[Validation of a decision tree for selective dry cow therapy of dairy for a digital expert system].

Abstract

In this study, a decision tree derived from scientific literature on selective dry cow therapy (ST), which was developed as a knowledge base for a digital expert system, was evaluated. The decision tree merges algorithmic (based on cell count results) and cultural (based on milk sample findings) approaches.During a two-year project period (August 2021-September 2023), ST was carried out on 19 dairy farms in southern Germany according to the decision tree, without specific requirements being placed on the herd's udder health before the start. A total of 1,369 dry-off observations were recorded. The dry-off cases were evaluated regarding implementation of the recommendations, cell count at the first milk recording after calving, proportion of new infections and cures during the dry period depending on the implementation of the dry-off recommendation and the udder health status of the herd.Across all farms, 38.4% of cows were dried off without the use of an antibiotic. The potential for saving antibiotics varied greatly between farms (range: 2.4-71.3%). In 75.9% of cases, a dry-off recommendation could be made based on the available udder health data; only in around 25% of cows did the dry-off recommendation require a microbiological examination of quarter milk samples. On average, the milk cell count after the dry period was less than 100,000 cells/ml in animals that were dried off with and without an antibiotic dry-off preparation. The proportions of new infections and cures during the dry period did not differ significantly between animals with and without an antibiotic dry-off preparation.The results show that with the help of the developed decision tree, ST can be carried out safely in dairy farms without endangering udder health. This decision tree can therefore serve as a reliable knowledge base for a digital expert system to optimize dry-off management in dairy farms.