Getting Personal With Cow Metadata

October 20, 2021
By Daniel Foy
LinkedIn

Share this article

Summary

Individual cows are the foundation for enhancing our understanding the opportunities for improvements in their welfare and resulting performance.
Integrated “big data” ecosystems are key to improving farm management by supporting decision-making and enhancing the value of their own data.
Leveraging new IoT (Internet of Things) technologies are vital to improving our understanding of cow behavior and its relationship to improving the welfare and resulting performance of each cow across any herd size.

When we look at our own human health technologies, companies like Google with their Fitbit, Apple’s Apple Watch, or the prospects of Amazon getting into a host of human health products, they are essentially using hardware and software technologies that produce, collect, and transmit data to find out more about you as an individual, what is your baseline, what could be better, if there is something wrong and can we catch it earlier.

This type of data is highly sought after and valuable. It can be used to focus on better systems for population medicine, government and industry services. Big data and IoT solutions offer the ability for almost instant epidemiology— creating a clear picture of what is happening in the world wherever it is collected. Given the implications, this could be the next human health revolution.

On-farm sensors and data platforms are key to how we operate as dairy or ranch businesses, even though many farms are still using first-generation systems. Current farm data systems tend to lack connectivity, restricting data flow between them, that are across multiple dashboards and spreadsheets, that take time to manage, while their interfaces are difficult to navigate with poor user experience.

This current generation of data systems delivers data in many formats (defined and undefined). The data come from many different independent databases, requiring skilled consultants and academics to assist with the management and understanding of even the well-organized farm’s data sets.

We currently have no clear format on what the structure of a centralized farm database may look like. There are some academic projects like Dairy Brain at the University of Wisconsin-Madison exploring this, but limited farm level examples. To be able to get diagnostic, predictive, and prescriptive analytics, data infrastructure is required so that data can be handled, processed, moved, and analyzed while allowing the farm to maintain ownership of it.

The farmer owns their data but has no way of saying if their data is right or wrong. They cannot validate their system or data independent of the supplier or manufacturer if there are bad or inaccurate inputs. These are all things we need to know so that we can be confident in these digital tools and their accuracy, otherwise, these tools will be useless. More importantly, this includes the ability to see who is using your data and for what purposes.

We’re All Individuals and So Are They

All cows are different. They are highly social, sentient beings that I am sure we all have our raft of unique stories about. These individualities tell us more than we think and most of the time these are things that we don’t see with the naked eye.

We used to know all cows by name or behaviors in our herd, but as we move to larger herds and more automated systems, we now have to know these same cows by their individual data and data patterns, which is a whole new perspective toward monitoring and managing our herds with an even sharper eye.

Being able to see changes in an animal prior to a visual onset of symptoms would change how we do everything. This is where precision farming technology meets precision medicine and can result in better diagnostic, predictive and prescriptive information for improved decision support. When we see a lame animal visibly limping, it’s often too late. We need to have a higher view of what happened to that animal recently as well as the animal’s current and past lactation, or during the transition period to better understand causation. This may include many different data points, from many different data systems. Somatic cell count, lying time, rumination, feeding patterns, and video cameras are just a few examples of the types of data and systems that can be used. Together, this is called metadata—data that provides information about other data. All these data or metadata can give a comprehensive overview and insight to see what was happening before any visual onset occurs.

From individual cow metadata, we can build up an overview of outliers within the group or herd to better understand what is happening and to help determine the best course of action to take. Sometimes being an outlier is normal, but when an animal changes from her own baseline or her “normal”, that’s when we need to look a little closer. But we also want to make sure that this is just the individual cow and not the group. The ability to compare and drill down is powerful. If we are going to get to the next generation of digital audits or welfare metrics on a herd, this will be based on individual cow data, then built up to an assessment at the group and herd levels.

We Still Need Group Analytics

This doesn’t mean that we all just forget about group or herd analytics. No, we must get better at looking at the micro and the macro within our livestock population— from each individual cow to herd, whole farms, county, state, or even an entire industry (“Field to Fork”).

