Prediction of lameness based on the percent body weight distribution to individual limbs of Karan Fries cows

The present study was carried out to predict various categories of lameness by using percent body weight distribution to all the four limbs of lactating Karan Fries crossbred cows. Load sensors were able to detect an abnormal distribution of body weight to individual limbs. Among the cows, 22.97% were afflicted with mild, 14.19% with moderate, 21.62% with lame, 6.75% with severe and rest with no lameness. Prediction of 5-point scale of lameness based on percent body weight distribution to all four limbs of cows through multi-nominal logistic regression could be done with an accuracy of 70.6, 47.1, 0.0, 50.0 and 0.0% into no lame, mild, moderate, lame and severe lame categories. Cox and Snell, Nagelkerke and McFadden pseudo R² values for the present model in prediction of various categories of 5-point scale of lameness using percent body weight distribution to all four limbs was observed to be 32.0, 33.7 and 12.9% respectively while these values for 3-point scale were observed to be 46.0, 54.0 and 32.3%, respectively. The percent correct classification of no lame, mild and severe lame category in 3-point scale was observed to be 91.8, 0.0 and 73.8%, respectively. The overall percentage of correct classification rate of 3-point scale was much higher (73.6%) as compared to 5-point scale (45.9%).