New standards in stochastic simulations of dairy cow disease modelling: Bio-economic dynamic optimization for rational health management decision-making

Abstract

Bioeconomic models applied to animal health issues are now commonly observed in literature. This section of literature is very heterogeneous and the underlying methods are very diverse, from very simple methods (partial budgeting) to very complex ones. The objective of the present study is to build a new dynamic stochastic optimisation bioeconomic model applied to the dairy cow sector, that goes beyond some limitations usually found in methods used up to now. First, based on a critical literature review, we highlight four issues of bio-economic stochastic simulation models (BESSMs) applied to dairy cow diseases at the farm level. These models appear as partial (the farm system is not considered as a whole), unbalanced (between the economic and biological parts of the model), closed (to the farm environment) and only partially dynamic. To address these 4 main issues and improve the methodological standards in the microeconomics of dairy cow health management, we secondly develop a new bio-economic sequential optimization model (BESOM), called DairyHealthSim. DairyHealthSim aims to better consider both the context of decision-making and the farming system dynamics to define the best health management strategies in a given context. The biological part of the model simulates the complex dairy production cycle with a holistic approach. It is defined on a cow-week basis, and the weekly probabilities for all cow events, including production, reproduction and diseases, are simulated. The economic part of the model is a mean-variance optimization framework that dynamically represents the farmer's input allocation decision process under constraints. The biological and economic parts are closely integrated and the model is running with back and forth between the 2 parts of the bioeconomic model. Third, an application involving farmers' strategies related to biological risk management, labour willingness and market demand is proposed for dairy production and mastitis management. The results highlight the added value of the farming system-driven system coupled to economic optimization approach. DairyHealthSim identifies the optimal scenario for the entire ten-year simulation period or is based on yearly optimization (sequential modelling). The two different optimal solutions found show the usefulness of considering the dynamics and complexities of the actual field situation. The opportunity cost between the best and alternative solutions demonstrates that some solutions are economic equivalents. In conclusion, compared to approaches where the outcome is reduced to the monetary impact of diseases, DairyHealthSim is far more precise and appropriate for supporting decision-making.