No more lame excuses - New technology can detect lameness early - and save farmers huge sums
Lameness has been described as the most serious welfare concern facing the European dairy industry.
It is not only a problem for the cow; it can also lead to significant financial losses for farmers. Teagasc research has shown that a case of clinical lameness is estimated to cost €160-€300.
Please log in or register with Farming Independent for free access to this article.
Dairy lameness is primarily caused by diseases or injuries to the hoof, which can be categorised as either affecting the claw horn or the surrounding skin.
Early and accurate lameness detection can improve the overall prognosis and the welfare of cows.
Unfortunately, early detection of lameness is extremely difficult; cows show little behavioural response to pain until injuries are advanced, and farmers find it hard to spot early signs of impaired movement in cows.
However, new technology being developed in Waterford Institute of Technology's ICT research wing, Telecommunications Software and Systems Group (TSSG) aims to solve this problem.
Researchers there are utilising leg mounted sensors - measuring step count, lying time and swaps per hour - in combination with machine learning algorithms to identify lame cattle at an early stage.
The data gathered is analysed in the cloud, and anomalies are sent to the farmer's mobile device, so affected animals can be treated immediately and avoid further effects.
By detecting early lameness before it can be visually captured, treatment costs are decreased while animal welfare is improved.
The 'project' is called MELD -Machine Learning for Early Lameness Detection - and is aimed primarily at dairy cattle.
The research is funded by Internet of Food and Farm 2020 and has expanded on a previous SmartHerd project that was supported by the Science Foundation Ireland Connect Research Centre.
As part of a real-world trial in Waterford, 150 dairy cows were fitted with a long-range pedometer.
Mohit Taneja, a doctoral researcher at TSSG, explained that accelerometric readings from those pedometers are converted into behavioural time series activities such as step count, lying time and how many times a cow gets up and down.
"These three activities feed into the model. In our research, we engaged an animal health expert to get label data to see if an animal is lame or not," he said. "We then use a machine learning methodology which uses these three activities to access whether a cow is lame."
Radio transmitters send the data to a farmer's PC where they can monitor their cows.
"You do not need high-speed broadband for the system to work. The software can be downloaded and can be integrated into existing farm management systems," said Mr Taneja.
The initial results indicate that the system will be able to predict lameness three days before it can be visually captured by the farmer, with an overall accuracy of 87pc. This means that the animal can either be isolated or treated immediately to avoid worsening the condition.
"We calculate that lameness in dairy farmers costs dairy farmers approximately €80m per year," said Mr Taneja.
"We found that milk yield increased by 2.5-3pc using our technology."
He added that one of the key benefits of detecting lameness earlier is that it reduces the need to use antibiotics.
"As the size and scale of farmers increase, it is getting harder to see each and every animal," said Mr Taneja.
"If you are identifying lame cows through human visual inspection you are detecting them at a very late stage.
"They are severely lame and will usually have to go on an antibiotic cycle for three, seven or 10 days.
"Obviously, the milk from that animal can not be sold by the farmer. There is a real economic impact so being able to intervene early is vital for farmers."
'The system works indoors and outdoors'
Paul Malone, security and trust research manager at Waterford IT's TSSG, who is also working on the lameness detection project, is keen to point out that the system can be used in both housed and grazing dairy systems.
"We have developed a system that is environment agnostic," he said. "Irrespective of whether a cow is taking a lot of steps or not many, the system will still be able to distinguish between indoor and outdoor behaviours.
"The system is also population agnostic - the size of the herd doesn't matter."
The team is now intending to actively commercialise the research.
"We are expanding this trial which we did on 150 cattle to two different vendors," said Mr Malone.
"We have designed the system in such a way that it can plug into other vendors' offerings. The business model is that we would offer this service on top of existing systems."
The expanded trial includes two beef herds in South Africa, a beef herd in Israel and two farms in Portugal.
For Stories Like This and More
Download the Free Farming Independent App