Which sEMG variable best distinguishes between non-lame and induced lameness conditions in horses?

This study aims to determine the most effective surface electromyography (sEMG) measure, from individual muscles, for distinguishing between non-lame and induced fore- (iFL) and hindlimb (iHL) lameness conditions. We compared bilateral sEMG absolute value and asymmetry measures from triceps brachii (triceps), latissimus dorsi, longissimus dorsi (longissimus), biceps femoris (biceps), superficial gluteal (gluteal) and semitendinosus. sEMG and 3D-kinematic data were collected from n=8 clinically non-lame horses during in-hand trot. iFL and iHL were induced on separate days using a modified horseshoe, with baseline data initially collected each day. MinDiff was calculated from poll (HDMin) and pelvis (PDMin) vertical displacement. sEMG signals were DC-offset removed, high-pass filtered (40 Hz), and full-wave rectified. Average rectified value (ARV) was calculated per stride and normalised to the maximum observed value of individual muscles at baseline (sEMGabs). The difference between right and left normalised ARV were calculated per muscle and stride (sEMGasym). Receiver operating characteristic analysis was conducted, using area under the curve (AUC) as an overall measure of discrimination between baseline and iFL/iHL. MinDiff showed excellent discrimination for iFL (HDMin AUC=0.98) and iHL (PDMin AUC=0.96). Triceps sEMGasym was best (AUC=0.84) for differentiating iFL, with other measures showing fair to chance discrimination (AUC range=0.49-0.76). Lame-side gluteal, semitendinosus and longissimus sEMGabs (AUC range=0.90-0.91) showed excellent discrimination for iHL, followed by biceps sEMGasym (AUC=0.83). Evidently, the choice of sEMG variable is muscle/condition dependent: sEMGabs performed acceptably across muscles/conditions, but triceps and biceps sEMGasym, respectively, appear superior for discriminating iFL and iHL