Data is available at various levels – from hatcheries and feed suppliers to farmers and veterinarians and finally from the slaughterhouses. It is only by gathering it and bringing it together that meaningful analysis can occur, delegates were told at the Porphyrio-sponsored symposium, "Biostatistics and big data at the service of modern poultry production," held at VIV Europe this year.
Integrated into a database and interpreted, poultry production data can offer insights for farmers and any other stakeholders given access.
There are three main uses for this processed data: improving production, veterinary guidance programs and, collaboration in research projects, for example, with governments, universities or pharmaceutical companies.
This aggregated and analyzed data can serve the various stakeholders in poultry production in different ways, helping them align their individual inputs to what is actually happening on the ground.
1. Producer’s perspective
David Speller, managing director Applied Poultry Group & Optifarm in the U.K., said that, at producer level, data is used for:
- Performance monitoring and improvement
- System fine tuning
- Outcome prediction, rapid response
- Idea, technology development
- Staff work pattern evaluation
Speller gives the example of examining water consumption data. He notes that, with water consumption recorded, growth rates and activity levels can be predicted, and continues that understanding water demand and availability might result in the introduction of an extra drinker into the barn, for example.
The poultry house environment’s impact on comfort levels can also be better understood. For example, perhaps ventilation is coming on too soon, birds feel cold and become less active. Disturbances in the house, such as noise and other stressors, can be discovered, and healthy feed availability can also be monitored.
“I can even know if the stockman worked on a Sunday or not, if the water consumption stays the same. This is the type of information we can get with simple things,” Speller said.
His company reviews multi-site data of farms in various parts of the country via a single platform. Overlaying data, such as body weight, he can see where something is going wrong.
“It is not going to solve your problems, but it will allow you to maximize where there is potential,” he said.
Data alone, without analysis and action, will not bring about change; farmers must respond, and do so as quickly as possible.
“It might be too late to react to data from yesterday,” Speller said. “As farmers, we should probably work at night and sleep in the day. Drinkers and ventilation always work well, but the minute I go, problems start.”
The internet of things can pay dividends, but there is no guarantee of a return on investment.
Producers must be prepared to have an open mind and to make changes in light of what the data reveals and, sometimes, there may be more questions than answers.
Speller recommends starting gradually, first working with simple but effective data monitoring and applying lessons learned, and working with professionals to achieve goals more quickly.
He also notes that big data is a fast-moving area and that failure to embrace the benefits that it brings could be like handing an advantage to competitors.
With extensive data gathering, areas of specific interest to each individual can be extracted, analyzed and acted upon. | strangebirdy, Bigstock.com
2. Veterinarian’s perspective
Veterinarians are one of the important decision-makers in poultry production, and there will be specific data sets of interest to them.
Dr. Johan van Erum, managing partner of Poulvet Group in Belgium, said they particularly need to check:
- Performance data: flock charts, zootechnical parameters
- Farm/management data: size, location, farm age, inventory of houses and house facilities.
- Health data: disease prevention, vaccination, treatments applied, necropsies and laboratory results, slaughterhouse results, and rejections.
- Epidemiological data
For Van Erum, data collection is of high importance for a poultry veterinarian to identify potential factors to improve poultry health, to adapt veterinary protocols to prevent disease and to help in antibiotic reduction.
Data may be collected for monitoring and improving flock health, but the veterinarian supports the farm manager, and so a greater understanding of the flock can be fed into the farmer’s management decision-making.
Significant data is important as food safety and transparency have become paramount, and it helps to meet government and consumer requirements.
Annual farm reports must cover management, performance, housing and health – the complete production process. It is an audit that identifies potential problems and risk factors specific to the farm.
Lastly, data interpretation is crucial, and the veterinarian plays a key role in this.
3. Feed miller’s perspective
Data has to fit with long-term business goals and, according to Joost Sparla, marketing and technical director poultry with ForFarmers of the Netherlands: “Collecting data is not enough; you have to make a plan, a strategy.”
Collecting data is simply the starting point in forecasting, then comes conversion into insights, but it is worth remembering who owns the data and who can use it. Once all of this has been established, a strategy can be executed.
An example of what big data can offer is a better understanding of the relation between feed production parameters at the feed mill and broiler performance in the field, which can be gained when data from both are collected and compared.
At the feed mill, pellet temperatures, steam addition, energy use, pellet durability, pellet hardness and, of course, the various raw materials used in formulation, are measured.
