Farmers who implemented this, also implemented that: Use of association-rule-learning to improve biosecurity on dairies

Biosecurity practices are the cornerstone of disease prevention and control programs. In Canada, their implementation is evaluated with a Risk Assessment Questionnaire (RAQ). Association Rule Learning (ARL) – a non-supervised machine learning algorithm – is widely used in marketing for consumer segmentation based on purchase patterns. This technique may help veterinarians to recommend biosecurity practices that are more likely to be adopted by producers. In this project, we applied ARL to 3825 RAQ completed by Québec dairy producers to generate 22 million rules that identified combinations of self-reported practices frequently applied together. We retained the best 63 rules predicting the adoption of 13 biosecurity practices with a confidence ≥ 70 %. ARL is useful in studying the relationship between biosecurity practices on dairy farms. By identifying biosecurity practices more likely to be implemented by a given producer, veterinarians can provide targeted recommendations that might improve disease prevention and control programs. • Association Rule Learning (ARL) is a flexible technique to analyse questionnaires. • ARL predicted 13 key biosecurity practices more likely to be implemented by farmers. • Implemented practices were predicted with a confidence higher than 70 %. • ARL allows prescriptive use of questionnaires and personalized recommendations. • ARL may enhance uptake of disease prevention and control programs.