Abstract

abstract: “Effective biosecurity practices are crucial for preventing the introduction and spread of infectious pathogens in swine production. To address the lack of quantitative supporting evidence for choose specific biosecurity measures, a novel interpretable machine learning toolkit, MrIML-biosecurity, was developed. This tool benchmarks farms and production systems by their predicted risk and quantifies the influence of biosecurity practices on disease risk at individual farm levels. It was found that 50% of 42 variables were associated with fomite spread, while 31% were linked to local transmission. The methodology involved training machine learning algorithms to classify farms based on their PRRSV status using survey data from 139 herds. Interpretations revealed significant contributions to predicted outbreak risk from practices related to employee turnover, density of surrounding swine premises, shared haul trailers, and distance from public roads. Quantifying and ranking biosecurity practices by their efficacy can facilitate more informed choices of biosecurity strategies.”