Why is LL beneficial to vets? 1. We are biased 2. We have limited temporal resolution, so... We miss things that happen quickly There is evidence of poor agreement between experts LL samples at 200 x/sec Objective methods not subject to bias 1993 GOAL: Develop an objective method of lameness evaluation in horses We did not know what to measure potentially unlimited number of things to measure Concentrated on kinematics and not on kinetics we colloborated with computer engineers this is a data mining problem vertical motion of the torso (head and pelvis) most senstive indicators of lameness and best differentiators of right vs left-sided lameness 1998 concentrated on torso movement We did not know what to measure..... max, min, area under curve, frequency? We collaborated with mechanical engineers (good signal processors) this is a fault detection problem Later: could determine when (impact vs pushoff) the fault (i.e. lameness) was occurring. 2000 Keegan KG, Pai PF, Wilson DA, Smith BK. A curve-fitting technique for evaluating head movement to measure forelimb lameness in horses. Biomedical Sciences Instrumentation 36 (ISA [International Society for Measurement and Control volume 395]:239-244, 2000). we collaborated with electronics engineers 2002 Keegan KG, Yonezawa Y, Pai PF, Wilson DA. Telemeterized accelerometer-based system for the detection of lameness in horses. Biomedical Science Instrumentation 38 (ISA [International Society for Measurement and Control volume 419]:112, 2002). 2008 2009 2010-2012 1st 9 beta testers 1st commerical version AAEP 2009 70+ units in use around the world what is the purpose of telling you all this? we didn't just think of this overnight this is a targeted approach to developing a lameness tool we didn't just adapt existing technology to the problem of lameness in horses everything about LL is the way it is because it works best that way for the practitioner Current Makeup of LL weight of sensors = 30 grams size of sensors = 3.5 x 3.4 x 2 cm range of transmission = 100 m live data collection 3 minute setup analysis and reporting takes seconds should not interfere with normal lameness evaluation LL gives the vet 3 pieces of information limb/limbs involved severity timing of lameness (impact vs pushoff or both) LL can be used to evaluate the effectiveness of a block use the dual report 95% confidence intervals for change in lameness Validation of LL LL is precise (repeatable) Test-retest 119 horses intra-class correlation coefficient Head ratio (0.87) MAXDIFFHEAD (0.92) MINDIFFHEAD (0.94) Pelvic ratio (0.92) MAXDIFFPELVIS (0.94) MINDIFFPELVIS (0.95) intra-class correlation coefficient for force plate = 0.84 95% confidence intervals MAXDIFFHEAD (6.8 mm) MINDIFFHEAD (6.7 mm) MAXDIFFPELVIS (3.1 mm) MINDIFFPELVIS (2.5 mm) LL is accurate correlation between force plate and inertial sensors very good 95% prediction interval of inertial sensors = +/- 6% VGRF LL is sensitive Head to Head Comparison with Subjective Evaluation Induced Lameness; 3 evaluators blinded to limb First to pick correct limb Forelimb LL picked correct limb first in 63.3% of trials Evaluators picked correct limb first in 3.3% of trials Tie in 33.3% of trials Hind Limb LL picked correct limb first in 80.0% of trials Evaluators picked correct limb first in 13.3% of trials Tie in 6.6% of trials LL is valid CURRENT USERS 70+ units world wide 16 veterinary teaching institutions USA, Canada, Sweden, Scotland, Ireland, Brazil, France, Germany, Libya LL FAQs You need to train to effectively use LL LL is marketed ONLY to equine vets LL is a practice builder pays for itself is used 3-4 times per month We are not just selling equipment software support maintenance warranty training upgrades conclusion LL is a precise, accurate, sensitive augmentation of subjective evaluation of lameness Further information can be found.... www.equinosis.com http://lameness-locator.blogspot.com/ TIME LINE OF DEVELOPMENT CLICK TO PLAY MOVIE CLICK TO PLAY MOVIE Go collect lots of data! known sound horses known lame horses natural induced 5 years, hundred of horses, high-speed camera based kinematic gait analysis wavelet analysis neural network classification CONCLUSION Type of Frequency Analysis CLICK TO PLAY MOVIE Vertical torso movement modeled as a surface vibration Perturbation of normal movement from lameness fault detection equation 1st wireless prototype CLICK TO PLAY MOVIE RF LH RAYS LONGER 1986 WORLD CUT ARGENTINA VS ENGLAND MARADONA'S "GOD'S HAND" 2010 WORLD CUP GERMANY VS ENGLAND NO GOAL LEFT FORELIMB LAMENESS with compensatory right hind RIGHT HIND LIMB LAMENESS with compensatory right front less severe right hind with compensatory right front mild left front with compensatory right hind mild right front mild right hind no lameness Evaluation of lameness while lunging compare lunge left to lunge right Evaluating flexion tests