Precision livestock technologies are yet to be fully reliable as a standalone means of detecting respiratory disease, a review by Brazilian scientists suggests.
The researchers from Pontifícia Universidade Católica do Paraná looked at 23 studies involving technologies used to identify signs of disease in cattle, pigs and poultry. The majority of the technologies monitored the sounds of animals to identify those suffering with illnesses, though others used thermal imaging and motion detection-based monitoring.
They judged each technology’s reliability based on it correctly detecting animals both with and without disease at a frequency of over 90% under field conditions, using a reliable reference test and without a risk of bias. The latter was judged based on whether there was important information missing, such as software, the devices themselves and how they were installed.
From the information analysed, they found that only three of the technologies met the required level of performance in the field. Of them, only a single one was not thought to be at risk of bias — a technology able to monitor the cough sounds of pigs to detect Mycoplasma hyopneumoniae, porcine reproductive and respiratory disease, and postweaning multisystemic wasting syndrome.
Better reporting will give clearer picture
“Some studies achieved high performance, but it is unclear if the technologies would perform well when applied in real-life situations. Many studies were only conducted on laboratory conditions or used an unreliable reference test,” the researchers explained in the journal Animals.
A number of the studies included only data on the technology’s identification of diseased animals (sensitivity), they noted, and not those that were well (specificity). This poses a risk, as low specificity could result in high numbers of ‘false positives’, unnecessarily raising the alarm for farmers.
They suggested that monitoring more than one clinical sign of disease is important in achieving a reliable diagnosis, though the reviewed studies largely focused on just a single marker.
“For example, monitoring cough and fever together could be useful for developing a technology able to better diagnose respiratory diseases. Other data, not collected by the technology, could also be used to overcome false positives, for example: season, feeding time and animal handling,” they said.
Despite this, the scientists expressed optimism about the potential for automated disease detection in livestock production, given the results seen in studies. They also acknowledged that they had potentially overlooked reliable technologies due to not including conference proceedings and company reports in the scope of their review.
They concluded by calling for clearer reporting of results and methodologies, to enable better understanding of the effectiveness of technologies in this emerging area of ‘agriculture 4.0’.
