“Dairy Brain” project leaders share updates at second advisory committee meeting

The UW2020-funded “Virtual Dairy Farm Brain” project held its second Advisory Committee meeting on Feb. 20 at the Wisconsin Institute for Discovery. The big-picture goal of the project is no small feat: to develop a way to collect and integrate all of a farm’s various data streams in real time and then use artificial intelligence to analyze those data to help farmers make better management decisions.

Twenty-six people attended the second Advisory Committee meeting in person, with an additional nine joining virtually. Attendees included project staff, participant farmers, other academics, and industry representatives. Industry professionals included companies from data analytics, animal health companies, breeding associations, dairy herd improvement representatives, software providers, milk processors, milking equipment companies, nutritional companies, sensor producing companies, and farm consultants (nutritionists, veterinarians, etc.).

The purpose of the meeting was to: 1) Bring the Advisory Committee up to date on Dairy Brain project accomplishments of the past year; 2) Enlist Advisory Board members in helping define data security needs; and 3) Obtain Advisory Board ideas for the project’s future for long-term sustainability.

Below are some of the updates that were shared at the meeting, provided by dairy science professor Victor Cabrera:

Originally, the plan was to work with only one farm on this project. However, we decided to include more farms to become more familiar with the diverse technologies that are being used by farmers, including different sensors, machinery and/or software programs. Having more farms will help us to scale up the project much faster in the future. We feel very proud that we now have 6 farms enrolled. 

We have, on average, between 70% or 80% of the farms’ data flowing to our server in the Wisconsin Institutes for Discovery. That in itself is a big accomplishment. We have had challenges of connecting data streams from different sources in real time, probably more than we anticipated, but we have a protocol in place that will help to progressively make more of this integrated data available to our analytical tasks. In parallel, we have identified and advanced analytics (algorithms) in 5 management areas that will be ready to be deployed once the integration of data becomes available.

More info about the 5 management areas: 

  • Nutritional Grouping: Provide more precise diets by better allocating cows to pens and re-formulating diets (This area is ~70% of the way to completion)
  • Ketosis: Find cows at risk of ketosis disease in early lactation. These cows can be treated effectively (This area is ~80% of the way to completion)
  • Monitoring Clinical Mastitis: Find cows at risk of mastitis according to their genetic traits (This area is ~50% of the way to completion)
  • Early detection of Clinical Mastitis: Flag cows at risk of mastitis five milkings before (This area is ~70% of the way to completion)
  • Net Present Value: Project the probabilistic cash flow of a cow in the long term. This value is critical to make cow-level decisions such as treatments, breeding, selection (This area is ~50% of the way to completion)

Finally: We have submitted a $1 million grant to the USDA-NIFA-FACT (Food and Agriculture Cyberinformatics Tools) program, which is under review at the moment.