Artificial intelligence could detect pneumonia in pigs, study finds
"What's exciting is that the AI also had perfect consistency, even though multiple people were involved in its training" - Robert Valeris-Chacin
A new study has explored the capabilities of artificial intelligence (AI) for detecting respiratory disease in pigs.
The research team, from the Texas A&M College of Veterinary Medicine & Biomedical Sciences, found that AI could support the detection of lesions in pig lungs – which could be a sign of pneumonia-causing bacteria.
Although the AI is not considered as accurate as a veterinary professional, its behaviour is considered to be similar to a person.
The technology could be used as part of European food animal production. Vaccine manufacturers often send veterinary professionals to the processing plants to monitor the success rates of their vaccines.
The new project, led by Robert Valeris-Chacin, sought to assess the capabilities of the AI to ascertain whether it would increase the efficiency and accuracy of the process.
Researchers also measured how consistent expert evaluators were in assessing pigs’ lungs, and how often they agreed with each other. This was examined in consideration that the study conditions differed from real life, where evaluators can touch the lungs to support diagnosis.
Experts were asked to evaluate hundreds of images of pigs’ lungs for bacterial pneumonia. Some of the images were repeated to confirm the consistency of the evaluators’ responses.
The expert evaluators, as well as the AI system, provided a total lung lesion score, lesion score per lung lobe and a percentage of the affected lung area.
The results revealed that, although the evaluators disagreed with each other quite often, their individual responses were generally consistent. The same evaluator proved very likely to score the same way each time they were presented with an image.
The artificial intelligence system was also found to have perfect consistency. It showed moderate accuracy (62- 71 per cent) in identifying lesioned and non-lesioned lung lobes.
Dr Valeris-Chacin said: "What's exciting is that the AI also had perfect consistency, even though multiple people were involved in its training,
"The company behind this AI wanted to create an AI that would mimic the way human evaluators score the lungs, and the AI is very promising in this regard."
The full study can be found in the journal VetRes.
Image © Shutterstock
RCVS Knowledge has called on vet practices to audit their post-operative neutering outcomes.