Population medicine, production medicine, and epidemiological systems are important to the future sustainability of the livestock production sector. Allowing agribusinesses and researchers access to individual animal and herd data like this will allow us to unearth new insights, solutions, and management processes to benefit farm businesses. It’s not only the health of each animal that benefits but also the profitability and viability of the entire sector.

But remember, data has value. Some even call it the “new oil.” Understanding ownership of data, how we keep it secure, and how the supply chain is using or accessing your data will further define and increase that value.

Different parts of the supply chain—service industry, processor, and consumer—have different needs from farm and ranch data. A high-level overview on patterns such as pen to pen, group to group, or averages of where their food or product has come from is good enough in some cases, but we can’t just say group or herd-level data is good enough. We need to look at what helps the individual animal, the farmer, and then everyone else up the supply chain and across the industry.

Understanding the group dynamic is also an important factor to consider within individual analytics, while also allowing for the identification of social behaviors, and how stress affects individual animal or herd health outcomes. This is especially important as we look at calves in groups, where socializing offers so many developmental benefits that aid in animal health later in life. It’s also important with periparturient cows where being able to express normal behaviors is important to the cow and can provide valuable insight into their health and welfare.

The example of incidence rates and prevalence shows how data from individual cows are used for different outcomes on a farm. Individual data is important and so is group or herd data. The incidence rate is numerically defined as the number of new cases of a disease within a period of time, usually as a proportion of the number of cattle at risk for the disease. The prevalence is the number of cases or events at a point in time or in a given time period. The incidence and the prevalence of a disease and an event are different ways to look at data and each provides an important understanding of the disease or event. A great example is lameness. The number of new cases of lame cows in a month is the incidence rate. Incidence and prevalence are linked in this formula: Prevalence = Incidence (new cases) X time (duration). The number of lame cows on a dairy today is the prevalence of lameness. The dairy management team needs to know the incidence rate of lameness to control lameness, extend the lifespan of the cows on the dairy, and stay profitable. The prevalence of lameness is used by welfare audits to assure consumers the dairy does not have too many lame cows. Incidence and prevalence rates require both individual and group level analytics; understanding these rates at a deeper level and in real-time is a must.

Without appropriate individual data input and individual level analytics we will get misleading cull rates on herds, which is also output as a group level assessment.

What are the Take-Aways?

To get back to knowing each of our cows better, we need to start with the individual cow’s metadata in mind, from which we can build our analytics up to a group, ten groups, a farm, or several farms while being confident in the quality and reliability of the data.

To get there, we are going to need to start by horizontally integrating current on-farm data systems into one central data ecosystem, that’s farm-specific. The current generation of data systems that are on farms do produce large volumes of data and some of these systems are producing more valuable data than others. We just need to get to the bottom of what systems these are and what is that data. That makes each individual cow the star of the show. With this approach, we can also better decide what future agritech solutions we need to deploy to improve the value of the data ecosystem. This will save time and capital.

Bringing farm data together from all on-farm disparate systems allows producers to control and take ownership of their data, which has tremendous value—on and off the farm. Working with producers, their data systems, and building a plan that includes understanding that data—why they are collecting it, how it flows, where it’s coming from, and what it means—is our next step in becoming livestock super stewards. This is going to take learning new kinds of languages, terms, and systems, but the great thing is that these are getting easier, more intuitive, more user-friendly, and with intelligent support.

This is not only going to help our on-farm teams better manage what’s going on by allowing them to see a farm through new, sharpened, digital eyes, this will assist us towards our goals of sustainability, welfare, and traceability.

Building out the necessary data infrastructure at the farm level allows data to feed and speed up other areas of agritech development in AI, camera AI, virtual assistance and virtual reality technologies, that big data at the individual cow level is critical for. Embracing these new types of data strategies on-farm will strengthen our pathway to becoming global leaders in the livestock sector and move us a step closer to “field to fork” intelligence of our products in the marketplace.