A Prescriptive Approach Based on Fuzzy Cognitive Maps and Genetic Algorithms for Disease Management in Beef Production

Livestock disease diagnosis and treatment often rely on the experience of veterinarians and the availability of clinical signs, which can vary significantly between cases. This paper proposes a novel prescriptive analytics approach based on Fuzzy Cognitive Maps (FCMs) integrated with Genetic Algorithms (GAs) to support decision-making in the treatment of common cattle diseases. The FCM captures expert knowledge through causal relationships between symptoms, treatments, and diagnoses, while the GA optimizes treatment actions to achieve desired health outcomes. We evaluated our approach using three case studies –babesiosis, anaplasmosis, and coccidiosis– on datasets comprising 3000 cattle records each. The predictive model achieved accuracies of 92%, 87%, and 87% for the respective diseases. The prescriptive model yielded high performance with average $R^{2}$ values above 0.93 and low RMSE values, demonstrating that the recommended treatments closely matched the optimal solutions. This work contributes a hybrid, explainable, and data-efficient framework that can be integrated into intelligent agriculture systems for improved livestock health management.