The proliferation of data sources and types enables scientists to identify new patterns and correlations in animal growth, development and behaviour. Hendrix Genetics, a company located in Boxmeer, the Netherlands, is investing heavily in the development of an Artificial Intelligence (AI) platform that will take the current state of animal genetics technology to the next level. Bram Visser, a data scientist at Hendrix Genetics, explains how ‘smarter metrics’ can benefit his company and the sector as a whole.
‘We’re essentially talking about two parallel trends. The first is data proliferation: the availability of data from all kinds of new sources. Of course, we already use metrics to measure anything and everything, but that’s largely a physical process, conducted by human beings. However, we now also have access to data collected by devices such as cameras, thermometers, microphones and RFID readers (radio-frequency identification – Ed.), which can be visualised using earmarks and motion sensors. All this data combined makes up a massive cloud of big data. And that brings me to the second trend: artificial intelligence (AI) can help us to identify and interpret various non-linear relationships between all these data sources and subsequently make it accessible to users in the form of information.’
‘We’re only on the cusp of this new trend, so I would venture to say that the sky is the limit. But just to give you an example: it enables us to observe, with far greater precision than in the past, the behaviour of individual animals. This includes finding answers to questions such as: where in the barn are the animals located at a given time; which animals are linked through a social network; which ones connect specific social groups with each other, and so on and so forth. Since it’s increasingly common for animals to be kept in groups, this represents valuable information. For example, you can isolate specific social groups during an outbreak of an infectious disease and record the outcomes.’
‘That’s precisely the challenge we’re facing and are currently working on. We’re not just talking about a huge amount of data, but also data that’s extremely variegated. It would be impossible for a human being to find any sort of correlation between this data, but an AI platform has no trouble doing just that. I’ll explain it to you this way: when people set out to look for patterns, they tend to do so based on past experience. But there are all kinds of patterns with which we are unfamiliar, but that do have some sort of inherent value. Machine-learning algorithms are becoming more powerful all the time, and when more data will become available in the future, they will be able to find correlations which we never would have entertained in the past and which help us to further improve the results.
We are therefore focussing part of our R&D efforts on developing an AI platform, as they’re not commercially available in a one-size-fits-all format. Although we do use all sorts of open-source development tools, part of our job involves reinventing the wheel. On top of that, this is such a rapidly changing field that you need to be able to incorporate new technologies quickly and be highly adaptive.’
‘For starters, there’s the fact that we can offer new systems to our clients that allow them to manage their animals more efficiently and effectively. If you have a system that, say, sends out an alert when a random animal has exhibited low appetite for several consecutive days, you can intervene in time, as there is clearly something wrong there. But there are immense opportunities in our own business as well. We currently analyse data using traditional methods to predict the genetic value of animals, but in the future computers will be able to learn, correct and improve 24/7. And what’s at least as important is that this trend is occurring worldwide, using objective and standardised methods. To give you another example: an animal’s energy levels are indicative of its genetic health. But what might seem like “high-energy” or “spirited” behaviour to one person could appear “nervous” and “skittish” to another. By introducing smarter metrics, you can eliminate the subjectivity of human observation. And I could name countless other examples, as I feel we’ve only just begun to scratch the surface when it comes to the infinite potential of artificial intelligence.’
Read more about Hendrix Genetics