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

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