2: Muscle activation to muscle force Step 3: Muscle force to muscle moment Step 4: Muscle moment to skeletal motion Emphasis is on the first two steps. Why? The authors obviously believe use of EMG signals is valuable and appropriate. Why? g pp p y Neural Signal to Muscle g Activation FirstFirst-Order Continuous Activation Dynamics Neural Signal to Muscle g Activation SecondSecond-Order Discrete Activation Dynamics Neural Signal to Muscle g Activation NonNon-linearization "Tunable" Model Parameters Tunable Activation dynamics , 1, 2, A Contraction dynamics Fom, lom, lst, o, vom
What is the problem here? And how can we get around it? Model Parameter Tuning Model Parameter Tuning Model Parameter Testing
Crossover Cut Straight Run Calibrate, then test! For Next Time Download and start working on OpenSim g p Tutorial 3 Download and read Wang et al. (1990) J Bone Joint Surg Am 72: 905-909 905Download and read Lin et al. (2006) Arch Phys Med Rehabil 87: 562-568 562- ...
View Full Document
This note was uploaded on 02/15/2012 for the course EML 5595 taught by Professor Staff during the Spring '08 term at University of Florida.
- Spring '08