Machine learning helps detect calf health problems earlier

Subcutaneous sensors and artificial intelligence could help identify health problems in calves before visible symptoms appear, ERR reports in an article by Rait Rand of the Estonian University of Life Sciences.

Cattle often hide signs of illness, meaning that by the time a problem is visible, it may already be advanced. Delayed treatment can require more medication and cause significant financial losses for farmers.

Researchers tested subcutaneous implants that can be injected under the skin near a calf’s neck or shoulder, much like a vaccine. In trials involving 720 calves, the implants were successfully placed in all but 10 animals. The researchers said the sensors stayed securely and safely in the animals for their lifetimes, and implantation required neither anaesthesia nor shaving.

Unlike rectal temperature checks, which can stress animals and are difficult to carry out frequently, the implants allow temperature to be monitored without repeatedly handling the calves. Earlier alternatives, including swallowed sensors and ear thermometers, were described as unreliable or inaccurate.

However, temperature under the skin is affected by the surrounding conditions: it can fall in a cool barn or rise in hot weather, obscuring a fever. Researchers trained machine-learning algorithms using more than 160,000 records from clinical trials. The data included subcutaneous temperatures, local weather conditions and reference measurements of internal body temperature from swallowed sensors.

The study aimed to use these combined measurements to make health monitoring more useful for detecting changes in calves before they become apparent to farmers.