PSX-B-11 Classification Performance of Multinomial Logistic Regression for Identifying Resistance, Resilience, and Susceptibility to Gastrointestinal Nematode Infections in Sheep

Abstract

Abstract The objective was to investigate the feasibility of using easy-to-measure phenotypic traits to predict resistant, resilient and susceptible sheep to gastrointestinal nematodes via multinomial logistic regression (MLR). The database comprised 3,654 records on 1,250 Santa Ines sheep from six farms. The animals were classified into three responses to infection classes (resistant, resilient and susceptible) according to fecal egg count and packed cell volume. The MLR was used to predict such classes using the information of age, month of record, sex, Famacha degree, weight and body condition score as predictors, and a leave-one-farm-out cross-validation technique was used to assess prediction quality across farms. The MLR was able to predict with satisfactory performance the resistant and susceptible animals in two of the farms, with resistant precision equal to 79 and 77%, and susceptible recall equal to 68 and 83%, respectively. In addition, at least one of the classes was well predicted in four farms, with susceptible recall equal to 71 and 89% in two of these farms, and resistant precision equal to 86 and 100% in the other two farms. The model was not able to satisfactorily classify resilient class. The proposed approach could help attenuate the negative impacts related to infections caused by gastrointestinal nematodes, contributing to design deworming strategies that take into account the risk of an animal being contaminated, consequently reducing the costs with anthelmintic administration and laboratory analyses based on blood or fecal samples. The results suggest that the use of easily measurable traits may provide useful information for supporting management decisions on the farm level that could potentially contribute to reducing parasitic contamination and production costs. In addition, the animals identified as resistant can also be incorporated as selection candidates into breeding programs for genetic improvement of flocks.Supported by the São Paulo Research Foundation (FAPESP) grants #2020/03575-8, #2018/01540-2 and #2016/14522-7, SP, Brazil.