“If you are able to connect this with the feed conversion ratio, daily weight gain and health status, you can improve your milling process with data analysis,” Sparla said. “With all the data, maybe you can see how pellet hardness is affecting grower or finishing period performance.”
Through data and algorithms, the feed mill can offer added value to the farmer. By continuously monitoring and analyzing the relationship between feed and nutrient intake and measured outputs, feeding programs can be improved.
4. Scientist’s perspective
Every broiler is different, and there is no template that fits all.
Professor Bart de Ketelaere, of KU Leuven in Belgium, said success is increasingly based on data – a lot of data.
Where biological material is concerned, for example poultry, big data and high-quality models are needed, although he echoes that they offer no guarantee of success.
It is not enough to simply gather data -- it must be the right data.
“When I say we need the right data, I say we need to understand the underlying data processes, to avoid spurious correlations,” he said.
Data should cover the range and variability of interest, to form a solid basis for prediction, avoiding events that may correlate, but that actually have nothing to do with each other.
Good models describing reality are also needed, as are good visualization tools to demonstrate value. For broilers, physiological-based models describing growth can be highly valuable.
Sampling is also very important. With poultry, this needs to be conducted frequently; providing weekly data is often not sufficient.
de Ketelaere also believes that, to be successful with big data, there needs to be close interaction between data scientists and those with a clear understanding of the underlying processes, for example poultry scientists.
Big data in and of itself will not solve problems, rather it informs better decision-making. | Peter Hermes Furian, Bigstock.com
Is big data only for larger poultry producers?
The often talked-about use of big data would appear to offer a panacea of opportunities for poultry producers. But does big data have to be all or nothing, something for only large producers able to make significant investments, or can its advantages also be harnessed by smaller operations on a more limited scale? WATT Global Media spoke to Richard Ten Cate of Dutch agriculture consulting company FarmResult to find out more.
Availability of monitoring systems continues to grow, allowing producers to see in real time what is happening in their flock and act accordingly. What is monitored, however, can be as broad as a system may allow or as focused as a producer may want.
For some, installation of a monitoring system may seem daunting, or financially out of reach for smaller producers, but this is not necessarily the case.
Take, for example, PoultryResult, which can be used across the largest integrated farm, recording and collating data ranging from details of inputs, to house conditions and final slaughter weights, with everything recorded automatically, or can simply be set up to store limited data entered by hand by the farmer.
The product is very much adaptable to each individual’s circumstances, working with equipment already operational inside the house, meaning that installation of additional sensors may not be necessary.
“We try to make use of those systems already in place, because those are the systems that control the house. We extract the data from those sources to use in the overall management of the broiler house,” Ten Cate said.
Information can be gathered on water, feed, weight, temperature, and so on. Put simply, if there is something that can be measured, information can be gathered and fed back to a dashboard.
Custom fit
When a poultry farm has multiple locations or multiple houses, a network can be built that that connects all locations to the dashboard. A pilot has already been carried out with a large South American integrator operating across two locations with 10 houses each.
At the other end of the scale, the system is also available to the small producer. If there are no sensors in the house, a sensor can be installed, and PoultryResult offers an iOS or Android app that a farmer can use to enter data by hand while doing the rounds each day, and the data is then fed back to the dashboard.
Anything that can be measured can be included in the system, facilitating a maximal or minimal approach, and allowing producers to use the system in ways with which they feel comfortable, Ten Cate said. Each producer will come with a unique set of skills and approach to running the house, and the system works with this.
While the system can, it does not have to monitor all aspects of the flock’s performance. If there is no requirement for daily performance data, then simply feed conversion, total gain and growth per day at the end of the cycle can be recorded, for example.
In addition to keeping check on the house, PoultryResult can be seen as having a preventative aspect, Ten Cate said. The minute a problem becomes apparent in one house, producers can act to ensure that a similar situation does not occur in others.
The dashboard alerts producers to potential problems via a simple red light. If something is wrong or deviating from normal, it is simply a question of clicking on the location and drilling down to the individual house to see what is happening and making any necessary adjustments.
Beyond operational management and comparing planned with actual performance, there is the option of sharing information, should a producer so wish, and Ten Cate emphasizes that collected data remains the property of the producer. This sharing may be with feed suppliers, for example, or pooled with other producers to enable benchmarking to take place.
Can big data make sense of the poultry industry?
