Students' Research Circle    
 
 
2022
2021
2020
2019
2018
» 2017
The conference
Veterinary Session
Veterinary Jury
Sponsors
Awards-list
HSF
Call for papers
2016
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
Home » Archive » 2017

TDK conference 2017

Predicition of puerperal metritis with a mathematical model in dairy cows
Muntyán Nóra Judit - year 6
University of Veterinary Medicine, Department and Clinic of Food Animal Medicine
Supervisor: Dr. Zoltán Szelényi

Abstract:

The puerperal metritis is a disease that affects all cow dairies and causes serious losses. The greatest damage is the loss of milk production, the increased medicine cost, and the cost of replacement heifers.

In the research we estimated the likelihood of metritis in a southern Hungarian dairy farm with the aim of a mathematical model published earlier, and we also examined the accuracy of the predictive model. Holstein-fresian cows took part in our examination (n=200). During the clinical research we measured the body temperature and milk yield from the second day for three days following the calving. On the first day after calving, we also measured the concentration of beta- hydroxy butirate of the blood from whole blood with a manual hand-held measuring device, using test strip validated for cattle. 10 days after calving, we examined the occurence of metritis by means of rectal examination. In addition, we collected data of the cows’ parity and calving ease, and the peripartal clinical diseases.

Based on the model we predicted metritis in those animals where the model showed more 50% likelihood. From the animals with metritis (n=21), local symptoms showing as well as toxic metritis occurred more frequently at the cows calving first time, 18 and 12 %, respectively. The model gave around two third of the metritis cows less, than 30% likelihood (Se: 33%, Sp:86%, Accuracy: 80%, Precision: 22%, OR = 3.1), and at higher likelihood value (80%) the estimation proved to be correct at only 40 %, approximately. The prediction for the likelihood of metritis and on the ROC curve analysis revealed the AUC = 0.625, at 50% cutoff value the accurracy is 0.79, which also showes the poor accurracy of the model. In case of toxic puerperal metritis the model is even less capable of differentiating positive animals from negative ones (Se: 83%, Sp: 15%, Accuracy: 28%, Precision: 19%, OR = 0.9).

As a conclusion, the mathematical model, in its current form, is not suitable to predict in clinical use. The main reason for this is that originally it described cows with higher parity, thus it can be used only hypothetically for calving heifers. It is likely to assume that the mathematical model needs to be revised, taking into consideration more variables. It would be necessary for an accurate prediction model that can be clinically used, and also the individualsare carrying out the examination must undergo training, which are indispensable in describing the alterations in accordance with the definition. In the future we are planning further examinations with an improved model for a more accurate prediction of metritis.



List of lectures